Organized intraethnic conflict and cooperation in divided stateless nations: the competitive dynamics of rival ethnonationalist factions

PDF

Download

Open document

Transcript (if available)

Content

Organized Intraethnic Conflict and Cooperation in Divided Stateless Nations
The Competitive Dynamics of Rival Ethnonationalist Factions

by

Michael R. Pfonner

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
POLITICAL SCIENCE AND INTERNATIONAL RELATIONS

May 2022

Copyright 2022 Michael R. Pfonner
ii

National aspirations must be respected; people may now be dominated and governed only by
their own consent. 'Self determination' is not a mere phrase; it is an imperative principle of
action.
- Woodrow Wilson

Declaration of a state [Kurdistan] is the legitimate right of the Kurdish nation.
- Mesûd Barzanî

“And now,” I said, “you traitor bent on evil,
I do not need your talk, for I shall carry
true news of you, and that will bring you shame.”
- Dante Alighieri
iii

This work is dedicated to my wife, Yousra Pfonner. It would have not been possible without
your unwavering support. Me te xosh dve, Ycera.

iv

Acknowledgements
There are many people who contributed in some way or another to this project. First, I
would like to express my gratitude to my advisors. Pat James, Steve Lamy, and Yao-Yi Chiang all
offered their assistance, expertise, and encouragement as this project developed. I also appreciate
the support from the faculty and staff of the Political Science and International Relations
Department and the Spatial Sciences Institute at the University of Southern California. Many
discussants and commenters at APSA, ISA, IPSA, and ARC offered helpful insight on earlier
drafts of this project. The project also benefitted from input from my cohort mates, especially Paul
Orner and S. Jason Giannaros.
I am also deeply appreciative of the family of Michael R. Young and their support of the
School of International Relations.
Several of my comrades from the Marine Corps suffered through my obsession with this
topic and patiently listened to my rants about it. In particular, I thank Scot “Scoot Jansen” Johnson
and Chris “Melman” Mennell of the GFC. BJ Kent also deserves special mention for his patience
and willingness to read obscure and bizarre papers.
Not all who are important to us will make it to the culmination of our achievements and
completing graduate school takes a large chunk of our lives. My uncle, Jimmy Riley, and cousin,
Adam Riley, died during this period. My cousin Dorothy also passed. Especially bitter is the loss
of my great-uncle, Elmer Pfonner. Along with the rest of the Pfonners of the West, he thoroughly
welcomed Yousra and me into the extended family when I first came to Southern California from
Western New York as a young Marine attending Infantry Training Battalion. The last pets of my
youth—Garm, Basil, Sif—also did not reach this point.
v

One other key figure who greatly influenced my interest in the topic of intra-ethnic conflict
was my father-in-law, Darbo Ido Avdi. He passed away near the completion of this project from
various illnesses that were caused by exposure to mustard gas during and after the Iran-Iraq War.
Darbo and his comrades who survived lingered for decades before dying horrendous deaths. They
sacrificed their health, lungs, and eyeballs safeguarding their families and their people from the
depredations of nefarious actors, and I have the utmost respect for them.
I also need to express my gratitude to my parents, Frank and Patricia Pfonner, for their
williningness to help with anything Yousra and I needed. My brothers, Frank and Joe, also read
earlier versions of this work and offered advice. Finally, Yousra deserves an especially profuse
thanks for her non-stop love, support, advice, and patience.

vi

Table of Contents

Epigraph...........................................................................................................................................ii
Dedication.......................................................................................................................................iii
Acknowledgements.........................................................................................................................iv
List of Tables...................................................................................................................................ix
List of Figures.................................................................................................................................xi
Guide to Abbreviations, Acronyms and Initialisms.......................................................................xiv
Abstract ......................................................................................................................................xxiv
Preface........................................................................................................................................xxvi

Chapter 1: An Overview of the Puzzles regarding Conflict and Cooperation among
Divided Stateless Nations...............................................................................................1
1.1: The puzzle of ethnonationalist rivalry within divided stateless nations….........1
1.2: Limitations to current explanations of intraethnic factional conflict……
and cooperation..........................................................................................11
1.3: A brief depiction of the argument and its contributions..................................24
1.4: Overview of the research design......................................................................27
1.5: Preview of findings…......................................................................................31
1.6: Overview of chapters.......................................................................................32

Chapter 2: The Conflict and Cooperation Theory: Explaining Patterns of Behavior among………
Rival Ethnonationalist Factions...................................................................................34
2.1: Defining and conceptualizing organized intraethnic conflict and
cooperation.................................................................................................34
2.2: Propositional pathways of conflict and cooperation for rival DSN factions.....40
2.3: The Conflict and Cooperation Theory: Depicting rival factional
behavior using the systemist technique.......................................................54

Chapter 3: An Analytically Eclectic Design of Intraethnic Conflict and Cooperation among
Rival Factions...............................................................................................................65
3.1: Empirical strategy...........................................................................................68
3.2: Research design...............................................................................................71
3.3: Case selection................................................................................................103
3.4: Plan of the case chapters................................................................................110

Chapter 4: Divided Stateless Nations: Intraethnic Conflict and Cooperation among Rival
Kurdish Political Factions..........................................................................................114
4.1: Historical overview of the Kurdish people and Kurdish political Factions..116
vii

4.2: A depiction of conflicts among rival KPFs...................................................128
4.3: The roots of DSN rivalry: Evaluating existing explanations.........................134
4.4: How do DSNs fight? A spatial analytic approach to DSN conflict patterns...139
4.5: How do DSNs cooperate? Examining who dies and where they die..............148
4.6: Summing up: Tentative findings from DSN conflict and cooperation...........158

Chapter 5: Single State Separatists: Intraethnic Conflict among Rival Chechen Factions..........165
5.1: Historical overview of the Chechen people: Resistance to outsiders and……...
internal cleavages.....................................................................................166
5.2: A depiction of intra-Chechen conflict............................................................174
5.3: The roots of SSS rivalry: Evaluating existing explanations...........................180
5.4: How do SSSs fight? A spatial analytic approach to SSS conflict patterns…..185
5.5: Do SSSs cooperate? Examining who dies and where they die.......................191
5.6: Summing up: Tentative findings from SSS conflict and cooperation............196

Chapter 6: Divided Irredentist Groups: Rivalry and Conflict among Albanian
Irredentist Factions?...................................................................................................203
6.1: Historical overview of the Albanian people: Resistance to outsiders and
internal cleavages.....................................................................................204
6.2: Background to the Yugoslav Wars and Albanian factional development....212
6.3: Origins of Intra-Albanian conflict—or violence?.........................................216
6.4: Cooperation among rival(?) Albanian irredentist factions...........................220
6.5: Summing up: Tentative findings from DIG violence and cooperation...........229

Chapter 7: Two States, One Nation: Conflict between Korean State Regimes?............................234
7.1: Historical overview of the Korean people: Resistance to and adoption
of outside influence, and the precedence of north-south division.............235
7.2: A depiction of rivalry between Korean state regimes.....................................242
7.3: Examining the origins of Korean rivalry........................................................248
7.4: Examining post-Cold War conflict and cooperation among KSRs................252
7.5: Cooperation against outsiders?......................................................................256
7.6: Summarizing the TSON case.........................................................................258

Chapter 8: Findings: Tentatively Supporting a Categorical Approach to Organized
Intraethnic Conflict....................................................................................................264
8.1: Summing up: A review of propositions and hypotheses................................264
8.2: DSN conflict and cooperation........................................................................266
8.3: SSS conflict and cooperation.........................................................................274
8.4: DIG violence and cooperation.......................................................................280
8.5: TSON conflict and cooperation.....................................................................286

Chapter 9: Advancing the Study of Organized Intraethnic Conflict .............................................293
9.1: The Conflict and Cooperation Theory and its implications for
the literature.............................................................................................293
9.2: Implications for policy..................................................................................305
9.3: Outlook for organized intraethnic conflict, cooperation, and resolution........310
viii

Notes............................................................................................................................................312
References....................................................................................................................................327
Appendices...................................................................................................................................380
Appendix A: Reference list for statistical software packages...........................................380
Appendix B: Historical demographic maps selected for map processing.........................383
Appendix C: An elaboration on the statistical analyses....................................................391
Appendix D: Case selection table.....................................................................................401
Appendix E: Preliminary tests of intra-Kurdish conflict and cooperation........................407
Appendix F: Preliminary tests of intra-Chechen conflict..................................................429
Appendix G: Preliminary tests of intra-Albanian cooperation..........................................444

ix

List of Tables

2.1: Conflict and Cooperation theory: Summary of propositions and hypotheses..........................52
2.2: Systemist notation...................................................................................................................55
3.1: Projected coordinate reference systems (CRS.........................................................................79
3.2: Typology of ideal types of intraethnic conflict and cooperative patterns...............................107
4.1: Temporal analysis: Change in Kurdish-inhabited territory, 1910-2011.................................122
4.2: Spatial analysis and modeling: Spatially lagged regression models of intra-Kurdish
conflict.............................................................................................................................142
4.3: Poisson point process modeling of Kurdish rival cooperation, 2014-2017............................155
4.4: Summary of evidence for DSN conflict based on KPFs.........................................................159
5.1: Temporal analysis: Change in Chechen-inhabited territory, 1878-2015................................173
5.2: Spatial analysis and modeling: Spatially lagged regression models of intra-Chechen
conflict.............................................................................................................................187
5.3: Summary of evidence for SSS conflict based on CSFs..........................................................197
6.1: Temporal analysis: Change in Albanian-inhabited territory, 1940-1991...............................212
6.2: Poisson point process modeling of Albanian rival cooperation, 1998-2001..........................226
6.3: Summary of evidence for DIG conflict based on AIFs..........................................................230
7.1: Temporal analysis: Change in Korean-inhabited territory, 1946-2020..................................241
7.2: Summary of evidence for TSON conflict based on KSRs......................................................258
8.1: Conflict and Cooperation theory: Review of propositions and hypotheses............................265
D.1: Case selection process..........................................................................................................402
E.1a. Summary statistics: Ordinary least squares (OLS) regression models of episodes of
intra-Kurdish conflict.......................................................................................................412
E.1b: Summary statistics: Logarithmically transformed regression models of episodes of
intra-Kurdish conflict with stepwise selection based on Bayesian information
criterion (BIC)..................................................................................................................414
E.2: Measures of spatial autocorrelation of intra-Kurdish conflict events: Moran’s I statistic......418
E.3: Measures of spatial clustering of intra-Kurdish conflict events: Getis-Ord General G..........420
E.4: Spatial analysis and modeling: Robustness checks of intra-Kurdish conflict........................422
x

F.1a: Summary statistics: Ordinary least squares (OLS) regression models of episodes of
intra-Chechen conflict......................................................................................................434
F.1b: Summary statistics: Logarithmically transformed regression models of episodes of
intra-Chechen conflict with stepwise selection based on Bayesian information
criterion (BIC...................................................................................................................436
F.2: Measures of spatial autocorrelation of intra-Chechen conflict events: Moran’s I statistic.....439
F.3: Measures of spatial clustering of intra-Chechen conflict events: Getis-Ord General G.........440
F.4: Spatial analysis and modeling: Robustness checks of intra-Chechen conflict.......................442

xi

List of Figures

2.1: Systemist exposition of the Roots of Rivalry (P1)...................................................................57
2.2: Systemist exposition of the Effects of Distance (P2)...............................................................59
2.3: Systemist exposition of Out-group Incursion (P3)...................................................................61
3.1: Georeferenced and processed historical demographic maps....................................................81
3.2: Analyzing populational-territorial change of ethnic predominance over time.........................84
3.3a: Choropleth map: Total population of the USA by state..........................................................87
3.3b: Choropleth map: Population of California by county............................................................88
3.3c: Choropleth map: Population density of California by Census tract per square mile...............89
3.4: Assessing the effects of distance on intraethnic conflict among rival factions.........................95
3.5: Assessing the effects of out-group incursion on cooperation among rival factions................100
4.1: Distribution of Kurdish populations before and after the Treaty of Lausanne (1923)............120
4.2: Temporal analysis: Change in distribution of Kurdish populations before and after
the Treaty of Lausanne (1923)..........................................................................................121
4.3: Kurdish hûnermendê and factional support...........................................................................129
4.4: Extent of the 1994-1997 intra-Kurdish conflict.....................................................................132
4.5: Extent of the 2011-2020 period of intra-Kurdish conflict......................................................134
4.6: External intervention into intra-Kurdish conflict in the 1990s...............................................149
4.7: External intervention into intra-Kurdish conflict (KNC- and KCK-affiliated groups)..........150
4.8: Turkish intervention into intra-Kurdish conflict....................................................................152
5.1: Distribution of the Chechen population in 1878....................................................................170
5.2: Distribution of the Chechen population in 1956....................................................................171
5.3: Temporal analysis: Change in distribution of Chechen populations before and
after deportation................................................................................................................172
5.4: Extent of the 1994-2009 intra-Chechen conflict....................................................................178
5.5: Extent of the conflict with the VN of the Caucasus Emirate...................................................180
5.6: Russian intervention into the intra-Chechen conflict continuum...........................................193
5.7: Russian intervention into the conflict with the Caucasus Emirate..........................................194
6.1: Distribution of the Albanian population in 1940....................................................................208
xii

6.2: Distribution of the Albanian population in 1991....................................................................210
6.3: Temporal analysis: Change in distribution of Albanian populations after the Second
World War and prior to the Yugoslav Wars......................................................................211
6.4: External intervention into ethnic Albanian irredentist movements, 1998-2012.....................223
7.1: Distribution of the Korean population, 1946-2020................................................................240
7.2: Inter-Korean conflict events, 1991-2020...............................................................................254
8.1: Origins of conflict between rival DSN factions.....................................................................268
8.2: Objects of contestation for DSN factions in conflict..............................................................270
8.3: Cooperation between rival DSN factions...............................................................................272
8.4: Origins of conflict between rival SSS factions.......................................................................276
8.5: Objects of contestation for SSS factions in conflict...............................................................277
8.6: Lack of cooperation between rival SSS factions....................................................................279
8.7: Origins of violence between members of rival DIG factions.................................................281
8.8: Objects of contestation for individual-level DIG violence.....................................................283
8.9: DIG cooperation....................................................................................................................285
8.10: Origins of conflict between rival SSS factions.....................................................................287
8.11: Objects of contestation for TSON regimes in conflict.........................................................289
8.12: Lack of cooperation against out-group forces between TSON regimes in conflict..............291
B.1: Pre-Lausanne map of Kurdish distribution...........................................................................384
B.2: Pre-Soviet map of Chechen distribution...............................................................................386
B.3: Pre-Socialist Federal Republic of Yugoslavia map of Albanian distribution........................388
E.1: Georeferenced historical distribution of Kurdish populations...............................................407
E.2a: Summary statistics for episodes of intra-Kurdish conflict: Pearson correlation
(distance from regional centers).......................................................................................409
E.2b: Summary statistics for episodes of intra-Kurdish conflict: Pearson correlation
(distance from borders)....................................................................................................410
E.3a: Plots of log-transformed regression models for episodes of intra-Kurdish conflict:
Regional Centers Model (1).............................................................................................416
E.3b: Plots of log-transformed regression models for episodes of intra-Kurdish conflict:
Borders Model (2)............................................................................................................417
E.4: Moran scatterplots of spatial autocorrelation of intra-Kurdish conflict events:
Empirical Bayesian smoothing (EBS)..............................................................................419
xiii

E.5: Kernel density estimation plot of cooperation between rival Kurdish factions......................424
E.6: Assessing spatial dependence and randomness.....................................................................426
E.7: Comparing the observed model to random models of rival KPF cooperation........................428
F.1: Georeferenced historical distribution of Chechen populations..............................................429
F.2a: Summary statistics for episodes of intra-Chechen conflict: Pearson correlation
(distance from regional centers).......................................................................................431
F.2b: Summary statistics for episodes of intra-Chechen conflict: Pearson correlation
(distance from borders)....................................................................................................432
G.1: Georeferenced historical distribution of Albanian populations............................................444
G.2: Kernel density estimation plot of cooperation between Albanian factions............................445
G.3: Assessing spatial dependence and randomness.....................................................................447
G.4: Comparing the observed model to a random model..............................................................448

xiv

Guide to Abbreviations, Acronyms and Initialisms

General and theoretical abbreviations

AIF Albanian irredentist faction
CCP Chinese Communist Party
CIA Central Intelligence Agency
CSF Chechen separatist faction
CSP Center for Systemic Peace
DIG Divided irredentist group
DoD Department of Defense (US)
DSN Divided stateless nation
HRW Human Rights Watch
ICG International Crisis Group
IGO Intergovernmental organization
IRB Internationally recognized border
KMT The Kuomintang, or Chinese Nationalist Party: also known as the CNP.
KPF Kurdish political faction
KSR Korean state regime
MAR Minorities at Risk Project
NGO Nongovernmental organization
SSS Single-state separatist
TCO/TOC Transnational criminal organization/Transnational organized crime
TSON Two states, one nation
UN United Nations
UNHCR United Nations High Commissioner for Refugees
US United States (of America)
USSR Union of Soviet Socialist Republics

xv

Methodological abbreviations

AIC Akaike information criterion
ANN Average nearest neighbor
BIC Bayesian information criterion
CRS Coordinate reference system
CSR Complete spatial randomness
DEM Digital elevation model
DPI Dots per inch
DV Dependent variable
EBS Empirical Bayesian Smoothing
EDA Exploratory data analysis
ESDA Exploratory spatial data analysis
EV Explanatory variable
GADM Database of Global Administrative Areas
GED Georeferenced Event Database
GeoEPR Georeferencing Ethnic Power Relations
GIS Geographic information system(s)
HDX Humanitarian Data Exchange
KDE Kernel density estimation
LRT Likelihood ratio test
MAUP Modifiable areal unit problem
MLE Maximum likelihood estimation
OCHA United Nations Office for the Coordination of Humanitarian Affairs
OLS Ordinary least squares
OSM OpenStreetMaps
OVB Omitted variable bias
POI Point(s) of interest
xvi

PPP Poisson point process
PRIO Peace Research Institute Oslo
SDA Spatial data analysis
SDBMS Spatial database management system
SM Spatial modeling
UCDP Uppsala Conflict Data Program

xvii

Actors and media pertaining to the Kurdish case

AA The Autonomous Zone of North and East Syria: the territorial unit in which
the PYD is the most powerful political party
ANF Ajansa Nûçeyan a Firatê, or Firat News Agency: a KCK- and PKK-
affiliated media source
DFNS Democratic Federation of North and East Syria: an older name for the AA
ENKS Encûmena Niştimanî ya Kurdî li Sûriyê, or Kurdish National Council in
Syria (KNC-S): a KDP-aligned faction in Syria
HPÊ Hêza Parastina Êzîdxanê, or the Êzîdxan Protection Force, formerly known
as the Protection Force of Sinjar (HPŞ): nominally independent, sometimes
affiliated with the KRG and KDP
HPG Hêzên Parastina Gel, or People’s Defense Forces, the armed wing of the
PKK; there is also an all-female equivalent called Yekîneyên Jinên Azad ên
Star, or Free Women’s Units, abbreviated as YJA STAR
IMK Bizûtinewey Îslamî le Kurdistan, or Islamic Movement of Kurdistan
ISG The Islamic State Group: a violent transnational Salafist organization
KCK Koma Civakên Kurdistan, or Kurdish Communities Union: an umbrella
organization containing many left- wing Kurdish factions in Iran, Iraq,
Syria, and Turkey
KDP Kurdistan Democratic Party: a nationalist-populist faction founded by the
Barzanî family; the most powerful party in the KRG; supports the KNC;
their armed forces are the KDP Peşmerge; denoted in media sources as PDK
for Partiya Demokrat a Kurdistanê, or the Democratic Party of Kurdistan
KKK Koma Komalên Kurdistan, or Peoples’ Confederation of Kurdistan; an
earlier name for the KCK
KNC Kurdistan National Council: the umbrella organization of right-leaning and
traditionalist Kurdish factions allied with the KDP
Komala Komeley Şorrişgêrrî Zehmetkêşanî Kurdistanî Êran, Komala Party of
Iranian Kurdistan: a Maoist insurgent group in Iran
KRG Kurdistan Regional Government: the governing body of the KRI
KRI Kurdistan Region of Iraq: the territory of the KRG
PÇDK Partî Çareserî Dîmukratî Kurdistan, or Kurdistan Democratic Solution
Party: a member of the KCK from Iraq
xviii

PDKI Hîzbî Dêmukratî Kurdistanî Êran, or Democratic Party of Iranian
Kurdistan: a PKK ally in Iran
PKK Partiya Karkerên Kurdistanê, or Kurdistan Workers’ Party: a Marxist
Kurdish ethnonationalist faction founded in Turkey that helped create the
KCK; its armed wings are the HPG and YJA STAR
PJAK Partiya Jiyana Azad a Kurdistanê, or Kurdistan Free Life Party: a KCK-
aligned faction in Iran
PUK Patriotic Union of Kurdistan, also known as YNK for Yekîtiya Nîştimanî
ya Kurdistanê: a center-left Kurdish ethnonationalist faction that splintered
from the KDP and forms the second most powerful party in the KRG; its
armed forces are the PUK Peşmerge
PYD Partiya Yekîtiya Demokrat, or Democratic Union Party: a Syria-based
member of the KCK that forms the most powerful party in the AA; its armed
forces are the YPG and YPJ
Rudaw A KDP-aligned media source
SAR Syrian Arab Republic
SNA Syrian National Army: a name used in some media sources to denote groups
related to the TFSA
SOHR Syrian Observatory for Human Rights
TEV-DEM Tevgera Civaka Demokratîk: Movement for a Democratic Society, the
ruling coalition within the AA whose most powerful party is the PYD
TFSA Turkish-backed Free Syrian Army: a rebel group in Syria that has
undergone many incarnations and contains a mix of secular, Islamist, and
Jihadist forces that are predominantly, but not exclusively, Arab
TSK Türk Silahlı Kuvvetleri, or Turkish Armed Forces: sometimes called the
TAF
YBŞ Yekîneyên Berxwedana Şingalê, or Şingal/Sinjar Resistance Units: Êzîdî
armed forces aligned with the PYD and PKK
YJÊ Yekinêyen Jinên Êzidxan or Êzîdxan Women's Units: formerly the
Yekîneyên Parastina Jin ê Şingalê (YJŞ), or Şingal Women’s Protection
Units: female Êzîdî armed forces aligned with the PYD and PKK
YPG Yekîneyên Parastina Gel, or People’s Protection Units: the armed forces of
the PYD
YPJ Yekîneyên Parastina Jin, or Women’s Protection Units: female armed
forces of the PYD
xix

Actors and media pertaining to the Chechen case

ASSR (Checheno-Ingush) Autonomous Soviet Socialist Republic; a semi-
autonomous SSR that was part of the Russian Soviet Republic
ChRI Chechen Republic of Ichkeria (Nóxçiyn Respublik Içkeri); a secessionist
government that succeeded the Checheno-Ingush ASSR
FSB Federal'naya sluzhba bezopasnosti Rossiyskoy Federatsii; Federal Security
Service of the Russian Federation; their primary responsibility is internal
security, though some external security functions are allowed
GRU Glavnoje Razvedyvatel'noje Upravlenije; Main Directorate of the General
Staff of the Russian Federation; now abbreviated as GU, but still often
referred to as GRU; their primary responsibility is external security
Kavkaz Center The media platform of the Caucasus Emirate
NCCChP/OKCLN All-National Congress of the Chechen People; a political organization led
by Dzhokhar Dudayev; the ruling party of the ChRI
PCChR Provisional Council of the Chechen Republic; an opposition movement at
odds with the ChRI
SPIR Special Purpose Islamic Regiment; a Chechen Salafist movement that was
allied with TOC organizations and opposed the ChRI
State News Agency The media platform of the post-split ChRI; also called Chechen-press
VN Vilayat Noxçiyçö; the Province of Nokhchicho; the Chechen portion of a
proposed division of the Caucasus by the Caucasus Emirate

xx

Actors and media pertaining to the Albanian case

ANA Armata Kombëtare Shqiptare (AKSh); Albanian National Army in North
Macedonia; the armed forces of the FBKSh
BND Bundesnachrichtendienst; the German Federal Intelligence Service
FARK Forcat e Armatosura të Republikës së Kosovës; Armed Forces of the
Republic of Kosovo; the armed forces of the LDK
FBKSh Fronti Për Bashkim Kombëtar Shqiptar; National Front for Reunification
of Albanians; the political wing of the ANA
FPRY Federal People’s Republic of Yugoslavia (1946-1963)
FRY Federal Republic of Yugoslavia (1992-2006)
KFOR Kosovo Force; a peacekeeping force organized by NATO
KLA Kosovo Liberation Army; the UÇK in Kosovo
KPC Kosovo Protection Corps; a security force that succeeded the KLA; also
known as the Trupat e Mbrojtjes së Kosovës (TMK)
LDK Lidhja Demokratike e Kosovës; Democratic League of Kosovo; a
nationalist and separatist organization that at first pursued peaceful
secession from Serbia under Ibrahim Rugova
LPK Lëvizja Popullore e Kosovës; People’s Movement of Kosovo; the political
wing of the UÇK, particularly the KLA
NLA National Liberation Army; UÇK in Macedonia
NLM National Liberation Movement; also called Lëvizja Antifashiste Nacional-
Çlirimtare (LANÇ); a communists resistance movement in the Second
World War
PDK Democratic Party of Kosovo (Partia Demokratike e Kosovës); a political
party that was founded by Hashim Thaçi and other former KLA fighters
SAS Special Air Service; a British special operations force
SFRY Socialist Federal Republic of Yugoslavia (1963-1992)
SHIK/SHISH Shërbimi Informativ Kombëtar/Shërbimi Informativ Shtetëror; National
Informative Service/National Intelligence Service; the post-communist
intelligence organs of Albania
xxi

UÇK Ushtria Çlirimtare Kombëtare; National Liberation Army ; the armed
forces of various violent irredentist groups, including the KLA, NLA, and
UÇPMB
UÇPMB Ushtria Çlirimtare e Preshevës, Medvegjës dhe Bujanocit; Liberation
Army of Preševo, Medveđa and Bujanovac; UÇK in the Preševo Valley
and adjacent areas in Serbia
UNMIK United Nations Interim Administration Mission in Kosovo; a UN
administrative organ that assists Kosovo with issues of governance

xxii

Actors and media pertaining to the Korean case

BRD Bundesrepublik Deutschland; West Germany; also known as the Federal
Republic of Germany (FRG)
CCP/CPC Chinese Communist Party; its official name was also called the
Communist Party of China
DDR Deutsche Demokratische Republik; East Germany; also known as the
German Democratic Republic (GDR)
DFRK Democratic Federal Republic of Koryo; a proposed framework for
reunification of KSRs favored by the DPRK
DMZ Demilitarized Zone; the demarcation between North and South
DPRK Democratic People's Republic of Korea; North Korea
GSOMIA General Security of Military Information Agreement; an intelligence-
sharing organization between the ROK and Japan
JSA Joint Security Area; a portion of the DMZ where DPRK and ROK
officials meet and negotiate, among other functions
KLA Korean Liberation Army; one of the armed forces of the KPG
KMT Kuomintang; Chinese Nationalist Party; the party that controlled Taiwan
until 1987
KNA Korean National Association; an early independence movement that at
times operated from exile
KNC Korean National Community; a reunification framework that is preferred
by the ROK
KPA Korean People’s Army; the military forces of the DPRK and WPK
KPG Korean Provisional Government; a government-in-exile during the
occupation of Korea by Japan
KVC Korean Volunteer Corps; one of the armed forces of the KPG
NLL Northern Limit Line; the maritime demarcation line between the DPRK
and the ROK
PLA People’s Liberation Army; the armed forces of the PRC
PRC People’s Republic of China
xxiii

PVA People’s Volunteer Army; a Chinese military force that assisted the DPRK
in the Korean War
ROC Republic of China; Taiwan
ROK Republic of Korea; South Korea
THAAD Terminal High Altitude Area Defense; an American anti-ballistic missile
system
UNTCOK United Nations Temporary Commission on Korea; an organization that
instituted elections in the ROK
WPK Workers’ Party of Korea; the political party of the DPRK
YKAP Yanbian Korean Autonomous Prefecture; an autonomous government in
China that has a population of Korean inhabitants

xxiv

Author: Michael R. Pfonner
Title: Organized intraethnic conflict and cooperation in divided stateless nations: The competitive
dynamics of rival ethnonationalist factions
Advisor: Patrick James
Committee: Patrick James (Chair, Dean's Professor of International Relations, University of
Southern California), Steven Lamy (Professor Emeritus of Political Science and International
Relations, University of Southern California), Yao-Yi Chiang (Associate Professor of Computer
Science and Engineering, University of Minnesota; Adjunct Associate Professor (Research),
University of Southern California)
Degree Date: May 2022

Abstract

Why do ethnonationalist factions that ostensibly share goals and foes fight each other despite what
we know about the pacific effects of coethnicity? When do they cooperate against out-group threats
and when will they not? I argue that a four-fold categorical approach which distinguishes between
varieties of organized intraethnic conflict will contribute to a better understanding of this
underexplored phenomenon. Categorization is based on whether factions claiming to represent
ethnic groups are divided or constrained by borders, and whether the group forms the core
community of a nation-state or not.
This work theorizes that factions claiming to represent a stateless nation that is divided by borders
will experience a form of organized intraethnic conflict that is distinct from conflicts that affect
other recognized forms. These include separatists in a single state, state-backed irredentists divided
by borders, and two states of co-ethnics divided by borders—divided societies—such as East and
West Germany.
Organized intraethnic conflict is rare, but it may be predicted and modeled depending upon its
categorization. The origins of conflict, the objects of contestation, and the conditions for
cooperation vary by type. This work’s primary contribution is the recognition of conflict involving
divided stateless nations (DSNs).
This work tests the proposal that DSN factional conflict is distinct from other forms using an
analytically eclectic empirical strategy with mixed methods. It employs linear modeling and novel
spatiotemporal analyses combined with process tracing and qualitative research.
Data is drawn from the Uppsala Conflict Data Program, NGO, IGO, and state sources. Qualitative
sources include factional press releases and media. Interviews with fighters and members of the
ethnic communities being studied are also incorporated.
xxv

Tests are applied to archetypical, extreme cases as a basis for baseline theorizing. The DSNs are
represented by Kurdish factions. Chechen factions stand in for constrained separatists. Albanian
factions are the irredentists. North and South Korea provide the divided society.
I find that the drivers of conflict for factions representing DSNs are more like divided societies
and unlike infighting among separatists and irredentists. DSN factions are unique with regards to
what they fight over. DSN factions will cooperate against outsiders if large numbers of unaffiliated
coethnics are killed in institutionally competitive territory. Surprisingly, non-combatant status is
only salient when accounting for factional affiliation. Rival separatists and divided societies will
not cooperate even under this condition, while irredentists will cooperate regardless of who is
killed and where.
Earlier generations of the literature have suggested that coethnicity is either a pacific force or a
source for division and exploitation by state actors. A key implication for the literature is that
precise effects of coethnicity can be predicted by understanding a group’s degree of territorial
division and statehood status.
This work has an important implication for policy as well. The prevailing view is to consider
organized intraethnic conflict as a singular phenomenon. Many pressing humanitarian crises and
interethnic conflicts develop from this phenomenon, yet without making necessary distinctions
these crises escalate. Proper identification could make for more positive conflict resolution
outcomes.

xxvi

Preface

This dissertation was the culmination of my doctoral education at the University of
Southern California in the Department of Political Science and International Relations. It was
written to fulfill the graduation requirements of the Graduate School at USC. The work for the
dissertation began following my qualifying exams in 2018.
This project was motivated by several factors. The first is deeply personal. My wife is from
one of the communities mentioned in this work. Many of her relatives were affected by the Islamic
State’s invasion of their territory in 2014. The subsequent interventions really generated the first
form of my overall research question regarding intra-ethnic conflict and cooperation among rivals.
US policymakers seemed to profoundly misunderstand the situation and made many mistakes in
the following years even as the conflict wound down.
I was on active duty in the US Marine Corps when the invasion occurred. Had I not been
contractually and legally obligated to report to formation every day I would have joined a Kurdish
militia to fight the Islamic State.
When I completed my term of active service, the conflict was essentially over. Some
Western volunteers I corresponded with had become deeply disillusioned with the infighting
amongst those militias. And so, I did the next best thing and studied the broader phenomenon of
intra-ethnic conflict. Working with Pat James and Steve Lamy, two scholars who have greatly
contributed to the study of ethnic conflict and human security, I discovered that it was not just the
policymakers who overlooked this aspect of key international crises. Many scholars whose work
I respect tend to treat these groups as monoliths. Vital fractures are routinely glossed over or, when
acknowledged, brushed to the side.
xxvii

The case of the Kurds is especially emblematic of this problem. Many (if not most)
scholars, even, tend to refer to any collective of Kurdish people as “the Kurds” and demonstrate a
disturbing lack of understanding that there are many differing views of what a hypothetical
Kurdish polity should look like. This is unfortunate, but expected, given the low level of attention
given to the phenomenon of organized intra-ethnic conflict.
Two other events motivated me and have sustained my belief that this phenomenon must
be better understood. The rise of Boko Haram and the resurgence of the Taliban are those two
events. The rule-making states of the liberal international economic order (LIEO) fundamentally
misjudge not only the centrality of ethnonationalism in other societies, but also the ways in which
ethnic group homogeneity is not a safe assumption to make.
The fall of Afghanistan took place as this project was completed. The LIEO first ignored
the importance of Pashtun nationalism, and then missed the morphing of that form
ethnonationalism into a successful civic nationalist movement based on cultural and religious
identity. The massive investment in Afghanistan over two decades was lost, and the peoples of
Afghanistan must now live with the consequences of poor decisions made by the LIEO.
The prevailing preference for explaining identity-based conflict is for economic theories,
reflecting the LIEO’s worldview. As a result, it is not easy to publish work that considers ideational
factors along with material ones. It is my hope that the work in this project helps contribute to a
shift toward a better-informed sense of the motivations behind many pressing crises of our time.
Given the rise of nationalism, separatism, and irredentism throughout the world, the LIEO must
adapt to remain relevant.
1

Chapter One
An Overview of the Puzzles regarding Conflict and Cooperation among Divided Stateless
Nations

1.1 The puzzle of ethnonationalist rivalry within divided stateless nations
A 19-year-old man named Saeed Khoudeida Qasim was stopped at a checkpoint with his father on
Saturday, May 14
th
of 2016 near Şingal, a city in northern Iraq, close to the border with Syria. The
checkpoint was manned by the Rojava Peşmerge, a militia affiliated with a Kurdish political party
from Syria and made up of Kurds displaced by conflict there. This group was part of a larger
umbrella of Kurdish ethnonationalist parties. Saeed was an Êzîdî—an ethnoreligious group most
heavily concentrated in the Nineveh Plains of Northern Iraq, but with populations found
throughout Kurdish-inhabited parts of the Middle East. Some Êzîdî consider themselves Kurds,
others do not. Kurdish political factions consider the Êzîdî to be the original Kurds. Those factions
have competed fiercely for influence among the Êzîdî population and for control over the territory
they inhabit. Saeed had joined a rival Kurdish ethnonationalist faction and served in their armed
forces. He had been inspired by that group’s defense of his people when they were attacked by the
Islamic State Group in August of 2014. His affiliation was known to the Rojava Peşmerge and
they shot him in the face.
1

This sad story is reduced to but a single data point in one of this work’s case studies, but it
stands as one data point among many in this analysis of intraethnic conflict. Though the two rival
Kurdish groups ceased their violent intraethnic political dynamics for a time and cooperated to
eject Islamic State fighters from Kurdish-inhabited territory in Iraq and Syria, they returned to
2

conflict once the out-group threat receded. Days-long armed battles over territory, targeted killings
in towns and at checkpoints, enforced disappearances of political rivals, arson committed against
rival party headquarters, and other forms of violence resumed and continue to this day, leading to
further displacement, forced migration, insecurity, and regional destabilization.
Logically, one would assume that ethnonationalist factions should cooperate in order to
achieve goals of self-determination, especially when they are under attack by out-groups. There
are indeed cases in which out-group threats caused groups engaging in organized intraethnic
conflict to unite. There are, for instance, examples from history. Consider the case of the Second
United Front. This alliance was between the Chinese Communist Party and the nationalist
Kuomintang. These groups held competing visions for what form an independent Chinese state
should take and, at such, were at war with each other. However, when Japanese forces took
Shanghai and Nanking in 1937, engaging in mass slaughter, pillage, and rape, the two factions
united temporarily to combat the out-group forces.
2
The Second United Front did not last long—
only through 1938. The rival dyad returned once more to in-group warfare once a string of Chinese
victories decreased the operational effectiveness of the Japanese forces.
However, there are other instances in which cooperation did not materialize, even in the
face of invasions by out-groups. Fatah and Hamas have engaged in conflict with each other, even
when faced with Israeli incursions. Chechen factions have served as counterinsurgents against
other Chechens in the service of the Russian state. The Cold War was rife with proxy conflict in
which divided societies like the two Koreas fought their ethnic counterparts as participants in
alliances comprised mostly of out-groups.
This work does not claim that organized intraethnic conflict is a frequently occurring
phenomenon. It is recognized to be rare in the literature. Fearon and Laitin (1996) and
3

Habyarimana et al. (2009) tell us that coethnics benefit from low information costs through shared
means of communication; they can locate one another easily through interpersonal connections;
and that there are incentives to distribute public goods fairly within the group. Organized conflict
among political factions within ethnic groups—rare as it is—is generally viewed as a result of
individual leadership differences; manipulation by the state; or regional, system-level, or global
transnational factors.
These explanations correspond, respectively, to Waltz’s (1959; 1979) first, second, and
third images. The image is the explanatory variable that influences the behavior of actors. Singer
(1960; 1961) referred to images as levels of analysis. The level of analysis is the tier of social
organization in which variables interact. The terms “image” and “level of analysis” will be used
interchangeably in this work. Arguably, other levels of analysis encompassing transnational social
and cultural movements, or regional security arrangements,
3
could be incorporated as existing
either above the system, or between the state and the system.
For the sake of parsimony, I include fourth-level factors with those of the third.
Additionally, the analyses in this work will proceed from the least aggregated to the most
aggregated, rather than from most to least. In other words, individual-level factors will be
examined first, followed by state-level ones, and then system-level factors. This is intentional.
Though this work pursues parsimony, it prioritizes accuracy and completeness. Each level might
well feature some evidence in favor of that explanation in a particular case. Going from least to
most aggregated will permit more detailed and holistic examinations of potential causes of
organized intraethnic conflict.
Like the reasons for intraethnic conflict, cooperation among rival factions is the subject of
three competing explanations. The first simply states that invasion by an outgroup will cause rivals
4

to join together out of a feeling of ethnic solidarity. A second explanation finds that out-group
intervention will more likely cause rival leaders to fragment, rather than cooperate. Finally, the
third explanation posits that the institutional distribution of power among rival groups will predict
whether fragmentation or unity will occur. These explanations also involve all three levels of
analysis.
The puzzle of organized intraethnic conflict
Let us return to the experience of the Kurdish factions from the above vignette. These are
groups that have experienced both conflict and cooperation and will serve as the basis for the most
significant case in this work. At times, they have been at war with each other. At other times, they
have worked together to eject outsiders from Kurdish-inhabited territory. These groups claim to
represent an entire people—the Kurds—rather than only the Kurds of a specific state. This
transborder dynamic is potentially different from other forms of separatist conflict because these
groups do not restrict their activities to a single state.
Kurdish groups also do not fall neatly into the existing explanations for conflict and
cooperation. It is in part the transborder aspect that helps to identify the limitation of those
explanations. Though much of the literature treats all ethnically related conflict in a similar
fashion, important works have disaggregated by type. Carment (1994) and Carment and James
(1995) categorize ethnic conflict as three types: post-colonial wars of liberation, separatist conflict,
and irredentist conflict to protect or absorb ethnic islands.
4
These varieties sometimes feature
intraethnic violence as part of their conflict processes. The conflicts of divided nations, such as
those of East and West Germany may also be included as two states compete for leadership over
a single ethnic group.
5

5

And so, there is a social science puzzle in which X occurs despite what is known and
explained by Y.
6
Rival factions representing divided stateless nations engage in conflict despite
what we know about the pacific effects of coethnicity. For instance, despite sharing an ethnicity
and a common foe, Hamas and Fatah nevertheless clashed in 2006-2007. Hundreds of fighters
were killed on each side and a lasting grudge between the two factions has developed.
7

And yet, in other cases, rival dyads do cooperate. An example from history—the Second
United Front between Chinese communists and nationalists—was described above, but there are
recent examples as well. The Nigerian army indiscriminately killed many Kanuri of all affiliations
8

when the Kanuri-dominated state of Borno began to lose ground in the conflict against the Kanuri-
dominated Boko Haram between 2009 and 2012. This led many Kanuri to turn to Boko Haram,
even if they were previously at odds with the latter group.
It is conceivable that explanations from different levels should be better at describing the
various types of conflict. Likewise, it is conceivable that different explanations for cooperation
could describe whether those patterns should materialize or not, and why or why not.
Existing research on intraethnic conflict approaches the phenomenon from the perspective
of Cold War proxy conflict or through the pursuit of in-group separatist leadership. In the case of
the former, competing political factions would be supported by Cold War rivals as a means of
draining the resources and probing the resolve of a foe and their satellites. There was also the
establishment of divergent economic and political systems, as well as international alliances,
which I term factional development. These developments resulted from splitting one state into two.
A recognized and accepted national border was created, thus splitting the group as well as the
state.
9
This led to competition for the mantle of the true representative of a nation and provided
rival system-level powers the opportunity to indirectly oppose their foes through proxies.
6

Sometimes, as with armed conflicts in Korea and Vietnam, this competition turned into interstate
war.
The armed forces of separatist groups have also engaged in intraethnic violence prior to
rebellion against the state as a means of uniting the group or establishing dominance.
10
States may
curtail such a development by intervening and pitting one side against the other. Ethnic defection
in the face of state power, in which one group splits and joins the state,
11
and manipulation by state
actors are ways in which intraethnic conflict can occur. System-level factional developmental
factors also offer an explanation for intraethnic conflict if factional development alters intragroup
information costs. Likewise, there are competing explanations for intraethnic rivals for why they
cease fighting and cooperate—all involve incursion by outsiders.
Although the Cold War is over, there are still instances of intraethnic conflict and rival
cooperation involving different forms of armed conflict. Some examples, such as the irredentist
Balkan Wars and conflicts in Nagorno-Karabakh, involve interstate warfare in which one state
seeks to assist coethnics outside of their boundaries. Others, however, such as the wars between
Chechen separatist factions, involve stateless nations in various conflicts amongst each other, as
well as with the host state. There are also surviving Cold War divided societies, like the two
Koreas.
However, newer politically relevant factions seem to exhibit elements of all of these types
of conflicts. The Kurdish factions of the Middle East and those of the Kanuri in West Africa—to
name just a few—seem to combine features of separatism, irredentism, and divided societal
competition in a manner that is not neatly addressed by existing explanations for the roots of
conflict. What these groups fight over, or why rivals may cooperate, are also the subjects of debate.
The major explanations for intraethnic conflict assert that organizational splits resulting from
7

individual leadership contestation, manipulation by state actors, and pressure from system-level
factors are the causes for intraethnic conflict. When might these explanations hold more utility
than others? Do divided stateless nations (DSNs) vary in their causes for conflict and cooperation
in comparison to single-state separatist (SSSs) factions that reside in a single state? Are they more
similar to divided irredentist groups (DIGs) that seek to join or add to an existing state? How about
the two-states, one nation (TSON) configuration? Is there also a discernible pattern for what rival
factions within an ethnic group fight over? Do they cooperate and why may they do so?
An argument for a categorical approach to the study of organized intraethnic conflict
I propose the study of the roots and processes of rival intraethnic conflict and cooperation
by creating a further disaggregation of conflict type: that of the DSN. I illustrate this distinction
through an analysis of four cases of factional conflict and cooperation based on permutations of
statelessness and division by an internationally recognized border. These cases include Kurdish
political factions (KPFs) as the DSN; Chechen separatist factions (CSFs) representing the SSS;
Albanian irredentist factions (AIFs) standing in for the DIG; and rival Korean state regimes
(KSRs) after the end of the Cold War for a TSON.
I advance an argument for considering intraethnic conflict categorically. Conflicts of
divided stateless nations are sufficiently distinct from those of separatists, irredentists, and divided
societies to be considered a category of their own. This work argues that DSN conflict originates
with system-level factors that cause factions to develop separately, like the TSON, but very
different from the SSS and DIG. However, DSN conflict is distinct from any of the other
recognized forms. For instance, conflict events focus on borders and regional centers, which is
similar to intra-SSS and intra-TSON conflict in some regards, but starkly different in others.
Additionally, DSN conflict is organized and coordinated at the level of the factions, unlike
8

violence within the DIG, which occurs at the individual level. Finally, rival factions within a DSN
will cooperate against outsiders—like the DIG, though cooperation occurs for very different
reasons. DSNs wish to compete with each other in the future, while the DIG has a representative
nation-state that can enforce a hierarchy among factions. The SSS and TSON do not unite against
outsiders once intragroup conflict occurs.
This work tests three propositions and their related hypotheses on the DSN, as well as on
the SSS, DIG, and TSON to determine whether differences are sufficiently distinct. These
propositions and hypotheses are built upon existing literatures of rival intraethnic conflict and
cooperation. They are formulated to answer the following research questions: do factions within
divided stateless nations experience intragroup conflict for the same reasons that other recognized
forms of intraethnic conflict do? What does this form of conflict look like in practice? Is intraethnic
conflict among rivals interminable or do rivals cooperate? Why would they cooperate or reject
proposals to do so?
These questions are argued to be important because of recent developments in global
affairs. Many of the most pressing humanitarian crises today involve ethnic groups that are divided
by national borders. The Rohingya and Uighurs are obvious examples of state-directed violence
against a divided group on ethnic and religious grounds. The Syrian Civil War and the Boko Haram
conflict also involve divided ethnic groups whose political factions have engaged in war against
each other, as well as with state and non-state actors.
Some factions around the world, such as the (divided) Pashtun-majority Taliban in
Afghanistan, have even established transnational administrative units that control territory and
population more efficiently than the states that claim said territory and population. The Taliban,
for instance, grew out of a Pashtun-nationalist movement that defeated other ethnically Pashtun
9

Mujahideen factions.
12
This faction—long considered the least attractive Pashtun faction as a
recipient for Western support—now controls the state of Afghanistan.
I intend for this project to contribute to the study of international security by arguing for
further exploration of cross-border intraethnic conflict and cooperation because of a spillover
effect resulting from this understudied phenomenon. Borders are often viewed as instruments of
peace,
13
while ethnic conflict is obviously destabilizing. Intraethnic conflict among newly relevant
groups clearly challenges the pacific nature of borders if ethnic groups become nationalistic and
revanchist over time.
The lens of New Wars: Population, territory, borders, and identity
Conflict fought with asymmetric means over ethnically based territory and population is a
feature of New Wars, rather than the conflicts between states that characterized older forms of
warfare. “New Wars” is a phrase coined by Kaldor (2012) that denotes a variety of warfare that is
distinguished from “old” conflicts by being fought over issues of identity, rather than ideology; by
being financed through means other than state-raised funds; by targeting populations through terror
more often than directly engaging military targets; and by involving both state and non-state
actors.
14
Mueller (2004) notes that this variety of warfare is often perpetrated by non-state actors
in concert with criminal elements of society. Like transnational criminal groups, emergent violent
non-state actors seek to engage in taxation and control of territory,
15
activities that are typically
the purview of the state.
16
Intraethnic conflict fits with the concept of New Wars as it is conducted
in order to pursue a faction’s stewardship of ethnic identity. It is also revisionist with regards to
the primacy of the Westphalian state, delineated by internationally recognized borders (IRBs), as
divided groups may challenge the legitimacy of such an institution.
10

Importantly, IRBs may not always constitute institutions of peace given the rise of DSN
conflicts. They may instead contribute to instability when politically powerful ethnic groups are
divided by them. As these groups develop pseudo-states possessing their own armed forces, they
may seek to expand their state-making activities in ways that will bring them into conflict with
coethnics and non-coethnics, as well as with host states and neighboring states. This is especially
salient given the global increase in subnational identity, as well as resurgence in nationalism more
generally.
17

This work will examine state-building activities, as well as populational-territorial control
in contested areas. Combinations of state and non-state actors are involved. It concentrates on the
post-Cold War period because of the assumptions of how and why New Wars are fought. This
work advances the study of the effects of borders by examining outcomes for ethnic groups
experiencing different permutations of statelessness and division. It examines intraethnic
dynamics within states and across IRBs for stateless nations and also for ethnic groups with a
representative nation-state. It will also advance the study of post-Cold War conflict by fleshing out
a category of ethnic conflict that is not yet distinguished—organized intraethnic conflict among
rival factions claiming to represent divided stateless nations, to include cooperation as a means of
continuing conflict at a later time.
A systematic appraisal of the types of ethnically related conflict may determine whether
the conflictual and cooperative actions of DSNs represent an increasingly common variety of such
conflict. I consider my contribution to be this; there is a discernible variety of conflict involving
DSNs that is distinct from each of the other three types. The roots of DSNs conflict are surprisingly
most similar to those of TSONs, rather than separatist groups. However, they are more like SSSs
with regards to some of their objects of contestation and unlike in others. Meanwhile, the manner
11

by which rival factions cooperate is completely distinct to any of the three; while both DSNs and
DIG rivals cooperate, they do so in very different ways and for different reasons.
Each of these four categories of conflict feature differing conditions with regards to
statelessness, division by borders, and factional development. These differences form the principal
reasons that this work makes the argues for treating organized intraethnic conflict as a categorical
phenomenon.
1.2 Limitations to current explanations of intraethnic factional conflict and
cooperation
Why do some factions within ethnonationalist groups fight? When will they put aside their rivalries
and cooperate? This section presents contributions and limitations to the existing literature. It
creates the basis for drawing hypotheses in the following chapter. The subfields that it covers
include the following: intraethnic ties—why coethnics experience generally pacific relations and
why they may not; the roots of intraethnic rivalry and the causes of conflict; and the explanations
for why rival dyads may cooperate or fail to do so. The objects of contestation fit more closely
with several hypotheses and the methodology selected for testing, and so the pertinent literature is
discussed in the following two chapters.
Intraethnic ties: Pacific effects of coethnicity, assumptions of group homogeneity, and
dimensions of ethnically related conflict
The baseline of intraethnic relations that we start from is one of amicability. Often, it is
inferred that coethnics tend to experience peaceful relations with each other. Information costs
between individuals are low because coethnics are known to the members of the group. Fearon
and Laitin (1996) inform us that members of the group are familiar or findable, and so coethnics
12

may benefit from enhanced trust when dealing with coethnics. Habyarimana et al. (2009) find that
ethnic groups distribute goods within the group efficiently for several reasons. First, individuals
within the group engage each other with a higher frequency than they do members outside the
group. Repeated interactions and the expectation of reciprocity foster trust through the expectation
of future benefit and mutual punishment for defection.
18
Second, they are able to locate each other
more easily because members are known to other members. Third, they communicate with fewer
barriers and greater predictability. In this perspective, cooperative relations are the norm because
repeated interactions and strong capabilities for communicative exchange should predict beneficial
outcomes for members of the same group.
For many ethnic groups, this is a reasonable set of assumptions to work from. However,
the potential weakness of this perspective is found in its implicit assumption of group
homogeneity. For instance, Brubaker (2004) argues that an ethnic group does not necessarily
possess universal goals and characteristics.
19
The experience of Afghanistan illustrates how clan-
and tribal-based loyalties suggest that cross-cutting allegiances can overshadow the larger ethnic
group’s supposed goals through social disputes.
20
In numerous Somalian conflicts, clan and sub-
clan allegiances provided obstacles to mobilization for both the Somali state, as well as for violent
Islamist actors like al-Shabaab and the Islamic Courts Union.
21

Such loyalties can make group cohesion difficult because enmity at the sub-clan level may
carry over into larger amalgamations of alliances. This could complicate the goals of what
Sambanis (2001) calls ethnic civil wars, in which the conflict is fought between ethnic
communities over power relations within a state. This work, however, examines state-making,
specifically the creation of a new state or state-like entity, or the enlargement of the same entities.
Thus, the categories of organized intraethnic conflict examined here will not involve interethnic
13

wars within one state as such. Nor will these categories automatically involve a traditional civil
war over control of an existing state framework unless it also fits one of these categories.
22

Goals may not be uniform for the groups that engage in different varieties of ethnically
related conflict. Carment (1993) finds that there are three basic types: wars of decolonization,
irredentist conflict, and separatist warfare. An additional type of conflict in which ethnicity is a
major component is that of divided states that are made of one ethnic group, such as the two
Koreas.
While wars of decolonization are largely concluded, the other three forms of conflict fall
within the framework of New Wars. Each of these varieties emphasizes a different goal. Irredentist
groups seek to join or add to an existing state, forging a “Greater X” territory composed of their
kin.
23
Separatist groups wish to achieve autonomy or statehood within an existing state.
24
Divided
societies pursue leadership over the ethnic group,
25
though rival regimes now use primarily
clandestine and asymmetric means, rather than directly engage in interstate warfare. Each variety
often includes a less-studied dimension of intraethnic conflict among groups that are neither united
nor monolithic.
The heterogeneity of ethnic groups can be problematic for lowered communication barriers
and information costs. The literature has identified how violent collective action can be used to
solve problems within an ethnonationalist separatist group as a means of achieving intragroup
leadership,
26
instead of providing goods more efficiently and aiding in collective mobilization.
27

Group size is a potential indicator for whether an ethnic group will witness intragroup violence,
with mid-sized groups at the greatest risk.
28
Small minorities lack the capability to mobilize force,
while large minorities are suppressed by the state. Politically relevant minorities are therefore
important to the body of research that addresses the earlier assumptions of group homogeneity.
14

Arguably, such a group will have to possess a certain range of relative populational, territorial, and
political influence.
Why factions fight: The roots of intraethnic conflict and dimensions of ethnically related conflict
There are three prominent explanations for the roots of intraethnic conflict: the presence of
leadership disputes; manipulation by the host state; or transnational and system-level factors, such
as competing alliance systems and incompatible political, economic, and socio-cultural
institutions.
Leadership contestation—a first-image explanation—is salient in explaining why factions
split. Asal et al. (2012) finds that leaders who collectively mobilize to use violence against rivals
are likely to cause a split within an ethnopolitical organization. Such splits could conceivably lead
to intraethnic conflict or warfare. Leaders within separatist groups also engage in intragroup
outbidding to mobilize support for their party at the expense of others.
29

Outbidding is a model of ethnic conflict in which leaders of multiple factions in one ethnic
group compete with each other to rally support for their endeavors by using appeals of emotional
attachment to the group’s identity. It is sometimes employed as a prelude to pursuing interethnic
conflict.
30
Individual leadership appears to be a potential motivator for intraethnic conflict across
factions, and not merely within them.
The state is a central second-image actor for explaining the occurrence of ethnic defection.
Coethnics pursuing intragroup leadership are sometimes viewed by the host state as effective
counterinsurgents.
31
A faction turning on coethnics seeks positional improvement relevant to their
rival and may be induced to fight coethnics at the behest of the state.
32
Ethnic defection in separatist
warfare occurs when one faction within an ethnic group turns on another faction in order to achieve
15

the goals of interethnic conflict.
33
A faction may engage in insurgent violence against dominant
coethnics
34
or may engage in warfare against other politically relevant factions with the aid of
dominant state forces.
35

The third-level explanation for why factions within ethnic groups engage in conflict with
one another comes from the international system. It includes transnational political, economic,
social, and cultural forces. Nation-states that were divided or created as two ideologically divergent
entities were split by IRBs and supported by competing patrons in the Cold War. These entities
also fought each other as proxies of those greater powers. The reason they did was because of
divergent development.
36
Over time, the two entities would create versions of their identity that
was incompatible with that of their rival. Some of these entities, such as the two Koreas and the
two Vietnams, engaged in direct warfare against each other with the assistance of coalitions of
allies that encouraged such divergent development.
In each explanation we see that such conflict can resemble interethnic violence under
certain conditions. It means that intraethnic conflict may be categorized with other forms of
ethnically related conflict, though this work does not assert that it is interchangeable with the
concept of interethnic conflict. Competition over the leadership of group identity is vital to rival
factions claiming to represent ethnic groups, whereas interethnic conflict involves competing
ethnicities, rather than competing identities within a single group. Nevertheless, ethnic identity
remains a salient component. Thus, intraethnic conflict should be considered a dimension of
ethnically related conflict.
A major contribution of this work is to determine whether politically relevant factions
representing a divided stateless nation share similar roots with one or more of the three recognized
forms of organized intraethnic conflict. If politically relevant factions within a stateless and
16

divided ethnic group engage in state-making activities, it is possible that a border may freeze the
state-making process and later contribute to intraethnic competition if conditions on either side of
the IRB change. Identification and explanation of this variety of intraethnic conflict is one gap in
the literature that this work aims to fill through a systematic comparison to recognized types.
Division by borders and rival intraethnic state-making activities
Although there is evidence of intraethnic conflict, intraethnic cooperation occurs as well,
often in the midst of interethnic conflict. Say, for instance, a population belonging to a particular
ethnic group is besieged by outsiders in one state. A neighboring state may be comprised primarily
of the threatened group. That state may then opt for augmenting their offensive capabilities, rather
than their defensive ones, and cross borders to intervene in order to rescue their “island” of
coethnics.
37
Carment et al. (2006; 2009) found that ethnic affinity influences which states may
become involved in such a conflict, with shared ethnic identity being a strong predictor of
intervention.
This variety of irredentist ethnic conflict could be sparked by active state oppression of
minorities, causing coethnics to come to their aid.
38
Conversely, outbreak of ethnic conflict may
be initiated by a state’s weakness; its inability to protect minorities from non-state actors may
require victimized groups to seek out the assistance of coethnics in neighboring states.
39
The
Yugoslav Wars witnessed many such cross-border incursions to assist coethnic factions in
neighboring states. Contrary to the processes leading to ethnic defection in some separatist wars,
factional interactions were more cooperative within irredentist groups that exhibited a
representative nation-state.
17

Irredentist movements provide an opportunity to closely study an element that is
underexplored in the ethnic civil war: the effects of borders. Simmons (2005) argues that borders
are viewed as guarantors of stability and trade when involved parties agree on their demarcation
and place trust in their counterparts’ ability to control the areas within their respective jurisdictions.
When states accept the territorial status quo, they benefit from lasting peace.
40
States at peace that
maintain borders that are “mutually accepted as legitimate”
41
experience greater prospects for
stability and profitable trade relations.
42
Stable borders decrease the likelihood of war between
members of a dyad, even when non-democracies are involved.
43

A hope for unification with coethnics may help prevent interethnic rivalry from evolving
into a “protracted conflict”
44
at the border, in which a termination of violence is unlikely.
Additionally, a peaceful transfer of territory may reduce the probability of continued hostility. One
such territorial transfer, Israel’s return of the Sinai Peninsula to Egypt, is regarded as having been
instrumental in ending conflict and establishing ties between two former rivals.
45
And so, borders
have been a primarily positive force for states.
Barth (1969) informs us that boundaries consolidate ethnic identification with the state
among otherwise-changing ethnic groups. However, with irredentist movements, this
identification is either nonexistent or it is not strong enough to overcome the identification with
the group over the state if sizable irredenta inhabit a state in which they are not part of the core
community.
There are other cases when imposed borders have divided ethnic groups into two states.
These states have rarely experienced peaceful relations. The two states, one nation phenomenon
involving the two Germanys, the two Vietnams, the two Koreas, and China/Taiwan brought about
18

dyads that were hostile to each other, in which wars, border skirmishes, and cross-border political
interference were somewhat frequent occurrences.
Zhao and Sutter (1991) find that the realignment of political and economic systems, as well
as the influence of allies, cause enduring rivalries to take hold among these dyads. Rivalries come
about due to competing networks of alliances that fostered conflicting economic and political
systems, as well as differing social and cultural influences.
46
Though there have been wars between
divided societies, interference, such as arrests of a rival’s sympathizers and the creation of political
and social movements in favor of the rival within the other’s territory were the most commonly
employed methods of political competition. Even economic integration could be used as a means
of exploitation by making the rival dependent on their enemy, thus creating political leverage that
can be exploited, as China is argued by some to do in its relations with Taiwan.
47

In the Cold War cases, it was the division by a border that preceded the conflict. This has
important implications for stateless ethnic groups if they too are divided by a border. Statelessness
is only indirectly studied in some literature on separatism.
48
Some factions within ethnic groups
do not need to be divided by a border in order to experience intraethnic conflict. Nevertheless,
those groups also seek control over the group’s institutions.
State formation is an ongoing process in much of the world. The Weberian (2015) state has
long been viewed as the only actor that can claim legitimacy in exercising force within a defined
territory. Tilly (1985) argues that a state is born through violence: both through forging its borders
with external actors and through the elimination of domestic rivals. Formation is deemed
successful if rulers turn to governance and protection of civilians instead of raiding for resources.
49

Politically relevant factions that claim to represent stateless nations may represent an increasingly
frequent vector for low-intensity conflict in the future, both within ethnic groups, as well as with
19

other groups and with states, as some of them seek to pursue more expansive forms of self-
determination and state-making. Those that become exceptionally powerful may even become
capable of effectively waging conventional warfare against rivals and states as we will see in the
Kurdish case.
Rival cooperation: Out-group incursions, factional fragmentation, and the institutional balance
of power
To sum up so far, there are three recognized varieties of warfare in which organized intraethnic
conflict can occur: separatist warfare, irredentist violence, and the conflicts between divided
societies. There are also three prominent explanations for the roots of intraethnic conflict. These
include leadership disputes and personality clashes, manipulation by state actors, or the influence
of transnational and system-level factors, such as alliances and division by borders, on divergent
development. Those explanations broadly correspond to the first, second, and third images,
respectively.
Now that we have discussed the roots of intraethnic conflict, we must turn to the cessation
of conflict. Rival ethnonationalist factions are not always in conflict with each other. Sometimes
they cooperate, despite their mutual antipathy. The study of rival cooperation should not be
separated from a systematic study of conflict, as this provides interesting variation that can better
distinguish these varieties of intraethnic conflict from each other.
There are three explanations for the phenomenon of rival cooperation that have emerged
from three generations of research. The first posits that out-group attacks force rival factions to
band together to avoid being conquered. Simmel (1955) observed that out-group invasion alone
could foster feelings of emotional solidarity against the outsider. As this body of research
20

progressed, it came to be understood that unity could result in a “rally ‘round the flag effect” that
could be employed as a means of reducing in-group opposition to the faction or its leader.
50
If this
is the case, the potential roles of smaller factions could be negated. This is because they lack the
ability or capacity to mobilize its members to address any grievances they may have with major
factions within the group.
51
As they are unable to resist the major factions, they would be required
to unite, despite any intraethnic tension that would otherwise preclude unity. Survival would trump
all else.
Using the lens of the first generation of research, we might expect that rival factions should
unite when confronting any threat from an out-group, even for cynical reasons. However, this is
perhaps overly reductionist. That criticism led to the development of a second generation of
research, which asserted that intervention by an out-group can instead contribute to factional
fragmentation, rather than unity. This view came about by examining the costs of mobilization for
factions considering uniting with rivals. As the costs cannot be borne equally by each faction, some
may be averse to cooperation because they stand to lose more by throwing in their lot with the
group if it will lose against an out-group.
52

This foundation of the second generation was built upon finer-grained exploration of
capabilities and power relations at smaller scales. This provided more nuance and local context to
what was known and assumed about conflict and cooperation among rival factions at the micro
level of conflict.
53
A complex web of competing loyalties, local complexity, and opportunities for
intraethnic conflict were revealed. This was the case even when factions faced a common enemy.
54

Kalyvas (2006), for example, found that local actors employ selective political violence to contest
territorial control in civil war, even directed toward coethnics.
21

Simply because a given conflict had an ethnic component did not necessarily mean it could
be classified as a strictly interethnic war based on the findings of this generation. Group solidarity
was not automatically required for survival. Conflicts with an ethnic component could feature a
wide array of ethnic and ideological dimensions, some of which could be in competition with each
other.
55
Collaboration is often encouraged by the out-group, with some factions siding with an out-
group or avoiding conflict for material gain, or to be spared violence. Lyall (2010) informed us
that some factions may even be sought out by invading forces to induce defection from the
ethnonationalist movement. In other words, coethnics could be sought to combat their ethnic kin
on behalf of the out-group force.
The second generation of research eschewed the Manichean view of in-group and out-
group hostility that was prominent in the first generation of research. A third generation of research
has since expanded upon the second. This generation focuses on how out-group pressure might
affect factional unity or fragmentation at the institutional level. Out-group forces may
accommodate or repress the institutional aspirations of rival rebel groups. If the out-group chooses
to support one particular faction, the result might strengthen the organization internally, making it
a more robust competitor to other factions. Additionally, by switching support to a rival faction,
or supporting two rivals simultaneously, Tamm (2016) finds that out-group intervention could
instead bring about fragmentation within the ethnonationalist movement.
Additional factors, such as various transnational alliance constraints, transnational ethnic
kin relations, the structure of regional power, and factional diversity can influence fragmentation
or unity. Seymour et al. (2015) find that the pressures these factors place on the institutions of rival
factions disrupt attempts at unity. It is logical to assume that alliances can hinder cooperation if
different factions have antipathetic patrons in the region or international system. Additionally,
22

ethnic kin in neighboring countries can cause fragmentation by introducing new institutions to an
area. The introduction of new institutions is key to this study of conflict and cooperation,
particularly as far as divided stateless nations are concerned.
Group members may be satisfied or dissatisfied with the institutional arrangement of power
among the factions that claim to represent the group. McLauchlin and Pearlman (2012) find that
satisfaction or dissatisfaction with this arrangement determines when ethnonationalist movements
experience fragmentation or unity when faced with out-group pressure. Satisfaction with the status
quo should lead to unity, whereas dissatisfaction should drive fragmentation. In other words, when
an out-group transgresses upon populations within the group, it can have different results. If the
group’s factions were content with how power is shared, institutionally, prior to out-group action,
the factions are likely to work together against the out-group. However, if one or more faction is
dissatisfied with the distribution of institutional power before out-group pressure is applied, they
are likelier to fragment because there may be an opportunity for a faction to increase their post-
repression share of power.
Transnational ethnic kin who introduce competing institutions to an area occupied by the
group—but controlled by a different faction—may be dissatisfied with the institutional status quo.
They certainly add new pressure to the faction that previously did not contend with another
faction’s influence on that territory.
56
However, cross-border groups may be invited or at least
allowed to operate in such territory.
The subject of invitation or permission gives us an unresolved question in the third
generation of research that this work aims to answer. What if that dissatisfaction with the status
quo contributes to cooperation, at least temporarily? To introduce their institutional alternative,
the rival faction must either invade or be invited. Why might they be invited? It is precisely this
23

conditionality that the cooperation section of each case aims to probe. In brief, the fear of losing
the territory inhabited by the group to an out-group may be greater than animosity to the rival. It
would be preferable in such instances to cooperate for a time so that competition at a later date
remains possible.
In this work I combine insights from each body of research into rival cooperative dynamics.
Specifically, I build upon the third generation of research, though the earlier generations are not to
be discounted out of hand. In-group solidarity, local contextual nuances, and institutional
arrangements are all salient here. Ethnonationalist movements are often a form of state-making.
Factions that represent ethnonationalist movements may combat out-group threats to protect their
population and territory in their endeavors to create state-like entities.
57
Rival factions may develop
incompatible institutions if they are divided by borders. Despite having mutually exclusive visions
for a political-territorial entity, they must nevertheless compete for influence among the same
people.
58

I explore periods of intraethnic conflict that experience out-group invasions. These
invasions feature surprising local nuances and institutional competition, thus requiring all three
generations to be considered on their own terms. My goal is to tease out conditionality for when
one generation of research may hold more predictive power for judging if and when rival factions
may cooperate. This should assist in disaggregating by the four categories of organized intraethnic
conflicts involving DSNs, SSSs, DIGs, and TSONs.
This work adds populational and territorial aspects of competition to the institutional
component of factional competition originating with the third generation of research. Studying
changes to the institutional status quo may offer insight as to when factions may unite or not.
59

Ethnonationalist factions that seek to implement their vision of a political-territorial entity must
24

convince coethnics that their vision is superior to that of their rival. They may seek to introduce
their institutions to the territories controlled by a rival to accomplish this task. Thus, the ability to
alter the populational and territorial institutional status quo in that faction’s favor may provide a
reason to cooperate if faced with out-group pressure; the loss of that community may decrease
trust in the institutions of either rival should the community and its territory be lost to the out-
group.
In this regard, the first generation of research is not entirely without merit. Solidarity may
well result from an out-group assault. However, it is not simply because of feelings of ethnic
solidarity; it is because of political calculations and the desire to further the faction’s institutional
share of influence among the in-group in newly accessible territory. Likewise, the second
generation demonstrated why factions may not cooperate, but was less clear on why factions will
cease their infighting and unite. The third generation has begun this important work, as this work
seeks to create a fourth generation of research.
1.3 A brief depiction of the argument and its contributions
This work asserts that intraethnic conflict is best studied as a categorical phenomenon. Conflicts
within DSNs are distinct from those within SSSs, DIGs, and TSONs. Division by borders and
statelessness (or the lack of division and/or statelessness) influence the origins of organized
intraethnic conflict; the objects of contestation between factions; and whether and why rival
factions may cooperate or not. This work’s central argument carries implications for literature, for
our research methodology, and for policy.
Dividing a stateless nation by borders encourages divergent development, which
potentially alters the information and transaction costs
60
involved in dealings with coethnics.
25

Irreconcilable political, economic, and socio-cultural institutions can come about. Factions
claiming to represent such a DSN may nevertheless compete for leadership across borders through
violence, incorporation of non-aligned members of the community, forced population transfers,
and/or superficial cooperation.
A period of competition may reach a stalemate because evenly matched factions may not
be able to overcome their rival. Attempts to conquer regional centers may stall. De facto borders
may be established. However, an exogenous shock, such as a defeat of one or more host states in
interstate war, or civil war, or other events may occur, thus allowing for out-groups to take
advantage of a security vacuum. Incursions by outsiders against the DSN force the factions to
decide whether to unite or allow the afflicted faction to face the threat alone.
It is the out-group’s targeting strategy that appears to influence the outcome. An out-group
that purposefully kills unaligned group members in territory that is desired by rival factions can
drive these rivals to unite. Those groups would prefer to compete with coethnics at a later time
than to lose the territory and its people to an out-group. The state-making activities of defining
territory and defending its populations are vital for any would-be factional leader that wishes to be
seen as legitimate by their coethnics.
Targeting rivals, even non-combatants that are nevertheless members of rival factions, will
not bring about cooperation. Without the loss of territory and population, the unaffected rival can
still retain an advantage over the affected rival in the future. This work’s theory is the subject of
Chapter Two.
This work’s contribution will be threefold; it has implications for theory, methodology,
and policy. First, theoretically, it builds upon existing research and provides a potential explanation
26

for this understudied phenomenon. It demonstrates against expectations that intra-DSN conflict is
best explained by system-level factors, rather than by individual- or state-level ones. Different
objects are contested by DSN factions than by other factions involved in other categories of
organized intraethnic conflict. DSN factions also cooperate for different reasons than others, thus
improving what we know about the utility of three generations of research. An older generation is
shown to retain its utility in some regards, while gaps in the second are filled in and a third is
expanded, setting the foundation for a fourth generation.
The second contribution is methodological. Though ultimately a product of mixed
methods, this work is novel in that it utilizes spatial and temporal analytical methods to assess
conflict and cooperation using numerous sources of armed conflict data that solve spatial statistical
problems commonly encountered in political science and international relations literature.
Factional competition, conflict, and cooperation are assumed to occur in areas populated
by the coethnics of the group in question. It would not make logical sense to compete for or assist
non-coethnic populations in terra incognita. Therefore, factors of space and time, such as distance
from the border, distance from regional centers, and the locations of cooperative or conflictual
interactions among factions are argued to be salient features for testing this project’s propositions
and hypotheses. The methodology will be explained in detail in Chapter Three. In brief, it develops
a generalizable, analytically eclectic framework that employs mixed methods. It may be used for
any case of all four categories of organized intraethnic conflict.
Finally, this work holds policy implications. As described above, decisions for supporting
a faction within a DSN have consequences. Consider the example of Afghanistan given earlier.
The US supported Pashtun factions who were more compatible with what American policymakers
envisioned for Afghanistan. By doing so, the US selected factions who could not create an effective
27

state. Pakistan, by contrast, selected a nationalistic faction that was remade into an Islamist one.
Pakistan was able to partially direct the growth of what came to be the Taliban, the DSN faction
that eventually triumphed against the Afghan state. Afghanistan must live with the consequences
of poor decision-making in the US, which could have had some influence on proto-Taliban groups
had they chosen a strategy of supporting realistic winners, rather than one of overly idealistic
nation-building.
61

A second policy-relevant contribution will be the development of a spatially referenced
database that can be expanded to include data for all ethnic groups, stateless or not, to serve as an
early warning predictor for conflict potential. This prediction would be based upon a group’s
capacity for political mobilization and disposition toward others. These implications are elaborated
further in Chapter Nine.
Though state failure is relatively rare, there are more states with cause for concern than
very sustainable ones.
62
If factions within divided stateless nations develop a conflictual identity
or consciousness based on ethnicity, they may mobilize for intragroup leadership. We may expect
to see intraethnic violence increase. It may cross borders more frequently if DSNs experience
divergent development in times of state fragility. We should recall Blainey’s (1973) assertion that
war is the natural human order, and we live in an aberrantly peaceful time. In an era where
identification with the state is often declining, DSN expansion and conflict may well become more
likely.
1.4 Overview of the research design
This work advances our knowledge of organized intraethnic conflict and cooperation among rival
factions by identifying a distinct variety of this phenomenon. It does so by establishing the current
28

state of knowledge with regards to why it happens and draws propositions and hypotheses from
the existing literature. It then develops a plan for assessing these hypotheses through quantitative
and qualitative means.
The three propositions and their ten hypotheses followed from the literature through both
induction and deduction. They pertained to the following research questions; first, do the roots of
intraethnic conflict among rival factions within divided stateless nations correspond more closely
with the roots of conflict for any of the other three varieties? Second, do the conflict objects of
divided stateless nations resemble those of some varieties more than others? Third, will rival
factions within divided stateless nations cooperate for reasons that correspond to those of the
recognized varieties?
The dependent variables that were relevant to these questions included conflict onset for
the first question; instances of armed conflict corresponded to the second question; and cooperative
action for the third question. Using the same order of questions, explanatory variables were the
existing explanations for rivalry, geographic proximity of key points of interest, and targeting
strategies by out-groups.
To answer the above questions, I used an analytically eclectic design.
63
This work
incorporated findings and methods from diverse theoretical traditions to test its hypotheses. The
work tested existing theory and developed new theory using both statistical and qualitative
analyses in a sequence that was applied to each case.
First, it statistically summarized the spatiotemporal extent of territory inhabited by ethnic
groups over time to identify a starting point for qualitative assessment of what factors may have
accounted for the roots of factional rivalry. This was accomplished through map processing, in
29

which a historical demographic map was georeferenced, transformed, and compared to recent
datasets on ethnic preponderance. Map processing demonstrated approximately how much
territory had been gained or lost since the imposition of IRBs. More important than precise
quantities of land, it depicted where changes in population and territory occurred. This stage
provided useful clues for process tracing. Potential mechanisms for the competing explanations
for the roots of rivalry were supported or refuted for each variety of conflict.
To explore how intraethnic conflict occurs, I combined standard statistical methods and
spatial analytic techniques with qualitative research to determine what points of interest may drive
the types of conflict. Models included distance from national capitals, regional population centers,
borders and border crossings, and primary highway systems as predictors of conflict density.
Pearson correlation tests, linear modeling, and logarithmically transformed ordinary least
squares regression models were used to identify whether there were interesting relationships that
could be identified in the data. If so, they were worth exploring in greater depth. Moran’s I tests,
Getis-Ord General G and, if necessary, Gi* tests were applied to see whether the data were
autocorrelated and clustered. These tests could ascertain whether spatial relationships in the data
existed but did not necessarily establish any causality. If autocorrelation and clustering were found,
a spatial weights matrix was created for use with a spatially lagged autoregression model with
maximum likelihood estimation that accounts for the special nature of spatial data. These tests
examined the relationships between covariates, rather than simply finding relationships within the
data. Robustness checks, such as a studentized Breusch-Pagan test, were also incorporated to
further improve the confidence in any inferences that were made. Factional material was also used
to support statistical findings.
30

When rival cooperation was identified I applied several spatial analytic techniques and
qualitative research to identify mechanisms that could explain why it materialized, or why it did
not. These tests began with kernel density estimation to examine where cooperative action
occurred. That was then subjected to K- and L-functions analyses to assess how far apart instances
of cooperative action could take place before they would be considered unrelated. An average
nearest neighbor analysis also tested for whether some of these instances could be considered
insufficiently related.
The preceding tests for cooperative action determined whether the data were related to each
other. This refers to second-order spatial effects, in which data are spatially dependent. First-order
effects refer to observations that result from underlying spatial processes. To assess these effects,
cooperative action was tested by a series of non-stationary Poisson point process models of
covariate effects, which include the use of a likelihood ratio test. These analyses assess whether
cooperative action could be due to one of the following underlying covariates with a spatial
component: deaths of coethnics have no effect—the null hypothesis; deaths of unaligned
coethnics; deaths of rival coethnics; and deaths of all civilians, regardless of political affiliation.
In each case, I created analytical maps to display the extent of the conflict and the intensity
of conflictual or cooperative events. These were useful for identifying further points for qualitative
research. They also provide the reader with a frame of reference for the overall spatial extent of
each case.
Qualitative research factored heavily into the explication of theoretical mechanisms in each
case, as context is so important when working with spatial data. Government, intergovernmental,
and NGO sources were useful for understanding the contexts of these periods of conflict and
cooperation. Factional media sources and interviews were beneficial in understanding the
31

motivations involved in whether intraethnic conflict or cooperation occurred. These served to
reinforce the statistical evidence in a manner that quantitative analysis alone could not capture with
the same richness that is offered by using mixed methods.
The results of many of these analyses are located in the appendices for each case to avoid
overburdening the reader with tables and figures. Generally speaking, the results of the temporal
analysis, the spatially lagged autoregression model, and the Poisson point process model appear in
the case. Other analyses will be briefly remarked on in the case chapter but displayed in the
corresponding appendix. Analytical maps are placed in the case chapters for each period of conflict
or cooperation. A detailed description of the design and a comprehensive evaluation of the data
used in this work is the subject of Chapter Three.
1.5 Preview of findings
The results of each case are briefly detailed in their respective case chapter. They are also the
subject of Chapter Eight.
In brief, DSN conflict occurs due to system-level factors, which makes them surprisingly
more similar to TSONs, thus suggesting that division by IRBs is more salient than statelessness.
Individual- and state-level factors are more closely associated with DIG and SSS conflict origins,
respectively.
We might reasonably expect DSN conflict origins to more closely resemble SSS and DIG
given their military capacity and similar levels of developments compared to advanced states like
TSONs. Moreover, the DIG is also divided, yet their intragroup violence has no system-level
influence. And so, both division and statelessness must be taken into account when explaining
organized intraethnic conflict.
32

DSN conflict objects themselves resemble a cross between SSSs and TSONs, but with
noticeable distinctions. The border is salient for DSNs and TSONs, but the regional center is salient
for DSNs and SSSs. Other objects are not shared between them. The findings with regards to one
particular object, primary routes, are inconsistent at present, but strategies for improvement are
offered.
Like DIGs—and unlike SSSs and TSONs—DSNs do cooperate with rivals against out-
groups. However, they do so for different reasons. DSNs cooperate when unaligned coethnics in
competitive territory are killed by outsiders. DIGs cooperate because there is a single
representative state that can hold disparate factions together, or at least prevent open warfare.
Cumulatively, these findings suggest that a categorical approach is preferable to viewing
organized intraethnic conflict as a subset of separatist or irredentist infighting. The differences are
expected to remain consistent across other cases of these categories. The results and strategies for
further studies are found in Chapter Eight and Chapter Nine.
1.6 Overview of chapters
In this introductory chapter I presented a sketch of the work as a whole. I identified a social
scientific puzzle, framed the literature, and provided a brief description of a theory of intraethnic
conflict and cooperation among divided stateless nations. Potential contributions were elucidated.
The research design that will be utilized in the case chapters was summarized, as were some key
findings.
The remainder of this work proceeds as follows; Chapter Two contains the theoretical
framework developed for studying the phenomenon of organized intraethnic conflict. I outline the
key concepts of this work, draw hypotheses from the literature to test on the four categories of
33

intraethnic conflict, and present a theory for conflict and cooperation among rival factions claiming
to represent divided stateless nations. Scope conditions are provided, and alternative explanations
are considered in Chapter Two.
The research design is fully detailed in Chapter Three. A justification for analytical
eclecticism with the use of mixed methods is offered. The key concepts from Chapter Two will be
operationalized and I will discuss the data used for testing the propositions and their associated
hypotheses. The plan for the case chapters is laid out in the sequence that the tests were performed,
which were briefly described in Section 1.4. Chapter 3 also elaborates on the strategy for case
selection that was used.
Chapters Four through Seven present the case studies of the types of intraethnic conflict.
The Kurdish factions, our divided stateless nation, are the subject of Chapter Four. Single-state
separatists are represented by Chechen factions in Chapter Five. I present Albanian factions in
Chapter Six. This group stands in for divided irredentist groups. Chapter Seven concerns the two
Koreas—the two state, one nation phenomenon that has outlived the Cold War.
The findings across cases will be compared systematically in Chapter Eight. Implications
for the literature, for our methods, and for policy are found in Chapter Nine. Chapter Nine also
concludes with thoughts on ongoing work, possible extensions, and suggestions for developing
broader research programs for future endeavors

34

Chapter Two
The Conflict and Cooperation Theory: Explaining patterns of behavior among rival
ethnonationalist factions

The last chapter covered a broad overview of this work. Several limitations of existing literature
were identified and the potential for contribution was established based on the development of a
theory of organized intraethnic conflict. This chapter is the exposition of that theory, entitled the
Conflict and Cooperation theory. First, I define and conceptualize key terms that will be used
throughout the analyses in Section 2.1. These are further constructed and operationalized in
Chapter Three. Second, I present the hypotheses that are drawn from existing literature in Section
2.2. Third, I propose a theory of intraethnic conflict among rival factions of divided stateless
nations (DSNs) in Section 2.3.
2.1 Defining and conceptualizing organized intraethnic conflict and cooperation
This section lays out the ontology of key terms and concepts associated with organized intraethnic
conflict and cooperation. The unit of analysis for intraethnic group dynamics will be the politically
relevant faction, which is analyzed by dyadic rivals. Combining insights from Asal et al. (2012)
and Warren and Troy (2015), I define the politically relevant faction as a hierarchical ethno-
political organization that possesses the potential to mobilize armed forces and maintains relations
with regional or international powers, thus making it, at a minimum, a regionally relevant political
actor. It may control a state or a territory in which it is a state-like entity. Also, it may aspire to
create either.

35

Intraethnic conflict is an umbrella concept that describes violent activities associated with
rival factional behavior. Such conflict necessarily includes an exchange of kinetic force, whether
or not fatalities are sustained. Besides organized armed conflict, it may involve violent interference
into the domestic affairs of the governing institutions of a rival faction. Such interference may take
the form of the killing of members of rival factions at checkpoints; car bombings; burning or
otherwise physically destroying bases of operation and political party headquarters; infiltration of
rival territory; or abductions, enforced disappearances, and arbitrary detentions of members of
rival factions.
By contrast, cooperation is the occurrence of cooperative action between two or more rival
factions. Cooperative action is mutually supportive combat against an out-group that poses a
threaten to the in-group. It is conducted by the armed forces loyal to the factions involved.
Permission should be given to a rival that wishes to establish political institutions
64
or a military
presence in territory in which they would not otherwise enjoy access. Thus, establishment of a
military base of operations that is permitted by the rival is considered cooperative action as well.
Rivalry is a condition that exists when two ethnonationalist factions claim to represent the
same ethnic group if they share a history of antagonistic relations that are likely to experience
armed conflict.
65
The ethnic group represented by the factions is a distinguishable group of people
who believe in common descent.
66
Belief is from the faction’s perspective, not necessarily the
community that the faction seeks to influence. Interviews with members of these communities
revealed that such a belief was sometimes dependent upon personal affiliation with a faction, or
the lack thereof.
The ethnic group that is the subject of study is also referred to as the in-group. An in-group,
however, is not used interchangeably with the faction. All members of the ethnic group are

36

considered to be within the in-group, but not necessarily members of the faction.
67
The out-group
is made up of the armed forces of actors that are not primarily composed of the in-group.
68

Members of the in-group who are part of the faction’s political party, armed forces, or areas of
historic control are considered aligned or affiliated with that faction. Conversely, members of the
ethnic group who inhabit territory outside of a faction’s control or in territories of recent control
are considered unaligned or unaffiliated if the population does not give widescale endorsement to
the legitimacy of the institutions of the relevant factions. Disposition toward factions was
established through third-party state, IGO, and NGO reports, as well as interviews with members
of the group.
As described above, the faction is a politically relevant ethnopolitical organization that can
collectively mobilize members of the group for violence and has a degree of influence on state and
regional political outcomes. To be considered rival factions, these groups must have an
oppositional relationship, though the nature of opposition may change from dyad to dyad. They
may hold irreconcilable visions for an ideal state or state-like entity that would represent their
people.
These differences can be arranged along political, economic, social, or cultural fault lines.
For instance, two rival factions may split based on political governance if one wishes to lead a
state based on Maoist confederalism and another prefers a hereditary authoritarian system.
Likewise, Marxist or market economic systems may be a point of contention between rival
factions. Social differences may result from stark divides over the manner by which factions
leverage transnational social movements. The cultural differences may rest along urban-rural
divides, tribal ties, or religious differences. However, rivalry may also occur if there are leaders
that dislike each other, despite being similar in other respects.

37

Categories of organized intraethnic conflict
These differences may not manifest equally depending upon the category of organized
intraethnic conflict being studied. As described in the preceding chapter, established varieties with
a recognizable element of intraethnic conflict or cooperation between rival factions include
separatist wars of secession, irredentist conflicts, and disputes between divided societies.
Separatist wars of secession are periods of armed conflict that are initiated by ethnonationalist
groups dissatisfied with the domestic political situation of the state in which the group resides. It
is waged against the state,
69
though there may be intraethnic warfare if the state successfully
induces ethnic defection, in which one faction turns against its coethnics to fight on the side of the
state.
70
A secessionist group wholly or mostly contained within one state are termed a single-state
separatist (SSS).
The state is an administrative unit that monopolizes the legitimate use of force within the
territorial boundaries it claims to hold jurisdiction over.
71
It protects its inhabitants from external
and internal threats, raises revenue, and provides a means of exchange.
In contrast, the nation is the ethnic group. Nation and state are not used interchangeably.
Many separatist factions belong to stateless nations. I define a stateless nation as an ethnic group
that forms the core community of no state recognized by the broader international community.
72
I
do not argue that all stateless nations pursue statehood, nor do I normatively argue for the granting
of statehood to all stateless nations. Some stateless nations do not aspire to self-determination at
all, while others are content with power-sharing or varying degrees of autonomy.
73

There are cases in which an ethnic group may make up the core community of one state
but have coethnics residing in a state that has a different ethnic group recognized as making up its

38

core community. Such an arrangement is instead a divided irredentist group (DIG) if it wishes to
be incorporated into the greater territory of a neighboring state, or if the neighboring stat wishes
to incorporate it.
Irredentist conflict occurs when an “ethnic island” is threatened by an out-group, causing
the state in which the threatened group forms the core community intervenes to protect the island.
74

Additionally, irredentist conflict can occur if the islands, or irredenta, attempt to join the existing
state or create a parallel state with the assistance of the existing state, whether or not the armed
forces of the state are directly involved. Indirect transnational intervention by the armed forces of
the state, such as training or arming of the irredenta, is nevertheless a component for delineating a
threshold between separatist and irredentist conflict in this work.
The third recognized variety of intraethnic rivalry concerns the divided society.
75
This
involves the division of one state into two along ideological lines. This feature of the Cold War
was a major component of the balance of power in a bipolar international system. East and West
Germany; North and South Korea; North and South Vietnam; and Communist China and
Nationalist Taiwan are examples of the two states, one nation (TSON) phenomenon. Conflict was
driven by system-level factors, such as alliances, political ideology, and economic systems. Each
state competed for the mantle of leader of the ethnic group. Some of these dyads have outlived the
Cold War. This phenomenon endures despite the end of bipolarity and the advent of New Wars.
76

Interstate warfare is declining, and New Wars are now the most commonly occurring types
of armed conflict. First and second image factors of so-called greed and grievance
77
are often
theoretically powerful when explaining these conflicts. However, surviving instances of Cold War
competition and proxy conflicts illustrate that the third image can still be a salient factor in
explaining intraethnic conflict in some instances, particularly when proxy conflict is involved.

39

Ideology is not completely removed from New Wars. In some cases—such as the TSON
phenomenon—irreconcilable political, economic, and socio-cultural differences become hardened
by division. Their development in these realms is influenced by this division and by the faction’s
transnational linkages to the regional balance of power and the international system. Thus, most
conflict is international in some regard.
78

One well-established New War variety of ethnic conflict is irredentism.
79
Like the two
states, one nation phenomenon, the irredenta are separated from the “Greater X” political entity by
an internationally recognized border (IRB). The IRB is a demarcating line that allocates
geographic ownership to states.
80
The nature of division by borders may change for each variety
of conflict described in this text. For instance, SSSs are “contained” by the state and are not split
by borders. The TSON arrangement is divided into two states that exercise sovereignty over their
domestic and international affairs. DIGs are divided by borders, but one segment of the population
forms a core community of a nation-state, while other segments are situated outside of the territory
of that nation-state.
I propose the recognition of intraethnic conflict and cooperation among rival factions of
divided stateless nations as a category of ethnically related conflict. A DSN may seek some form
of self-determination within one or more states in which it resides, like the SSS. However, they
may have state-like entities that can attempt to assist or influence ethnic islands on the other side
of the border that divides them into separate populations, similar to DIGs, but dissimilar to the
SSS. Additionally, DSNs must contend with system-level factors that can influence divergent
pathways of political, economic, and socio-cultural development like the TSON. However, the
nature of division and statelessness vary across all of these groups, and so they will likely behave
differently from each other if a systematic study is carried out.

40

This work categorizes these types of conflict and determines to what extent DSNs share
roots of conflict to the other varieties or not. It also ascertains whether the ways in which rival
DSN factions fight each other or cooperate are similar to the recognized varieties.
2.2 Propositional pathways of conflict and cooperation for rival DSN factions
Now that we have examined key concepts and terms in section 2.1, let us discuss the
pathways through which they may interact. The pathways take the form of three primary
propositions. Each proposition contains hypotheses that make up more precise components of the
whole. These hypotheses allow DSNs to be tested in a manner that compares the extents to which
they may differ from SSSs, DIGs, and TSONs. DSNs are argued to exhibit patterns of conflict and
cooperation that constitute an as-yet unrecognized category of ethnically related conflict.
Alternative explanations will be drawn from competing levels of analysis; theories of contestation
in war; and earlier (often discounted) generations of literature, respectively.
An analytically eclectic approach
This work utilizes analytical eclecticism as proposed by Sil and Katzenstein (2010). This
approach seeks to draw together seemingly disparate bodies of literature when the phenomenon of
interest does not fit squarely within a single established field. Analytical eclecticism is not a
replacement for other modes of research, but is rather a pragmatic, widely applicable, and complex
(rather than parsimonious) intellectual complement to a given research program. It is intended to
solve issues of real-world relevance through the production of policy-relevant middle-range
theory.
Middle range theory stands at the confluence of theoretical and empirical research.
81

Analytical eclecticism is ideal for the pursuit of middle range theory because it considers a wide

41

variety of possible causal processes occurring at differing levels of analysis.
82
This work
incorporates levels of analyses, or images, and several potential causal processes from a variety of
literary subfields, thus making analytical eclecticism a valuable tool. Furthermore, it seeks to
produce policy-relevant material that can potentially improve the management of organized
intraethnic conflict.
The roots of rivalry and the causes of intraethnic conflict onset
The first proposition is dubbed Roots of Rivalry. Recall that rivalry refers to an antagonistic
relationship between factions that may take the form of warfare, or other activities that can take
the place of war in less militarized societies.
83
Rivalries that morph into armed conflict should
have discernible reasons for doing so. Therefore, the proposition for the Roots of Rivalry (P1) is
given as:
The roots of DSN rivalry that develop into conflict are likely to be similar to one
recognized form of intraethnic conflict than the others.
I described how the most common explanations for intraethnic conflict align with the three
images above. The first image suggests that internal leadership disputes and personality clashes
lead to the splintering of ethnonationalist organizations. However, intraethnic outbidding is also
used against leaders of rival factions to achieve personal goals of empowerment or enrichment.
84

This explanation for conflict is named the Competing Leaders hypothesis (H1a):
DSNs are more likely to engage in conflict to resolve issues of friction between
individuals belonging to rival DSN factions.
A second grouping of related explanations examines local contexts and the relation
between the political factions pursuing leadership within an ethnic group and the state in which

42

the group resides. Intraethnic conflict could be the product of a state that successfully manipulates
rivals. A state—whose core community is comprised of out-groups—will play the grievances of
one off of others. Ethnic defection and collaboration with the state will be encouraged. Sometimes,
a state will seek to employ the faction as counterinsurgents. Recall that lowered information costs
and easy modes of communication are shared among coethnics. This makes defectors uniquely
suited to locating and eliminating each other, which constitutes an inversion of the reasons we
expect intraethnic cohesion.
85
We would expect the attributes and actions of the state to determine
the onset of intraethnic conflict, a second image explanation. The State Manipulation (H1b)
hypothesis reads as follows:
DSNs are more likely to engage in conflict at the behest of out-group state actors than
other explanations.
The third cause of intraethnic conflict is drawn from an earlier body of literature. Though
compelling explanations of conflict, the first and second image explanations may not necessarily
apply fully to DSNs. DSNs are divided into multiple states and their competitive patterns are
likewise transnational in scope. Leaders and fighters may have not even interacted, let alone
developed personal animosities. Also, states may not have an interest in manipulating these groups
into fighting each other so much as we may expect them to employ ethnonationalist factions as
proxies against other states. Because of these irregularities, I include literature that addressed a
similarly divided arrangement of ethnic groups that feature intraethnic conflict: the TSON
phenomenon.
86

Ethnic groups that are divided by externally imposed borders sometimes find themselves
situated on different sides of regional and system-level alliances. They may develop parallel
political and economic systems, while their social and cultural realms become entwined with those

43

of their allies. West Germany, South Vietnam, Nationalist Taiwan, and South Korea developed
varying levels and varieties of democracy, capitalism, and socio-cultural links to their Western
backers, and they also maintained certain elements of their pre-division cultures. Their
counterparts, on the other hand, implemented forms of communist-inspired political-economic
systems, social links to communist backers (through largely similar channels as the capitalist ones),
and retained other forms of their cultural identities.
Eventually, these identities hardened and were no longer compatible. However, ties were
not completely severed. Both sides were required to compete with each other over the same ethnic
group to prove that their vision for a political-territorial entity was superior to their dyadic rival’s.
87

This third image explanation for intraethnic conflict is called the Divergent Development (H1c)
hypothesis.
DSNs are more likely to engage in conflict if they develop divergent and irreconcilable
societal systems.
The roots of rivalry can be established through a temporal analysis that includes statistical
analysis and historiographic process tracing of the population from its ethnogenesis to the onset of
the conflicts explored in this work. Mechanisms for these explanations can then be identified and
supported, or alternatively, rejected.
The effects of distance on intraethnic conflict patterns
We have discussed the reasons for conflict onset and how those of violence among factions
representing DSNs may be similar or dissimilar to the reasons offered for intraethnic conflict more
broadly. However, we should also examine the character of intraethnic conflict for a more
complete picture of each category. This is the second of three components of the theory regarding

44

DSN conflict and cooperation among rival factions. DSNs, SNs, DIGs, and TSONs may all exhibit
different basic conflict patterns. Given the assumption from the New Wars literature that DSN
factions will possess relatively low levels of military capacity, geographic proximity to points of
interest (POI) may well be powerful predictors of instances of intraethnic conflict. With one
exception, the Effects of Distance proposition (P2) reads as follows:
As distance from a point of interest increases, instances of conflict are more likely
to decrease than increase.
In other words, there should be more conflict events closer to the POI. This proposition
can be broken down into more specific ones. First, a key finding from the literature indicates that
separatist conflict in geographically large states is greater in magnitude when a separatist group is
concentrated farther from a national center than when they are nearer,
88
unlike civil wars in which
the state capital is the object of conquest. This finding may or may not be directly applied to
factional competition because the entirety of the conflict would logically take place in or near the
territory inhabited by the ethnic group in question, but it is possible. The expected outcome varies
by the type of intraethnic competition. TSONs, especially, may not be limited by the same low
levels of military capacity as other varieties of intraethnic conflict. Although it is possible that
DSNs are more like TSONs in their in their objects of contestation, it is more likely that they are
more akin to separatist combatants. And so, the lone exception to P1, the National Capital
hypothesis (H2a) is as follows:
As distance from the national capital increases, instances of conflict are more likely
to increase than decrease.

45

However, if the conflict occurs between two states, we would expect the national centers to be
more likely to serve as an object of contestation. Capital cities have long been recognized as a
main target of states in war as they are the center of administrative power. They are a primary
center of gravity described by von Clausewitz (1917, p. 646-647) that warring states aim to capture
as a means of achieving victory over another state.
89

Additionally, we may expect a regional population center to be violently contested by rival
factions if group leadership is the goal of both factions in the dyad. Like national capitals, Starr
(2005) found that regional centers serve as symbols of a faction’s control over territory and
population, thus attaching territory and power to identity. While it is possible that a DSN may
exhibit similar conflict patterns to a TSON, it is not very likely as the violence should be more
localized. Regional population centers inhabited by members of the in-group are perhaps a better
predictor of conflict instances, which is somewhat of an inversion of the findings of Cederman et
al. (2009), in which an increase in distance from the corresponding national center leads to an
increase in conflict events. Hence, the Regional Center hypothesis (H2b) is most likely to strongly
predict the behavior of factions within DSNs:
As distance from the regional population center increases, instances of conflict are
more likely to decrease than increase.
One important consideration to make comes from the nature of DNSs. They are divided by
IRBS. Dyads that include nation-states, such as DIGs and TSONs, are also divided by IRBs. A
conflict pattern analysis by Goemans and Schultz (2017) finds that territorial division in the
colonial period influences contemporary conflict. War is more likely to occur when IRBs take the
shape of straight lines that result from division in the colonial era, rather than when they are the

46

product of a more natural process of division. Recall that the conflict of TSONs is rooted in
divergent development that is itself aided by division.
Such findings demonstrate that borders should be included in an analysis of the character
of intraethnic conflict and cooperation among rival factions. Instances of armed conflict may
plausibly be expected to occur close to the border because of the limitations of low military
capability many factions face as a type of New War. An armed force crossing a border may be
intercepted before reaching the regional or national center because, logistically, it must transport
men and materiél over great distances without advanced modes of troop transportation. Even if
such technology is available, state air power limits its utility.
Furthermore, if system-level factors are indeed salient, rival factions may rely on allies in
the region for access to territory in order to attack a rival elsewhere. Even during the Cold War,
proxy conflict sometimes appeared similar to New War intraethnic competition for leadership, but
often without the limitations of low military capability because divided societies formed powerful
states, like the two Vietnams or two Koreas. These dyads were capable of waging total war against
their rival. However, this is unlikely for groups that do not have highly developed militaries. Thus,
the second hypothesis, Unpacific Borders (H2c), is as follows:
As distance from the IRB increases, instances of conflict are more likely to decrease
than increase.
Despite having low levels of military capabilities, these factions may still use vehicles,
such as technicals and armored vehicles to transport mechanized forces. They may require access
to well-built roads. Highways are also vital if a faction wishes to control territory and populations

47

because they enable survival in war.
90
Therefore, a fourth hypothesis is included. The Primary
Route hypothesis (H4) states:
As distance from the primary route increases, instances of conflict are more likely to
decrease than increase.
The effects of distance may be assessed statistically in the manner described in the
following chapter. Additionally, factional media should reference these POI as being vital to their
goals. Alternatively, evidence for such an impact may favor rejection of this hypothesis, or simply
not be present.
Out-group incursion and rival cooperation
Rival factions do not always engage in armed conflict with each other. Sometimes they
engage in cooperative action against an out-group invader. The preceding chapter contained a
survey of the literature and presented recent and historical examples, such as those involving
Kanuri factions and the Second United Front, respectively. It is necessary to test the three
competing explanations for rival cooperation to systematically distinguish between categories of
organized intraethnic conflict. Ethnic solidarity and power relations; micro level nuances that
reveal complex webs of loyalties; and institutional diversity within a single ethnic group provide
potential explanations for why rival factions may cooperate even while they aim to expand their
populational-territorial influence.
I synthesize the three explanations to build upon the third generation of research. The
condition of out-group incursion from the first generation of research is used as a starting point to
look for whether or not there are cooperative outcomes. However, in line with the second
generation of research, I assume that the Manichean view of in-group solidarity and out-group

48

vileness is perhaps too reductionist. In keeping with the second generation, I assume that local
complexities will be influential, but that they do not necessarily result in fragmentation. I follow
McLauchlin and Pearlman (2012) in emphasizing the institutional balance of power within the
group as a decisive factor in predicting whether cooperation occurs or not. It is plausible that, if a
faction wishes to introduce their institutions to a new population or territory, it will have a better
chance of doing so amidst an out-group incursion. Whether or not the other faction allows them to
do so may depend on local nuances and factional relationships. Thus, the proposition for rival
cooperation, Out-group Incursion (P3) is given as:
Rival ethnonationalist factions are more likely to experience unity or fragmentation
during periods of incursion by an out-group than during periods of institutional
equilibrium.
Out-group incursion occurs when a recent exogenous shock upsets the territorial and
populational status quo. Such a shock, often in the form of an ad hoc partition of an existing state,
91

provides opportunities for separatist or irredentist governments to expand. Factions that pursue
goals of populational-territorial and institutional expansion amidst such a shock would likely court
populations of coethnics that live in newly accessible territory, presenting the chance to create ties
between the population and the faction.
92
Additionally, the variation in categorical outcomes—
unity, fragmentation, no change—ensures that this proposition provides opportunities for
falsification.
Rival factions may seek to fill a void in security. Members of other groups or hostile state
forces may do the same. Out-group incursions thereby present opportunities to study strategies of
institutional expansion chosen by rival factions. Out-group forces that kill newly accessible

49

unaligned populations of coethnics may induce rival cooperative action. The following argument
forms the basis of this work’s contribution to the third generation of research.
If a population of coethnics is not fully aligned with either faction and it is subject to mass
killings by out-group forces in territory desired by both factions, we would expect to see unaligned
members decline to endorse the institutions of any faction that permits such an outcome. Therefore,
we should expect to see the most intense cooperative action take place in territory that experiences
the greatest density of violence against unaligned coethnic. Perhaps the type of atrocity or absolute
number of deaths matter more than density, but those are subjects for further research that I
elaborate upon in Chapter Nine. Factions should cooperate to eject the out-group threat to retain
institutional competitiveness in such areas. This is because cooperative action allows a rival faction
to present their institutions as a competitor once the external threat recedes. It is preferable to
compete with a coethnic rival later, where the benefits of coethnicity may come into effect, than
to lose the population and its territory to the out-group permanently.
The first hypothesis associated with P3, which is dubbed Who Dies and Where They Die
(H3a), states that:
Rival factions engaged in conflict with one another are more likely to cooperate
against an external foe that deliberately kills unaligned coethnics in institutionally
competitive territory.
Territory containing greater acts of violence against unaligned coethnics relative to rival
coethnics should experience the most intense cooperative action. Cooperation provides access to
territory that was otherwise inaccessible to one or more factions before the out-group invasion.

50

The newly arrived faction may alter the institutional balance in the geographic territory in which
they now operate once the threat recedes.
However, what if the threat is relatively low? Or suppose that the territory being invaded
contains populations of coethnics that are firmly aligned with a faction. If an external actor invades
such territory, that faction’s rival may be less likely to come to their aid. This is because the out-
group incursion degrades the institutional power of the rival without unacceptable risk
93
to
unaligned populations. Such an interruption of the institutional status quo within the ethnic group
may benefit the rival not being invaded.
If populations and territory of unaligned coethnics are not threatened, an unaffected rival
may benefit by withholding assistance as their rival’s influence and prestige is degraded. Even if
assistance is offered to safeguard populations and territory dominated by the affected faction, the
affected faction has little reason to accept unless they are at risk of losing it entirely. This is because
accepting assistance from rivals would provide that rival a foothold in otherwise firmly held
territory.
In such cases we should expect to see the rival making the offer claim that the other is a
traitor to the broader nation for refusing assistance. The Heroes and Traitors hypothesis (H3b) is
given as:
Rival factions engaged in conflict with one another are less likely to cooperate against
an external foe that deliberately kills coethnics aligned with a rival faction in territory
firmly held by the rival.

51

An alternative explanation: Non-combatant deaths
Perhaps alignment of non-combatant co-ethnics is not overly influential when rival factions
must decide whether to cooperate or not. Recall that the first generation of research informs us that
emotional affect might be the driving force in unity. The irredentism literature identifies the need
to protect co-ethnics as a motivator for the combating of out-groups.
94
States whose core
community is comprised of the threatened group opt for offensive capabilities to protect “ethnic
islands” from out-groups that threaten the group in other states.
95

This work’s analytically eclectic design draws upon other bodies of literature. For instance,
the human security literature describes how atrocities that “shock the moral conscience of
mankind” rouse states to intervene in conflicts on humanitarian grounds.
96
It is possible that all
non-combatant deaths might drive rival cooperation. Co-ethnics may wish to protect vulnerable
members of the group, regardless of their political affiliations. Therefore, the Non-Combatant
Deaths hypothesis (H3) states:
Rival factions are likely to cooperate to combat an out-group force that kills non-
combatants, regardless of alignment.
These hypotheses incorporate aspects of identity, both ethnic and political, as well as
geographic location. An empirical framework that can accommodate both geospatial requirements
and highly contextual nuances for the conflicts being studied is required. The hypotheses of Out-
group Incursion (P3) may be assessed statistically. Factional media should also reference the
necessity to protect the in-group so as to posture themselves as leaders once the threat recedes.
Central to all three hypotheses regarding cooperation among rivals is the assumption that
rival factions seek to claim in-group leadership. Accepting assistance from rivals is expected when

52

the alternative is the loss of population and associated territory to out-groups. Institutional
competition with rivals in the future is preferable to permanent out-group control. In the tradition
of the third generation of research, the institutional status quo may be altered in the future to benefit
a rival who assists in the short-term. All three propositions and their associated hypotheses of the
Conflict and Cooperation Theory are summarized below in table 2.1:
Table 2.1
Conflict and Cooperation theory: Summary of propositions and hypotheses
Proposition Name Prediction
Hypothesis
P1 Roots of Rivalry The roots of DSN rivalry that develop into
conflict are likely to be similar to one
recognized form of intraethnic conflict than
the others

H1a Competing Leaders DSNs are more likely to engage in conflict
following competition between individual
leaders than other explanations

H1b State Manipulation DSNs are more likely to engage in conflict
at the behest of out-group state actors than
other explanations

H1c Divergent Development DSNs are more likely to engage in conflict if
they develop divergent and irreconcilable
societal systems
P2 Effects of Distance As distance from an object of distance
increases, instances of conflict are more
likely to decrease than increase

H2a National Capital As distance from the national capital
increases, instances of conflict are more
likely to decrease than increase

H2b Regional Center As distance from the regional population
center increases, instances of conflict are
more likely to decrease than increase

53

Proposition Name Prediction
Hypothesis
H2c Unpacific Border As distance from the IRB increases,
instances of conflict are more likely to
decrease than increase

H2d Primary Route As distance from the primary route
increases, instances of conflict are more
likely to decrease than increase
P3 Out-group Incursion Rival ethnonationalist factions are likely to
experience unity or fragmentation during
periods of incursion by an out-group than
during periods of institutional equilibrium

H3a Who Dies and Where Rival factions engaged in conflict with one
They Die another are more likely to cooperate
against an external foe that deliberately kills
unaligned co-ethnics in institutionally
competitive territory than during periods of
institutional equilibrium

H3b Heroes and Traitors Rival factions engaged in conflict with one
another are less likely to cooperate against
an external foe that deliberately kills co-
ethnics aligned with a rival faction in
territory firmly held by the rival
H3c Non-combatant Deaths Rival factions engaged in conflict with one
another are more likely to cooperate against
an external foe that deliberately kills non-
combatants regardless of territorial control
or factional affiliation

Note. The general propositions of P1, P2, and P3 frame the sequence of testing; the hypotheses
correspond to the specific tests explained in the research design.

54

2.3 The Conflict and Cooperation Theory: Depicting rival factional behavior using the
systemist technique
This section will present the overall logic of the Conflict and Cooperation theory in graphic form.
Before doing so, let us turn to the systemist technique. This means of graphic communication of
theoretical mechanisms was selected to convey this work’s theory because of its low barrier to
entry, efficiency in conveying theoretical mechanisms, and relevance to policy.
The systemist technique
Systemism in IR was created by James (2019) and is adapted from Bunge (1996). It is intended to
enhance communication across subfields in IR, while accommodating different learning styles
through efficient illustrative practices.
97
Pfonner and James (2020) have demonstrated
systemism’s ability to manage cognitive load while focusing on completeness in a manner that is
compatible with analytical eclecticism in ways that competitor approaches—like argument
mapping and game theory—cannot. As such, systemism is applicable to both quantitative and
qualitative work.
Systemism produces diagrams using diagrams.net, a free online tool. Such diagrams
convey linkages between and across macro and micro level variables using all levels of analysis.
Capitalization and position within the diagram denote whether a variable is found in the macro or
micro level. Variables located in the upper plane are identified with all capital letters. These are
macro variables. Those in the lower plane labeled with lower case letters are micro variables.
The diagram contains a system that is being studied, as well as the environment in which
the system exists. Variables and theoretical mechanisms are represented by shapes. These shapes
each correspond to a preselected color. Combining shapes with colors reinforce visual cues as to

55

the role of that type of variable. The colors are different shades of brightness by design, so that
red-green colorblind colleagues can use the same visual cues in grayscale. Arrows denote the
direction of the causal relationship, where the relationship can be read as X is a putative cause of
Y, depending on the direction of the arrow. The variables and types of linkages are depicted below
in table 2.2:
Table 2.2
Systemist notation

Initial Variable

The starting point of a series
of relationships.

Generic Variable

A step in the process being
depicted.

Divergent Variable

Multiple pathways are
created from a single linkage.

Convergent Variable

A single pathway is created
from multiple linkages.

Nodal Variable

Multiple pathways are
created from multiple
linkages.

Terminal Variable

The end point of a series of
relationships.

Connection Stated in Study A linkage explicitly made by
the author.

Connection Crossing Over

Two separate linkages that do
not interact.

Interaction Effect

Two variables that depend
upon the effect of the other.

Note. This table is adapted from James (2019) and Pfonner and James (2020).

56

Green ovals represent initial variables. These are where a series of relationships is generated. White
rectangles are generic variables, which illustrate mechanisms along a given pathway. Orange
diamonds—divergent variables—are a pathway in which a single input may split into multiple
contingencies. The blue parallelogram is the opposite. This is a convergent variable that receives
multiple inputs, but only produces one output. Purple hexagons, or nodal variables, receive
multiple pathways and also generate multiple pathways. The red octagon is a terminal variable,
which signifies the end of that series of relationships.
The linkages are represented by arrows. Unidirectional arrows display the direction of the
putative cause in a chain of events. Lines that do not interact but must cross due to space constraints
are identified as an arc, as is common with schematics in the mechanical sciences. Mutually
reinforcing linkages are identified as a bidirectional arrow, which represents an interaction effect
between variables. Other symbology exists for more specific applications, such as explications of
critical theory, which may need variables that are more co-constitutive in nature.
There are three general figures, each corresponding to a proposition of this work. They are
expanded upon in Chapter Eight in light of the results of the tests for the propositions’ hypotheses.
Each diagram involves the region in which an ethnic group is found. That region may also be
affected by the international system. The first proposition is displayed in figure 2.1:

57

Figure 2.1
Systemist exposition of the Roots of Rivalry (P1)

Note. Created by the author using https://www.diagrams.net/

We begin with the initial variables on the left side. In descending order of aggregation, we begin
the series of relationships with system-level factors of division and statehood, state manipulation,
and leadership disputes. These represent our hypotheses in reverse order. Each of these is followed
by generic variables that represent mechanisms occurring along the pathways for each hypothesis.
“NATURE OF DIVISION AND STATEHOOD → divergent factional development → coethnics
experience conflict with out-group” can be read in the following manner: “the system-level forces

58

that divide an ethnic group or keep in intact, as well as whether it is rendered stateless or not,
influences the degree to which factions within the group develop convergent or divergent
institutions. This, in turn influences the effects on decision-making with regards whether relations
will remain pacific or not when factoring in information costs if the group enters a conflict with
an out-group.” Similar pathways exist for the other two levels of analysis.
The remainder of the pathways occur after the purple hexagonal nodal variable. Depending
on the relative impacts of the competing hypotheses, pacific relations may be maintained, or
organized intraethnic conflict onset might be the outcome. This may happen in the micro or the
macro, depending on the distribution of the group in question and the nature of its alliances, if it
has any.
Let us now look at the second proposition. This is the subject of figure 2.2:

59

Figure 2.2
Systemist exposition of the Effects of Distance (P2)

Note. Created by the author using https://www.diagrams.net/

In figure 2.2 we see how the Effects of Distance (P1) affect the density of conflict events. Material
and ideational factors may be affected by the relative capacity of the faction in question. Strictly
material factors are represented by generic variables, including greed explanations, which will be
treated shortly in the subsection dealing with alternative explanations.
The convergent variables of the national capital and regional center span both material and
ideational motivators for conflict. Note that the linkage from “MATERIAL FACTORS:

60

CONTROL” has an arc that indicates this pathway does not converge with the linkage coming
from “IDEATIONAL FACTORS: GROUP LEADERSHIP” to “REGIONAL CENTER” but
rather converges with the linkage from “national capital.” Pathways converge, but they do so when
the same POI is involved.
As distance from each POI increases, there should be a measurable decay impact for the
density of conflict events, the strategy for which is fully described in the following chapter.
Moving on to Out-group Incursion (P3), let us examine a diagram that explains why rivals
may or not cooperate amidst a threat from outsiders. This is displayed in figure 2.3:

61

Figure 2.3
Systemist exposition of Out-group Incursion (P3)

Note. Created by the author using https://www.diagrams.net/

Again, we begin on the left side of the diagram. Some manner of exogenous shock, such as an ad
hoc partition or the defeat of a host state in war, can alter the populational-territorial status quo
between factions of an ethnic group. Preexisting competitive dynamics and this change in
equilibrium can make new territory accessible for the factions to compete over. That territory may
also be affected by out-group incursion. In this instance, that incursion comes from the

62

international system. However, it is possible that it could come from the region, state, or sub-state
as well, depending on the group(s) in question.
The nodal variable has multiple pathways that represent the decisions undertaken by out-
group forces with regards to targeting strategies. Each of the generic variables represent a
hypothesis. They lead to divergent variables; here, the group has to decide how to react to the out-
group’s targeting strategy. They either cooperate, or do not. Cooperation, like out-group
intervention, can occur in either the system or the region, again depending on the group.
Scope conditions
Let us discuss the scope of the phenomenon described here. This work is intended to be applicable
to competing ethnonationalist factions that engage in organized intraethnic conflict, whether for
individual-, state-, or system-level causes. These factions are required to claim to represent
coherent movements and possess the ability to mobilize armed forces. Periods of out-group
intervention are included in the scope because we would not otherwise expect to see rival factions
cooperate if they are pursuing intraethnic dominance.
There is no specified time interval with regards to the duration of conflict when predicting
if cooperative dynamics may take place. No two ethnonationalist movements are identical, so this
is expected to vary.
98
We may reasonably expect exogenous shocks and out-group attacks to
provide a potential to study whether conflict ceases, and whether cooperation begins.
Conflicts may have recognizable patterns based on the disposition of involved actors before
and during the conflict, as well as their behaviors.
99
Therefore, the permutations of statelessness
(or not) and division by borders (or not) provide four conflict patterns that this work seeks to
explain.

63

These four categories include organized intraethnic conflict involving the divided stateless
nation, the single-state separatist, the divided irredentist group, and the two states, one nation
phenomenon. The divided stateless nation is the primary contribution of this work, though the
scope should apply to each category.
In concrete terms, this work is applicable to describing and explaining the origins of
conflict for the four varieties of organized intraethnic conflict. Future work will further support or
disconfirm the findings here. It also provides a framework for predicting which objects of
contestation will be fought over by factions in each category. This framework also holds utility for
predicting when and why factions will cooperate against outsiders or not. Finally, it offers
prescriptions for managing and resolving DSN conflict.
Chapter Nine continues this discussion of scope conditions once the results have been
discussed. Later, this work will offer future suggestions for defining the universe of cases that
would cement the full scope.
Alternative explanations: Additional material and ideational considerations
This work considers both material and ideational factors in its assessment of the objects of
conflict. One potential alternative material explanation comes from the so-called greed family of
explanations for conflict.
100
This work is not opposed to greed-based explanations. However, I do
not include a specific hypothesis that incorporates greed-based objects of contestation. The main
reason is practicality; some objects are readily conducive to a spatial framework, such as oil
refineries.
101
Others, however, like human smuggling routes or illicit arms caches, cannot always
be properly identified. This makes generalization across all cases nearly impossible.

64

However, these greed-based objects may be easily identified in the source material for
conflict events. Reports of conflict over TOC objects, or minerals, etc. will provide valuable
sources of inference as to the impact of these motivations. These are considered throughout the
case studies in Chapters Four through Seven.
A second alternative explanation for cooperation comes from an ideational approach.
Shared culture may improve the likelihood of lasting peace following organized intraethnic
conflict, which counteracts the hardening of identities and lessens the probability of renewed
conflict.
102
If this view is correct, shared culture may be a moderating influence for groups that
have a state (or states, in the case of the TSON) that may (re)unify. This may also influence the
probability of cooperation, though divergent development will likely alter what elements of culture
remains shared.
It could allow for a merging or hierarchical ordering of factions. Also, perhaps shared
culture can be a calming force if territorial transfers can appease irredentist movements or other
categories more so than other, whereas factions remaining stateless have no such mechanisms for
reconciliation. Thus, a perpetually stateless (or dyadic state) rivalry may be more likely to become
a protracted conflict, even within groups.
103

Both alternative explanations are remarked upon throughout the case studies. Chapters
Eight and Nine will also address them in a systematic fashion.

65

Chapter Three
An Analytically Eclectic Design of Intraethnic Conflict and Cooperation among Rival
Factions

3.1 Empirical strategy
This chapter presents the empirical strategy that I employed in the case studies to follow. The
strategy follows analytical eclecticism which incorporates theoretically diverse lenses. It
pragmatically combines elements of each so that we may derive answers to questions that are only
partially addressed by different subfields within a discipline. It aims to create middle-range theory
that can solve real-world problems.
104
This work draws from multiple traditions and applies their
theoretical underpinnings to hypotheses generated by existing work. Thus, analytic eclecticism
with the use of mixed methods, may be the optimal approach to study a wide range of possible
factional behaviors through a diverse range of tests.
Induction, deduction, and the structure of analysis
This work is partially inductive due to its analytically eclectic nature that incorporates a
variety of inputs in an attempt to identify patterns among outcomes. However, there is a strong
element of deductive reasoning because it incorporates propositions from the literature in order to
develop new ones. This work tests existing theory, yet it also generates new theory based upon its
key findings.
Findings regarding conflict within DSNs will provide a new set of theoretical implications
and hypotheses. These may then be applied to a sample of DSNs in order to ascertain whether

66

there is significant variation within them or not. Thus, this work’s findings remain falsifiable and
provide an opportunity to expand their potential for stronger external validity if applied to more
cases in the future.
The spatial analytic approach is somewhat novel for the study of international relations.
Incorporating explicitly geographic approaches is simultaneously long-standing and sporadic.
Agnew (1994) recognized that social science frequently views territorial space as a function of
peripheral subordination to the political center. Under such a view, territoriality and boundaries
are viewed as equally important for all states. Likewise, a subnational unit’s relationship to an
administrative unit is taken for granted as necessarily subordinate. The state acts as a “container”
105

for the society—or societies—within. In newer states, and especially those with arbitrarily
imposed “containers,” this overlooks contextually complex local arrangements and processes, as
well as the heterogeneity of populations that would otherwise be assumed to be homogenous under
more state-centric approaches.
Human geographers recognized the limitations of viewing the nature of the territorial state
in such a manner, and so approaches accounting for geographic scale have seen application within
the discipline, particularly for the study of greed and grievance in civil wars.
106
The use of GIS
allows for discovery of locally important variables that may be absent at the national level.
107

Modern GIS allow for complex analyses of spatial patterns and modeling to be applied to
international relations in ways that were not possible until very recently.
108
The creation of
historical gazetteers of geographic features allows scholars to examine spatial trends in ways that
were once only possible through prohibitively labor-intensive qualitative analysis.
109

With regards to armed conflict, specifically, an application of conflict pattern analysis by
Goemans and Schultz (2017) finds that territorial division in the colonial period influences the

67

onset of conflict in the modern era. The manner by which IRBs are created and imposed can
portend future conflict. For instance, post-colonial spaces divided by borders that were arbitrarily
drawn, specifically if they are straight lines, are more likely to witness conflict at some later date
than spaces in which post-colonial borders take local group differences into account.
The potential for application of spatial approaches to identity-based conflict is recognized
in the literature.
110
Spatially analytic methods have been used to test both interstate and intrastate
conflict to reexamine the role of material and ideational factors influencing the onset of war.
111

This work provides an extension of this geographically based variety of methods by applying it to
competition, conflict, and cooperation through applications of spatial statistical methods at varying
levels of analysis.
I used a variety of statistical methods broadly encapsulated by exploratory data analysis
(EDA); exploratory spatial data analysis (ESDA); spatial data analysis (SDA); and spatial
modeling (SM).
112
EDA consists of summary statistics—Pearson correlation coefficients, OLS
regressions, and the like. Such techniques are employed as a means of initial exploration of the
data to find basic relationships within them. ESDA is similar but concerns spatial data. The object
of ESDA is to establish whether data have any spatial relationships, though not to assess the causes
of such relationships. That is the task of SDA. Some ESDA techniques include KDE and ANN.
Spatial regression analysis can be an example of both SDA and SM because it tests for potential
drivers of relationships in the data, not merely the presence of them, while they are also expressed
as formal models. Generally speaking, only the results of SDA and SM are displayed in the case
studies. The results of EDA and ESDA may be found in the Appendices for their respective cases.
Early pre-processing stages were performed using PostgreSQL with PostGIS, which are
components of a spatial database management system (SDBMS). The SDBMS is a tool for

68

managing, querying, and organizing spatial data. Data were then exported to Quantum GIS (QGIS)
for further analysis and visualization with the use of a Python bridge. Later analyses were
performed with ArcGIS Pro and R. All R packages used are cited in Appendix A. Analyses and
visualizations created with ArcGIS Pro have been duplicated and stored for use with QGIS in case
of future institutional limitations. QGIS is open-source software, while ArcGIS Pro is proprietary
software. All statistical methods were accompanied by qualitative methods, as well.
The remainder of this chapter proceeds as follows: it presents the research design that was
used in the case analyses to follow; variables are defined and constructed; the methods and their
relations to the hypotheses from the preceding chapter are described; the variables are
operationalized and the sources for data are explored; the case selection strategy is developed; and,
finally, previews of the cases are offered.
3.2 Research design
This section first covers the variables that are at the center of this work’s analysis. It then presents
the methods that were employed, as well as the operationalization of the variables as data.
Procedures for testing the three propositions are put forward. A strategy for case selection is
explicated, and justifications are made for the cases that were ultimately chosen. The section
concludes with an overview of the case chapters.
Variables
This subsection provides an overview of the explanatory variables (EVs), dependent
variables (DVs), their constructs, and their modes of operationalization. I also describe the sources
used for data collection in this subsection so that the subsequent discussion of methods possesses
greater clarity.

69

Dependent variables: Conflict and cooperation
This work has two DVs. The first DV is organized intraethnic conflict among rival
ethnonationalist factions, which was described above as exchanges of force and violent
interference into the domestic affairs of the governing institutions of a rival faction. There are two
constructs for organized intraethnic conflict. The first DV construct is conflict onset. Conflict onset
is the beginning of armed confrontations between rival factions. Onset is a dichotomous variable;
it either begins or it does not. Historiography and factional media are used to identify whether
conflict has occurred.
The second DV construct is the individual intraethnic conflict instance. The instance is the
identifiable conflict event involving two or more factions. This construct has a continuous
operationalization that occurs at the spatial scale of the individual-level event. The data for this
variable is drawn from the Peace Research Institute, Oslo (PRIO) Uppsala Conflict Data Program
(UCDP). Specifically, the Georeferenced Event Dataset (GED) v20.1 was used.
113

The GED consists of spatial data that is constantly updated from wire, government, IGO,
and NGO sources. These data are georeferenced points where some manner of armed conflict took
place. Each vector contains spatial and non-spatial information, including location; dyadic actors
involved; several estimates of casualties; type of violence; the source (typically a press or a
government agency); state/province or other subnational disaggregation; date; and confidence with
regards to geographic precision; and many other categories.
The Armed Conflict Location and Event Data (ACLED) Project
114
was assessed, but was
rejected in favor of the UCDP, which appeared to be complete and had a superior data export
portal. Additionally, the Correlates of War (CoW) Project was not considered. Though there is an

70

intra-state version
115
and a non-state version,
116
the data is not ideal for this type of work. The
CoW datasets are recorded at the scale of the conflict itself, rather than at the individual event,
thus the GED was the logical choice.
The second DV is cooperation between rival factions. This was defined as two or more
factions reducing or halting armed conflict in order to achieve a common goal. Cooperation is
constructed as cooperative action involving two or more factions. Such action involves the
provision of personnel and equipment by one faction to a rival faction in order to repel an out-
group assault by fire and close combat. Permission to operate in rival-controlled territory is
required to be considered cooperative action, thus the establishment of a rival base or checkpoint
(if sanctioned by the rival) was included as a cooperative event.
Cooperative events are a continuous variable at the individual spatial level. Events were
drawn from the UCDP GED and supplemented by qualitative research that included reports from
the US Department of Defense (DoD) and other governmental sources; NGOs, such as Human
Rights Watch (HRW), Syrian Observatory for Human Rights (SOHR), International Crisis Group
(ICG), and others; IGOs, such as the UN High Commissioner for Refugees (UNHCR); and party-
affiliated media outlets and press releases by the factions themselves. Each event was required to
hold geographic references with a degree of spatial confidence comparable to that of the GED in
order to be georeferenced.
Explanatory variables: Origins, distance, and invasion
There are three broad EVs that are associated with the propositions. These broad EVs are
root causes of rivalry (Roots of Rivalry—P1), geographic proximity (Effects of Distance—P2),

71

and invasion by outside forces (Out-group Incursion—P3). I will present the constructs of each in
turn.
The explanatory causes of rivalry have three constructs. Leadership disputes among
ethnopolitical organizations comprise the first. A leadership dispute must have some observable
rift between individual leaders that is based on personal ambition or mutual animosity.
The second construct is manipulation by state actors. Such manipulation should consist of
outreach by state actors to one faction as a means of initiating conflict with another faction. It is
not enough that state and rival factional goals coincide or that the state can take advantage of an
existing rift. The state must either be the actor that drives the beginning of conflict, or the faction
must be clearly subordinate to the state for evidence of this hypothesis to be considered supported.
The third rivalry construct is divergent factional development. Factional development is
the trajectory of the political, economic, social, and cultural realms of the population and territory
that the faction controls. Divergence refers to the differences in those realms that emerge within
the same group over time. The source(s) of divergence must be observable. For instance, the
divergence may be due to a faction’s membership in an alliance network. Or, perhaps, the
divergence may be a result of varying levels of integration into multiple states involving a single
ethnic group, or some other cause.
All causes of rivalry were assessed for plausible mechanisms through spatiotemporal
analysis, as well as historiographic and qualitative process tracing. This is described fully in the
coming presentation on methods.
The preceding three constructs were used to test the onset of conflict. Let us turn to the
next EV, which is geographic proximity, or, more simply, distance. Due to constraints on the

72

material and military capabilities that identity-based conflictual rivals generally experience under
the New Wars lens, distance should be a powerful limiting factor on the intensity of armed
confrontation between rivals. The objects of contention will likely be affected by the distance
between crucial, identifiable features. There are also three constructs associated with this EV. The
first is distance from the national capital. This concept has proven to be salient in predicting
conflict events in the literature,
117
so it will be tested here. Conflict events are more likely to be
farther from the national capital than closer to it.
The second construct is distance from the regional center. Major cities or towns that are
inhabited by the ethnic group in question are considered regional centers if they are the most
consequential seat of government and administrative control for the faction or ethnic group.
The third construct is proximity to the national border. Though the border itself may be
used as it is a type of demarcation for state forces, this is ideally either an official or unofficial
border crossing, if such an object’s existence may be ascertained,. A known smuggling route for
transnational organized criminal (TCO) groups can be counted as an unofficial border crossing.
However, border crossings that were known to be opened after the period of conflict were not
considered for the analyses for obvious reasons of uncertainty.
The entirety of the border itself could be used if necessary. Also, the internal border can
also be used for a type of robustness check when other data is lacking. Assessing whether borders
between semi-autonomous regions might also be useful for studying borders that are subnational
in character. This additional approach was used for the Chechen case, in which there are national
and subnational borders involved that divide peoples.

73

The locations for national capitals, regional centers, and border crossings were drawn from
gazetteers, governmental, IGO, and NGO sources. The administrative lines of the IRBs and
subnational borders came in the form of polygon shapefiles from the Database of Global
Administrative Areas (GADM) version 3.6.
118
This data source is compiled from government
sources specifically for scientific use. It consists of national boundaries and various subnational
administrative units. These may be states, provinces, governorates, counties, districts, and
subdistricts, depending on the polity.
Improving Euclidean distance
Distances between conflict events and POI were measured in kilometers. Euclidian or
planar (straight-line) distance was ultimately used because reliable, comprehensive road network
data for all time periods could not be established. Euclidian or planar distance is not ideal,
119
so I
also employed a basic road network to create spatial weights matrices.
This was an imperfect but necessary compromise. Using a modern road network from
OpenStreetMap (OSM, 2020) may have actually given less reliable results. By including roads
that may not have existed at the time of the case there is a risk of introducing greater uncertainty
with regards to the reliability of the statistical results. Using recent OSM data for historical cases
could also introduce temporal bias, and so it is potentially less useful than Euclidean distance. I
also wished to assess cases as equally as possible.
For these reasons, Euclidean distance was selected, but with an important modification.
Some cases involved armored vehicles transporting mechanized forces that can only be driven on
well-built highways. Highways are important for territorial and populational control. In turn, such
control is necessary for nonstate armed groups, whether motivated by greed, grievance, or simple

74

strategic control.
120
Therefore, Primary Route data from the Humanitarian Data Exchange (HDX),
a platform of the United Nations Office for the Coordination of Humanitarian Affairs (OCHA)
and NextGIS
121
was also included. These data consist of polyline files of highways that were
known to have existed for the time periods in question.
A digital elevation model (DEM) was considered for use in the analysis. DEMs represent
the geographic area of interest and contain topographic information, such as elevation. It is
intuitively conceivable that mountains would prevent a problem for movement. This is, however,
not a major concern, particularly for the New Wars lens. For one, both regular and irregular forces
of light infantry routinely cross geographic obstacles like mountains and rivers, which provide
advantages in asymmetric warfare.
122
Roads might be efficient means of travel; they are also death
traps in asymmetric warfare. In my own experience, goat trails were often preferred for traversing
large mountains, rather than roads.
Mountains feature strongly in some of the cases in this work. They were used by forces
possessing low levels of military capacity to their advantage, both defensively, and offensively, as
firing in enfilade or defilade is easier to accomplish for a force that is familiar with the local
elevation conditions. Road usage also presented a hazard that could be exploited by enemy forces
with greater levels of military capacity. I expand on this in the findings when discussing the
seeming low impact of primary routes.
Additionally, the analytically eclectic design of this work already sacrifices a degree of
parsimony for accuracy and completeness. Adding a DEM would further subtract from the former
without benefit to the latter two elements of mid-level theory. For these reasons, a DEM was not
used.

75

The third and final type of EV is invasion by an out-group, for which there are three
constructs. The first consists of the intentional killing of unaligned coethnics. These are members
of the group who are killed by out-group forces but belong to a community that is not politically
aligned with a faction. The second construct is the killing of in-group members who are aligned
with a political faction. Importantly, civilians may count as rivals if they are known members of a
political party. Members of armed forces belonging to political parties automatically counted as
rivals. The third construct is the killing of any non-combatant without accounting for political
affiliation.
As with conflictual and cooperative events, these data were drawn from the GED and from
qualitative research. Supplemental events were required to be spatially specific to at least the level
of a named inhabited place to be included in the analysis. If the event was spatially referenced at
the subdistrict level (the equivalent of the county), it could be included if it had an identifiable
feature, such as the road between two villages, or a specific mountain, or other such object. Deaths
and affiliation for the GED events were probabilistically assigned based on research into every
individual-level source for each event.
Methods: The Roots of Rivalry (P1) and conflict onset
In this subsection I describe the tools used to conduct the case analyses in Chapters Four
through Seven. Generally, these tools proceeded in three stages, with each stage corresponding to
a proposition in the order in which they were presented. For each stage I begin with an overview
of the goals for that analysis before describing the methods in detail. Broadly speaking, each stage
has a statistical and a qualitative component to it.

76

The Roots of Rivalry proposition suggests that factional conflicts within DSNs should
originate with a phenomenon similar to one or more of the other varieties of intraethnic conflict.
This was assessed through a mixed methods approach. First, a spatiotemporal analysis was
undertaken. Potential effects of the establishment of modern IRBs on the ethnic group as a whole
were identified. The analysis was performed to determine whether there could have been
populational and territorial potential for rival creation, conflict, and cooperation for each ethnic
group in the analysis based on changes in population and territory over time. IRBs could potentially
split groups into two or more states in which they form a core community of no state; split groups
into two or more states in which they form the core community of both states; formally render a
group stateless within one state; or divide a group into a single state and its irredenta.
This step compared the group’s extent before and after the imposition of IRBs, the visual
result of which displayed how much their territory changed across time. It also provided cues for
the qualitative component of this first stage of the analysis. That component involved process
tracing to examine whether the imposition of IRBs lead to the outcomes proposed by the
hypotheses of the Roots of Rivalry proposition. As an example, if a group’s population expanded
in one region, yet disappeared in another, there was a reason to investigate whether a disruptive
population transfer that could have had an impact on the development of rivalry occurred.
Suppose, for instance, that each of the recognized forms of intraethnic conflict appear to
provide evidence for separate hypotheses. In that case, we might surmise that explanations from
different images may be more useful for that pathway than others. Thus, if the root of DSN rivalry
appears more similar to one mode of intraethnic conflict than others we have a starting point for
either defining DSN conflict as a separate variety or eliminating it as being sufficiently distinct.

77

Testing the nature of rival conflict and cooperation among multiple DSNs in future work will
further contribute to our knowledge of this phenomenon.
The spatiotemporal analysis began with map processing, which is the conversion of
manually created historical demographic maps into a machine-readable format in order to extract
data that may be used in a GIS. I required three historical ethnographic maps showing the
distribution of Kurds, Chechens, and Albanians prior to the establishment of modern borders. This
is necessary because the dataset I employed for populational and territorial distribution—the 2019
version of the Georeferencing Ethnic Power Relations database (GeoEPR)
123
—only dates to 1946.
However, this meant that a historical map for Korean distribution was not required, as the current
borders for that group were imposed in 1948.
The goal of this stage of the project was to use demographic maps from before the
imposition of IRBs to process, georeference, and relate them to the GeoEPR dataset. Thus, it
created pre-1946 GIS layers for the three ethnic groups, in vector format. The GeoEPR is a polygon
shapefile from ETH Zurich. This vector dataset consists of polygons representing the spatial extent
of politically relevant ethnic groups across space and time, specifically where that group was
demographically dominant in a given year or period. Since each conflict has a temporal element
that is directly tied to the spatial, this is valuable for determining whether the group in question
inhabited contested areas at the time of the behavior. Essentially, I used historical maps to create
pseudo-GeoEPR layers further back in time than are currently available.
A caveat with regards to the use of historical demographic maps; some utilized straight
lines to depict of areas with mixed populations. This may introduce uncertainty into the statistical
analysis of population distribution before and after the imposition of IRBs. I dealt with this
challenge by excluding the area of mixed population if the group in question was in the minority

78

and included it if that group was in the majority to align the data with that of the GeoEPR. This is
an imperfect solution, but nevertheless proved useful for displaying population change over time.
Following Chiang et al. (2014), I used strict selection criteria for choosing what historical
demographic maps could be included.
124
Maps were required to be from reliable creators, such as
well-regarded cartographers, government agencies, or geographical societies. Only high-quality
images scanned in appropriate data formats were considered. Images encoded as TIFF, GeoTIFF,
MRSID, or JPEG2000 with a minimum resolution of 300 DPI were deemed suitable. All maps
were ultimately downloaded from either the Library of Congress or David Rumley’s Collections.
Finally, all maps were required to have the following elements: legends with clear distinctions for
the group in question; a date close to the imposition of IRBs; a map scale; and features suitable for
the use of control points. These features were important for georeferencing, which was the next
step.
Georeferencing is the process by which a map is transformed to align with the projected
coordinate reference system (CRS) of an already georeferenced dataset.
125
Historical maps may
lack a CRS. They may use an archaic projection, or they may simply use a projection that is not
suitable for analysis. All layers, or datasets, in the GIS must be projected to the same CRS to be
considered fit for use.
The historical map was then loaded into the GIS. The CRS was reassigned to the
appropriate one. Each case uses a different CRS, as the most precise CRS may vary from region
to region. Table 3.1 contains the CRS for each region:

79

Table 3.1
Projected coordinate reference systems (CRS)
Ethnic Group Projected CRS
EPSG
Albanians 9141 KOSOVAREF01 / Balkans zone 7
Chechens 2531 Pulkovo 1942 / 3-degree Gauss-Kruger zone 15
Koreans 5179 Korea 2000 / Unified CS
Kurds 3893 ED50 / Iraq National Grid

Note. The EPSG (European Petroleum Survey Group) Geodetic Parameter Dataset designation is
a standardized code for a projected CRS.

The next step was to transform the map. This stage aligned the map and overlaid it onto
the known features of other georeferenced data. In other words, it brings differences in projections
into alignment and “corrects” those differences. I assigned control points to known features on
both datasets and then used rubbersheeting to “stretch” the historical map into place, making it
conform to the curvature of the Earth and the other projected CRS, which results in a better fit.
126

Control points are objects that are common to both maps. They should be features that are more
permanent in nature. As an example, a historical map and a known referenced map may each have
a mountain peak. That would be a potentially useful control point because it can be referenced on
each map.
For known features I did not use coastlines, as these are subject to change by erosion or
anthropogenic transformation. Nor did I use riverine objects as these are sometimes diverted by
human activities, such as the building of hydroelectric dams or agricultural development. One
feature I used instead was generally quite ancient: the old town or citadel of cities. As these are

80

often thousands of years old, they have remained in the same place over time. Gazetteers, such as
those compiled by the US Central Intelligence Agency (CIA) that are available from the Library
of Congress, assisted in locating the coordinates of old towns and citadels. The accuracy of the
coordinates was manually verified using Google Maps.
Another type of control point was an IRB when applicable. Once established, IRBs are
infrequently changed. The IRB involving the specific ethnic group was not always usable, as it
may not have been imposed yet. However, contemporary IRBs between other states at that
particular time were sometimes present, so they were used when available. Mountain peaks were
another useful type of spatially constant control point as the historical maps contained terrain
features and were able to be located in both maps.
Control points were established at the corners of the historical map and at several places in
the interior. A total of eleven control points were used for the Kurdish and Chechen cases. Twelve
were used for the Albanian case. As there were no maps to transform for the Korean case, no
control points were used for it.
Following the establishment of control points, the maps were subjected to rubbersheeting.
Rubbersheeting accounts for the curvature of the earth and adjusts the historical map to have a
smoother fit over the georeferenced map. Appendix B contains full-sized images of the maps used
in this analysis. Figure 3.1 below displays the processed maps to illustrate the visual result of the
above steps:

81

Figure 3.1
Georeferenced and processed historical demographic maps

82

Note. From top: Albanians, Chechens, and Kurds. These are listed alphabetically, though they will
be examined in reverse order in the case chapters. The vivid colors (purple, purple, yellow,
respectively), are the processed extent for each population. The same overall color
corresponds to the original color for the demographic map. See the relevant appendix for
each case for the original map.

Once the map has been georeferenced, the distribution of the ethnic group in question was
traced in the GIS. It was then statistically compared to the distribution of the ethnic group after the
imposition of IRBs, hence the temporal component. The territory that remained unchanged, as well
as the territory that was gained or lost, was computed over time. A new layer was also generated
from this data for visual inspection. The final maps are shown in their respective case chapter
along with accounts of migration and other points of interest that resulted from the analysis. Each
is accompanied by a table displaying the approximate amount of territory gained or lost by the
group in square kilometers and percentage. At this point, the analysis was ready for process tracing
based on the evidence provided by the populational and territorial change.

83

Process tracing involves establishing qualitative linkages between the EV and the DV,
127

which, in this instance, consisted of the Roots of Rivalry (P1) hypotheses and the onset of
organized intraethnic conflict among rival factions, respectively. The temporal analysis provided
cues for where to focus historiographic and archival review for the group as a means of discovery
as to why the factions may come to exhibit certain behaviors toward each other. For instance,
ethnic groups that are divided across multiple states but may remain stateless themselves could
have substantial territorial gains in one state, but losses in others. This may characterize the
relationships between factions differently depending on the nature of displacement.
The temporal analysis of the change in population and territory homes in on where process
tracing of the mechanisms of the three hypotheses for P1 might be best supported. For example, a
loss of territory could have meant that urbanization, forced relocation, or other means of
depopulation occurred and should be noted in the historical record. A gain in territory could have
intoned a relocation in which ethnic ties became transnational in scope, or, conversely, were
severed. The workflow for the first stage of the analysis is depicted in figure 3.2:

84

Figure 3.2
Analyzing populational-territorial change of ethnic predominance over time

Note. Created by the author using https://www.diagrams.net/

Evidence for each pathway was identified or rejected as part of the process tracing stage.
Evidence for first image factors, like personal disputes between leaders; second image factors,
such as the successful instigation of intraethnic conflict by state actors; or third image factors,
especially system-level divergent development, were noted in the historiography and in the
factional media found in each case. A useful starting point for this evidence came from examining
the changes in population and territory revealed by the temporal analysis.

85

Methods: The Effects of Distance (P2) on conflict objectives
The second stage of the empirical strategy examined how intraethnic conflict occurs.
Particularly, the Effects of Distance (P2) were tested through a variety of statistical methods
commonly employed in the disciplines of International Relations and Political Science, as well as
less common spatial analytic techniques. As with the rest of the empirical strategy, these methods
were complemented by qualitative analysis.
In brief, this stage consisted of data collection, data pre-processing, EDA, EDSA, SDA,
SM, and qualitative research into the mechanisms suggested by the results of the statistical
analyses. First, data were collected for each variable.
128

Pre-processing was required following data collection. Some GED events were coded with
lower confidence rankings than others. Those observations were checked with the source and then
manually verified in Google Maps. Their dates were also compared to the time frame for known
theatres in the conflict. If manual verification yielded an event that was fit for use, its confidence
was considered upgraded. Observations were discarded if sufficient information was not available,
and confidence remained low. For instance, events were considered inaccurate if the georeferenced
location led to the centroid of an areal unit—such as a governorate when the event is supposed to
have occurred in a specific town—and thus were not included.
Once accepted, these data required additional pre-processing to be considered fully fit for
use. Most of the descriptive information was cleaned because of redundancy or because it was
irrelevant. Then, the names of administrative districts required alteration to match the spelling of
the same units across all datasets per case so that they could be spatially joined without issue, or
for building spatial weights matrices.

86

Any use of district-level data must be undertaken with care because of the modifiable areal
unit problem (MAUP). The MAUP refers to the statistical bias that may introduced to analyses
based on the selected level of aggregation.
129
Areal units are often of non-uniform sizes. They are
often assigned based on human importance, rather than for any inherently geographic reason. For
instance, Census tracts may be wildly different in size, so aggregating observations into the tract
will affect outputs of statistical tests, especially when units of the same level of aggregation are
compared to each other, despite their differences in physical size.
Consider choropleth maps, for example. Such a map contains areal units that are color-
coded according to the aggregation of observations within the confines of that areal unit, often
based on the number of observations within that unit. This may convey an inaccurate picture of a
phenomenon because of the MAUP. Physically large, yet sparsely populated areal units that
nevertheless have a few large population centers may appear to have low occurrences of some
phenomenon. In reality, however, it could have many observations within one or a handful of
locations. However, its color code denotes uniformly few occurrences because of its large size.
Such a maps portray a uniform distribution throughout an entire areal unit. Figure 3.3 and its
associated subfigures display some choropleth maps, and the narrative below explains some issues
with this variety of map and associated statistical approaches.

87

Figure 3.3a
Choropleth map: Total population of the USA by state

Note. The 48 contiguous states and Puerto Rico are displayed. Visualized via the US Census
Bureau with the “American Community Survey data (SEX AND AGE—Total
Population—Estimate) 2018 Data Profile” dataset.

88

Figure 3.3b
Choropleth map: Population of California by county

Note. Visualized via the US Census Bureau California Hard-to-Count Index Interactive Map with
the “California-Level Population, Education, and COVID-19 Data – California County
Data” dataset.

89

Figure 3.3c
Choropleth map: Population density of California by Census tract per square mile

Note. Visualized via the US Census Bureau California Hard-to-Count Index Interactive Map with
the “MHE_CA_PopulationDensity_tract_2015_tiles” dataset.

In figure 3.3a, we see a choropleth map of the population of each state in the US.
Obviously, the size of an individual state can vary greatly. The small number of bins (the categories
for population size in the legend) accords relative importance to states that may have a greater
disparity in population than visual inspection alone—without additional research—suggests. The
bin for California and Texas is a good example of this issue. Texas had approximately 28 million
inhabitants in 2018, while California had roughly 39 million. That gulf is massive, larger than the
entire population of most states, yet the choropleth bin displays them in the same category, which
can have an impact on statistical analyses.
130

90

An additional problem with this display of data is its tendency to impart uniform
distributions to areal units. Let us move to figure 3.3b. The population of California by county
suffers from the issue of bin size, like with figure 3.3a. It also demonstrates the disparity between
the sizes of counties. It illustrates how a county appears to have a uniform distribution of
population. The massive county just east of Los Angeles is San Bernardino County, which is the
physically largest county in the contiguous 48 states. Visually, it appears to be a heavily populated
county. I have a degree of familiarity with that county, having spent time at the Marine Corps Air
Ground Combat Center, Twentynine Palms, in the early 2010s. From contextual experience—
often a key factor in determining spatial analytic parameters—I can deduce that figure 3.3b has an
issue with displaying accurate population statistics, as the county is sparsely inhabited outside of
the tiny southwesternmost portion. Statistical analyses involving interpolation at the county level
will be heavily biased because of the level of this aggregation.
Figure 3.3c represents an improvement. This is displayed at a much smaller level of
aggregation, namely the US Census tract. Looking at the same county, San Bernardino, just east
of Los Angeles, it is apparent that much of the county is uninhabited or sparsely inhabited. This
provides one potential solution to mitigating the MAUP: the use of the smallest level of
aggregation to make statistical inference. However, this solution may not scale upward if the same
analysis is required at a greater level of aggregation—say, for instance, census tracts in multiple
states that have tracts of different sizes. Thus, this is an imperfect solution that may complicate
further tests.
Another solution to mitigating the MAUP is to utilize disaggregated point data at the
individual event level. In other words, analyzing events where they occur, rather than aggregating
them to an areal unit for the purposes of analysis. Aggregation for visual purposes may still be

91

useful to reducing the sheer amount of data for illustrative purposes. For instance, the analytical
maps displayed in each case chapter are visually aggregated for the convenience of the reader, but
they are not statistically analyzed as such.
There are also, however, potential issues with this method, as the availability of
volunteered geographic information (VGI) and the democratization of geospatial science brings
new uncertainties.
131
The solution to this additional consideration is to interpret the reliability of
the author of such data. To avoid the VGI problem entirely, I used data at the individual level but
did not include VGI. Only governmental, IGO, NGO, academic (namely the GED), wire, or
factional sources were considered to reduce questions of reliability.
The final pre-processing stage consisted of generating “near tables” of distances measured
in kilometers from key points of interest (POI) to individual events. The POI came from qualitative
research into the conflict and were selected based on the theoretical framework’s logic for
inclusion. They consist of national capitals, regional centers, official and unofficial border
crossings definitively proven to be in use at the time, and transborder mountain passes used by
smugglers and TCOs that were known to be affiliated with a faction. There may have been other
crossings available in some form, but if they could not be verified, they were not included.
Additionally, crossings that opened after the time period were obviously excluded.
Following pre-processing, I moved on to EDA. Summary statistics were computed and
examined. Testing for correlation was the first step. This stage consisted of the generation of a
histogram, a scatterplot, and the covariance between pairs of variables. A Pearson correlation score
between 0 (no correlation) and 1 (perfect correlation) was generated along with a statistically
significant pseudo p-value.

92

The correlation relationship was divided into two models. The first was a regional center
model, where the correlation between the occurrence of conflict events and distance from the
national capital was compared to the occurrence of conflict events and the distance from the
regional center. The second model was the border model, which tested the correlation between the
occurrence of conflict events and the nearest border-related POI. Both models also contained the
correlation between the occurrence of conflict events and the distance from the primary route
(highway). The results of these correlation tests are found in the appendices for the relevant case.
The next stage of EDA was ordinary least squares (OLS) regression. The OLS models were
fit in accordance with the correlation models. The National Capital (H2a) and Regional Center
(H2b) hypotheses formed the first model, while the Unpacific Borders hypothesis (H2c) was the
basis of the second model. The impact of Primary Routes (H2d) was included in both models.
The occurrence of conflict events was the dependent variable in both models. To improve
the quality of these models, they were then subjected to stepwise selection based on Bayesian
information criterion (BIC) and logarithmically transformed to make the continuously varying
distance tables conform to a normalized distribution. The results of the OLS and log-transformed
models may be found in the appendices for the relevant case. Plots for the log-transformed models
are also in the appendices.
Spatial data are special; it is assumed that they do not meet the linear model’s assumption
of independence. This is to say that an underlying spatial process is sometimes dependent on
itself.
132
Thus, the utility of the OLS model—even if log-transformed—is suspect without
exploring the data for the presence of the spatial processes of autocorrelation and clustering.

93

Tobler’s First Law of Geography states that all objects are related to each other, but nearer
objects are more related than distant objects.
133
Moran’s I statistic is an ESDA technique that tests
for spatial autocorrelation.
134
Spatial autocorrelation means that objects in close proximity have
similar values. To compute Moran’s I, a spatial weights structure was created for the conflict data
in each model based on the geographic neighborhood of observations. The neighborhood followed
a nearest neighbor structure, which is to say that the weighting system consisted of the number of
neighbors each observation has. The number of neighbors was derived from the square root of
observations in a given dataset. All such tests were subject to a Monte Carlo simulation of one
thousand iterations.
The null hypothesis was CSR, in which there was no autocorrelation. Resultant Moran’s I
statistics may range from -1 to 1, in which 0 is CSR. A score of -1 means complete dispersal; data
are completely unrelated. A score of 1 indicates total autocorrelation. Spatial autocorrelation is
indicated by positive Moran’s I statistics that are coupled with positive z-scores—which are
standard deviations above the mean—and statistically significant pseudo p-values.
The global Moran’s I statistic test was refined through Empirical Bayesian Smoothing
(EBS).
135
This technique is designed to eliminate outliers that may result in a skewed global
Moran’s I index. Moran scatterplots were also generated for the Moran’s I tests. The presence of
autocorrelation or lack thereof is reported in the case studies. The tables and scatterplots for these
analyses are located in the appendices for the relevant case chapters.
When there was a lack of spatial randomness, i.e., there appeared to be spatial
autocorrelation, the data was subjected to Getis-Ord General G statistic and Getis-Ord Gi* statistic
tests if necessary.
136
The General G statistic is a global measure that computes high or low
clustering trends within the entire conflict event dataset., while the Gi* statistic identifies localized

94

hot spots or cold spots. Hot spots occur when high values are surrounded by other objects with
high values. Objects with low values that are surrounded by other objects with low values are cold
spots.
The G and Gi* statistics also include a z-score and a p-value to determine whether the null
hypothesis of CSR may be rejected. I indicate whether data was clustered in the case study
chapters. The report for the General G statistic may be found in the appendices of the relevant
cases. The Gi* statistic was computed if the General G returned no evidence of clustering, or CSR.
In this instance, the Gi* statistic is a visual result that displays local events of clustering that are
not necessarily statistically significant, which is why the General G statistic may not have
identified the presence of clustering.
When spatial autocorrelation and clustering was detected I moved on to a spatially lagged
autoregression model (or spatial lag model) which spans SDA and SM techniques. The spatial lag
model consists of a maximum likelihood estimation (MLE) regression model that incorporates the
neighborhood structure of the data’s spatial weights matrix.
137
The inclusion of this matrix
modifies the dependent variable by accounting for spatial autocorrelation, thus making it more
suitable for inferential reasoning than the OLS and log-transformed regression models alone. If
the spatial lag model resulted in an Akaike information criterion (AIC) that was lower than the
other regression models, it was deemed a better fit. The results of the spatial lag models are placed
in the relevant case chapter.
The DV was the occurrence of conflict events. The EVs are drawn from the hypotheses for
the Effects of Distance (P2) grouping: the distance from the national capital, distance from regional
centers, and the distance from the border or a border crossing. Also, the distance from the primary
route was included to determine whether highway access was a salient factor.

95

Additionally, a test for residual autocorrelation and a studentized Breusch-Pagan test were
performed.
138
These robustness checks account for the data’s potential to violate the assumption
of linearity. Thus, a potentially homoscedastic distribution is rendered as a heteroscedastic one. I
remark on the robustness checks in the case studies, and the results for them are found in the
relevant appendix. The workflow for this stage of the analysis is found in figure 3.4:
Figure 3.4
Assessing the effects of distance on intraethnic conflict among rival factions

Note. Created by the author using https://www.diagrams.net/

96

Methods: Out-group Incursion and cooperation among rivals
A systematic review of intraethnic conflict would be incomplete without accounting for
rival cooperation. Groups in conflict with each other may abruptly cease combat. An earlier
generation of literature suggests this is most likely to occur during periods of out-group incursion,
though this outcome does not always materialize. This subsection describes the methods used to
test the effects of an out-group incursion on cooperative action.
This work is inherently geographically oriented as it deals with issues of territoriality and
populations of coethnics. To test the effects of Who Dies and Where They Die (H3a) on rival
cooperation, I began with kernel density estimation (KDE) analysis, which is an ESDA technique.
Planar KDE identifies clusters of events within a distribution of data based on the weights accorded
to each instance.
139
The weight is created by the number of neighbors each point has. The kernel
function calculates the decay impact as the neighbors become located further apart, which
determines the visual density of events. The decay impact makes it efficient for considering
Tobler’s First Law of Geography as it recognizes the relationship between near and far objects.
The second component of KDE is the user-selected bandwidth, or the distance by which
events are considered neighbors. This may be influenced by the data, or by qualitative knowledge
of the phenomenon being assessed. In this instance, the bandwidth, or radius, was the number of
nearest neighbors per point. The count of nearest neighbors was used to generate clusters, or
kernels, of events. The resultant size and color coding for each kernel was based on each event’s
proximity to other events. The kernels were those of factional cooperative action. For instance, if
rival factions both sought to repel an out-group from a threatened population center, the kernel
was large and brightly colored because there were many cooperative events over a short distance,

97

but darker when cooperative events were more dispersed. The results of the KDE for cooperative
action are located in the appendices for the relevant chapters.
The KDE was tested for the probability that complete spatial randomness (CSR) was at
work. Though unlikely, there is a possibility that omitted variables were at work, thus making
cooperative action appear to be the result of one of the hypotheses for Out-group Incursion. The
tests for CSR consisted of K- and L-functions analyses.
140

The K-function compared the observed values from the KDE plot to a hypothetical trend
of values, along with a confidence envelope that represented the range of expected values under
randomness. The more the observed trend varied from the hypothetical trend, the less likely those
values were to be random. The L-function is very similar to the K-function but is more useful for
smaller numbers of observations. It returned the observed values as a line, with an envelope for
CSR, along with a confidence interval. The point on the line in which the observed values are
farthest from random should point to an ideal distance between kernels. That is to say, conflict
event clusters that are further apart than the L-function predicts should not be considered related.
The results of the K- and L-functions are found in the relevant appendices.
One additional test for CSR included an average nearest neighbor analysis (ANN), an SDA
technique. The ANN generates 100 randomly generated Monte Carlo simulations of alternative
selection models.
141
The alternative selection models are made with random, independent data.
These data are then compared with the observed values. The deviation is assigned a pseudo p-
value. Results of ANN analyses are located in the appendix for the case chapter.
When the deviation of the observed values from expected values was large and statistically
significant, the Second-order effects of spatial dependence was supported. This is to say that the

98

events were related to each other. Great differences between observed and expected values
indicated that spatial randomness could not adequately account for the patterns in the data that
emerged. Such a result also meant that it was appropriate to test for first-order effects.
First-order effects are the product of underlying spatial processes.
142
A first-order effect is
identified when a clustering of events is caused by the clustering of a different event. By way of
an example, let us consider Love Canal, a former neighborhood of Niagara Falls, New York.
143

Electro-chemical manufacturers, defense agencies, and others deposited toxic waste in the area
that eventually became Love Canal, unbeknownst to the future inhabitants. They and their children
suffered serious genetic damage, birth defects, and illnesses.
Love Canal is an excellent and tragic illustration of both first- and second-order effects.
The intense clustering of leukemia absent any causal process in one particular neighborhood
demonstrates a second-order effect. Such a clustering could not be possible under spatial
randomness. The instances of leukemia, or events, are related to each other and have a discernible
decay impact as one moves further from the area, which fulfills the requirements of a second-order
effect. The underlying process is the depositing of toxic waste. This is also a type of event that
occurs in a clustering pattern. By considering the underlying spatial process, the resultant
clustering of leukemia is actually a first-order effect because it would not cluster without the first
instance of toxic waste clustering.
Let us now relate this to the variables of the study at hand. Instances of armed conflict or
rival cooperative action are variables that may be spatially clustered. Intense clustering indicates
that the events are related to each other, a potential second-order effect. This is especially the case
if there is a discernible decay factor and events become less clustered as distance from a kernel of
conflict or cooperative action increases. Killings committed by out-groups may also occur in

99

clusters. If clusters of killings of neutral co-ethnics, rivals, and all non-combatants can be
identified, the clustering of conflict and cooperative action can be compared to those clusters of
deaths. Thus, conflict or cooperation could be first-order effects that result from out-group action.
For the Love Canal analogy, the actions of out-groups are equivalent to the depositing of varieties
of waste. Cooperative action and conflict events (or the cessation of the same) are results of those
out-group actions, much like the negative health outcomes are the results of the depositing of
certain types of waste.
If the KDE and the K- and L-functions appeared to be non-random, and the ANN was
statistically significant, second-order effects were established. To test for first-order effects, a non-
stationary Poisson point process (PPP) model of covariate effects was used in tandem with a
likelihood ratio test (LRT).
144
This SDA and SM technique tests for first-order effects that
underlying spatial processes may have on a dependent variable with user-specified parametric
regression models. The DV was made up of cooperative action events. The EVs were the deaths
of unaligned coethnics, rival coethnics, and all non-combatants.
And so, the models for this stage consisted of the following: 1—the null hypothesis (CSR),
in which no variable has an effect on cooperative action; 2—the deaths of unaligned coethnics
drive cooperative action; 3—the deaths of rival coethnics drive cooperative action; and 4—the
deaths of all civilians, regardless of affiliation, drive cooperative action. Recall that civilians may
count as rivals under the second and third models if the known dead belonged to a political party
or its armed forces, while they do not count as rivals under the fourth model. The results of the
PPP and LRT are probabilities of cooperative action based on each hypothesis. The results are
reported in the case chapters.

100

The formulae for all quantitative tests are found in Appendix C. Figure 3.5 below displays
the workflow for this third stage of the analytical process:
Figure 3.5
Assessing the effects of out-group incursion on cooperation among rival factions

Note. Created by the author using https://www.diagrams.net/

There are three additional details to discuss with regards to the study of conflict and
cooperation. First, qualitative evidence was included to support the statistical analyses. As with
the other sources of qualitative evidence, this consists of governmental, IGO, NGO, and factional
sources. When possible, factional sources in the original languages were used, either with my own

101

language skills or with the aid of interlocutors. Interviews for context were also conducted when
possible with faction leaders, fighters, and unaligned persons.
145

Second, analytical maps were produced using a GIS for each period to provide a
visualization of the processes at work. These maps were created with data collated from the DoD,
SOHR, HRW, UNHCR and the GeoEPR. The maps depicted the maximum extent of territorial
claims by the relevant factions, as well as territory threatened by out-group incursions, fatalities
incurred per conflict, and key geographic features.
Territorial control and analytical cartography are inherently linked.
146
This work accounts
for populations both outside and within territorial control by states. Conflict events in the analytical
maps have been graphically aggregated to the nearest population center to avoid burdening the
viewer with “hyperactive optical clutter”.
147
Displaying each event would not give the viewer a
more illuminating picture of the conflict. They were not statistically analyzed as such, as it would
be imprecise.
Recall also in the discussion of the MAUP above, a choropleth map has limitations.
Namely, choropleth maps are useful for a visual reference to a specific problem, but they tend to
accord equal importance to geographic units.
148
Therefore, administrative areal units were not used
to display territorial control.
Analytical maps were utilized for each period to provide an overview of the deaths of rivals
and those not apparently affiliated with a particular faction as a visual reference for the reader. As
such, they are not intended to model each development in the conflict itself through precise time
steps. Yearly or monthly time steps are still insufficient in modeling extremely fluid
battlespaces.
149

102

Those killed by out-group action were displayed as a variable bar graph representing the
total killed for all factions, while cooperative action was displayed by intensity using KDE.
Territorial control was always color-coded to represent the party colors of the relevant faction.
150

Analytical maps are displayed for the reader’s convenience in each case chapter.
Finally, the exact spatial and temporal scales in each case chapter broadly conformed to a
particular scheme to make them comparable to each other. There were some variations between
exact resolutions and scales because of the nature of group size and dispersion. The first stage of
the analysis was performed at the level of the maximum spatial extent of the group. Its temporal
scale ran from the available period of demography until the beginning of the conflict periods in
the second and third stages.
The second and third stages of the analysis were performed at the spatial scale of the
conflict. For instance, analyses in Chapters Four and Six were conducted at the subnational scale,
but within multiple states. Chapter Five was at the subnational scale within a single state. The
national scale was used in Chapter Seven. The temporal scales for both stages ran from the onset
of conflictual or cooperative action until the conclusion of the phenomenon of interest, if such a
conclusion has indeed been reached.
Only data that matched the spatial and temporal scales were used. While the GED contains
data from 1975 until the present, those that lay outside of the conflict periods were removed from
the dataset. Additionally, those that lay outside of the spatial scale for the conflict (say, for instance,
an assassination of a dissident on another continent) were removed by clipping them from the area
of study.

103

3.3 Case selection
An individual case should study a population that is conceivably somewhat representative of a
broader group, for which evidence of a given phenomenon should be observable from a variety of
methods when explored in depth.
151
This work required the construction of four ideal case types
to provide a degree of generalizability to other groups possessing similar characteristics to the ideal
types. The four types are: a divided stateless nation (DSN); a single-state separatist (SSS); a
divided irredentist group (DIG); and the two states, one nation (TSON) phenomenon.
A quantitative analysis alone does not provide the richness of an in-depth analysis of
theoretical mechanisms. Therefore, I included qualitative analyses of four representatives of the
ideal types. Each case study opens with an overview of the history of the group. The roots of
conflict are assumed to stretch further back than the end of the Cold War, and will be taken into
account, but the dissertation will mostly focus on post-Cold War conflicts because the New Wars
literature informs us those conflicts with a dimension of identity have somewhat superseded
traditional interstate warfare. The case studies present the evidence for the causes and processes
of intraethnic rivalry, conflict, and cooperation. I used a small-n design in order to combine
qualitative research and statistical analysis to produce a fuller picture of the mechanisms at work.
Four permutations of ethnic groups with politically relevant factions were chosen based on
separation by a border and statehood status, rather than by which ethnic groups engage in
intraethnic conflict or cooperation. This was to avoid selecting on the dependent variable.
152
This
work’s contribution, the DSN, consisted of a stateless nation divided by border into multiple states
in which it formed the core community of no state.

104

The remaining three permutations came from the literature. One was required to be a
stateless nation not divided by a border, which is the SSS. This was expected to provide a contrast
to outcomes of statelessness that may or may not be influenced by a border. The SSS was derived
from the ethnic defection literature
153
and Horowitz’s (1981) exposition on separatist conflict. The
next case involved a state actor with coethnics residing in a neighboring country divided by a
border, whether there was direct or indirect assistance by the state actor to the non-state irredenta.
This variety is the DIG, which came from the study of irredentist ethnic conflict.
154
Finally, an
ethnic group divided into two states constituted the fourth case, which is the TSON. The TSON
type origin was drawn from the divided society literature.
155

Since selection involved ethnic groups, a means of narrowing down the potential cases
became essential. Depending on criteria for what constitutes an ethnic group, there were potentially
tens of thousands that may have been eligible for selection. I proceeded by including involvement
in a conflict in which ethnicity was considered a dimension, but not specifically intraethnic
conflict. This avoided selecting on intraethnic conflict (the DV) and also avoided considering the
overwhelming majorities of ethnic groups that do not engage in the patterns sought for study here.
At the same time, it demonstrated that there are factions within that ethnic group that are capable
of mobilizing for violent collective action. Additionally, all stateless groups or the irredenta of
groups with a representative nation-state were required to have a form of self-determination claim,
either autonomy or independence. This was to ensure that selected cases could conceivably fit into
the categories drawn from the literature; a group would probably not be properly termed separatist
if it does not seek separation.
For preliminary data, I consulted the Center for Systemic Peace
156
for periods of ethnic
violence and warfare. The presence of politically relevant factions was a requirement for

105

consideration, which is why I excluded communal ethnic violence from consideration as it is not
necessarily organized by such factions. Since warfare with an ethnic component is a facet of the
New Wars literature, I chose from post-Cold War instances, though conflicts with Cold War
origins were also considered. The goal of the project is not to point out how conflict in general has
changed since the Cold War, as it assumes the New War literature is at least partially correct as
evidenced by the decline in interstate war and the persistence of intra-state conflict. Rather it is to
examine DSN conflict and cooperative patterns employing this literary subfield’s assumptions.
For ongoing conflicts with Cold War origins, cases were considered if there have also been periods
of conflict after the end of the Cold War.
The criteria for selection used in this work involved the following queries. Is the group in
question stateless? Is it contained within borders or divided by them? Does it have a claim to self-
determination? Is the crisis in question communal, factional, or state-directed?
The last question revealed whether the groups involved formed the core component of the
government of the host state, which may have been disqualifying—unless the group is a TSON. It
also delineated groups with a unified political-military force from politically irrelevant groups
competing over the more traditional goals of ethnic conflict, such as those identified in greed and
grievance explanations.
Although including communal violence would be an excellent extension of this project, it
is, at present, outside of the scope. Explicit relations to the host state are an integral part of the
analysis. Any group that is or has been involved in post-Cold War communal ethnic warfare or
violence may move from eliminated for this study to any of the case types with future
developments, such as the formulation of a self-determination claim or the advent of political-
military unification into a relevant faction as defined here. Past conflict may lead to future

106

conflict,
157
especially through a hardening of identities resulting from such violence,
158
so this is a
possibility for future work. Appendix D contains more detailed information on the case selection
process, including a ranking system used for defining a universe of potential cases.
The information gained from asking the above questions was gathered and modified from
data available from the CSP. Conflict Year(s), State(s), Type(s), and Group(s) were retrieved
directly from the CSP. However, I expanded State(s) to include cross-border conflicts, Type(s) to
include sectarian ethnic warfare and labeled communal ethnic violence as a discrete category. I
rejected cases of civil war. In this instance, civil war refers to the contestation of control over the
central government that lacks an overt ethnic component. The categories of Stateless, Border, and
Self-Determination Claim are based on research into each ethnic group in question. I identified 53
conflicts involving more than 60 distinct groups. Table D.1 in Appendix D displays the full table
of available cases.
Groups with ambiguous values were eliminated for this particular study, simply because
settling the issue of ambiguity is a research agenda that lies beyond the scope of this project. They
are designated as potential cases for future study. Groups involved in communal violence were
excluded because such groups lack a unified political-military faction, which is the unit of analysis
in this project. Groups involved in conflicts with Cold War origins were considered if there were
violent flare-ups in the post-Cold War period. Ultimately, none of these were selected for other
reasons. For instance, the Palestinians are perhaps not truly stateless, given how many states
recognize their sovereignty, and I lack the expertise on the Sindhi that I have relative to the Kurds.
Selecting for the four types, cases were considered if they fulfilled the following
conditions:

107

1. A stateless nation divided by a border (DSN).
2. A stateless nation not divided by a border (SSS).
3. An ethnic group divided by a border but forms the core population of a state (DIG).
4. Two states comprised of the same ethnicity divided by a border (TSON).
These conditions for selection ensure that the dynamics for each permutation of statehood (or lack
thereof) and division (or lack thereof) may be captured. Table 3.2 displays the types and their
features, as well as the ultimately selected example of each:
Table 3.2
Typology of ideal types of intraethnic conflict and cooperative patterns
Term Division Core Ethnic Other ethnic groups
by IRB population group
of a state(s) selected

Divided stateless nation Yes No Kurds Balochi, Kanuri,
(DSN) (KPFs) Rohingya, Sindhi,
Uighurs, …
Single-state separatist No No Chechens Acehnese,
(SSS) (CSFs) Dagestanis,
Ingushetians,
Oromo,…
Divided irredentist group Yes Yes/No Albanians Armenians, Azeris,
(DIG) (AIFs) Russians, Serbs,
Somalis,…
Two states, one nation Yes Yes/Yes Koreans China/Taiwan,
(TSON) (KSRs) divided societies
(Viet, Germans,…)

Note. The DSN is proposed as a new form of intraethnic competition. Ethnonationalist factions
are the units of analysis, rather than the group as a monolith, which may have
considerable variation in the degree to which claims of self-determination along ethnic
lines may be supported or unsupported.

108

There were 18 potential conflicts involving DSNs. Conflicts involving Pashtuns and
Palestinians were rejected based on the ambiguity of statelessness and self-determination claims.
Calls for a Pashtunistan are less frequent and not as clearly articulated as those of other DSNs,
while Palestine is recognized as a sovereign state by 138 UN member states. As such, these are
reserved for future study pending a definitive classification. Several, such as Uighurs and
Rohingya are primarily victims of state-directed ethnic cleansing, without sufficiently unified
political-military factions, and so are not (currently) ideal candidates for this work. Should they
develop such features and greater cross-border interactions take place, those groups will be ideal
candidates. The Kanuri ethnic group spans four countries: Nigeria, Niger, Cameroon, and Chad.
This group is the primary—though not sole—component of Boko Haram, and they would be a
prime candidate for a future comparison across DSNs. The Sindhi and Balochis are also candidates
for a cross-comparison in future work, as are several groups in the Sahel and the Caucasus. The
final choice came down to my knowledge of Kurdish customs and a dialect of Kurmanji, or the
Northern Kurdish language, which make that group the optimal choice.
Additionally, the Kurdish case fits Gerring’s (2007) definitions of an extreme case, as well
as a within-case study, because there are multiple conflicts with various actors that provide several
cases, rather than only one. They are an extreme case because they have experienced a long history
of division by borders into four regions, and the vast differences between their host states make
for a series of cases within one group that may provide a wealth of benefits in observing the micro-
foundations of causal inference offered by a variety of methods.
The nine potential SSS cases were: Abkhazians, Acehnese, Chechens, Chiapas,
Dagestanis, Ingush, Mai Mai, Oromo, and Ossetians. The conflicts involving the Acehnese, Mai
Mai, Oromo have ethnic components, but are also strongly ideological. The indigenous armed

109

insurgency in Chiapas, Mexico, was short-lived and so is not, perhaps, as informative as the
separatist movements within Russia’s borders. From these candidates, data on the Chechen wars
was most readily available and they have been involved in comparatively more conflicts, so the
Chechen factions were selected.
For the DIG cases there were conflicts involving eight groups from which to select. These
included Russians in Ukraine; Somalis in Ethiopia; and Armenians and Azeris in the Nagorno-
Karabakh conflict; Croats, Serbs, Bosniaks, and Albanians in the various phases of the
decomposition of the former Yugoslavia comprise the rest. Russian irredentism in Ukraine and the
groups involved in the Nagorno-Karabakh conflict were eliminated because the SSS also involves
Russia and the Caucuses. I aimed for a more geographically diverse sample. The conflict in
Ethiopia’s Somali Regional State did not have as much data relative to the conflicts in the Yugoslav
Wars.
The interrelated conflicts involving the remaining groups may be classified as the Yugoslav
Wars, in which there was the Slovenian War of Independence, the Croatian War of Independence,
the Bosnian War, the Kosovo War, and ethnic Albanian insurgencies in Serbia and then-
Macedonia. In each of these conflicts, external forces, such as NATO, the United Nations
Protection Force (UNPROFOR), and the Kosovo Force (KFOR) provided significant material and
logistical assistance to groups that were perceived as besieged. While an ideal case would not have
these external forces in order to study the group in isolation, it is not possible in these instances as
there is an inescapable international component to such conflicts.
159
Comparatively fewer ground
troops were involved in the Kosovar War
160
than the other conflicts, so the Albanian case was
selected.

110

Finally, the selection of the TSON case was straightforward. Only the China-Taiwan and
North Korea-South Korea dyads survived the Cold War as examples of the one nation, two states
phenomenon. While there are only two cases, these are nevertheless central actors in potential
hotspots for a new global conflict. As only the two Koreas are separated by a land border, this
dyad was selected. A future extension could examine whether separation by land, as opposed to
sea, leads to different outcomes across TSON dyads. The history of the group and its Cold War
conflict will be studied, but post-1991 relations will be the primary focus of this case study.
3.4 Plan of the case chapters
The case studies will proceed in the following order of ideal types: DSN, SSS, DIG, and TSON.
Each case was subjected to the same tests, both quantitative and qualitative, to the extent possible.
This section presents an overview of each of the cases examined in this work.
Kurdish political factions (KPFs)
The Kurds are a stateless nation that have been officially divided by borders for a century yet have
existed in the historical record for millennia. Since their division by a border, and long before it,
they have been embroiled in wars of conquest, independence, and secession. They qualify as a
within-case study of a DSN as they feature prominent examples of conflict occurring since the end
of the Cold War in multiple states involving multiple factions.
The KDP-PUK rivalry included what is commonly referred to as a civil war. This conflict
occurred within the borders of Iraq, but it involved Kurdish factions coming from Turkey and Iran
to fight with other Kurdish factions. The KNC-KCK rivalry, in which two transnational umbrella
groups have clashed and continue to clash with each other, is another such case of a stateless nation

111

engaging in conflict and cooperation across international borders. In this instance, the states are
Iraq and Syria. Both periods of conflict were interrupted by conflicts with out-groups.
The KPFs, which represent the DSN, provide the main contribution of this work. If DSN
conflict and cooperation are dissimilar to the other forms, it should be considered a variety of
ethnically involved conflict. These factions are the subject of Chapter Four.
Chechen separatist factions (CSFs)
A stateless nation with a history of separatism, but not divided by an IRB, Chechen factions have
witnessed several wars with Russia since the end of the Cold War. The primary time periods for
study will include the First and Second Chechen Wars—sometimes used collectively as a conflict
continuum due to the very brief interwar period—as well as the more recent conflict between the
Chechen government and the Caucasus Emirate. Chechen factions represent a quintessential
SSS—a stateless nation not divided by an international border—with politically relevant factions
that ruled the Chechen Republic and the Chechen Republic of Ichkeria, among others. The CSFs
have frequently fought against each other. This group also witnessed incursion by an out-group in
the form of the Russian military.
The study of CSFs will allow for a comparison between the DSN and the SSS to ascertain
how similar or different their intraethnic relations are in terms of development, competition,
conflict, and cooperation through their shared quality of statelessness. Division by borders,
however, are kept separate. CSFs are not divided by borders, but KPFs are so divided. CSFs are
examined in Chapter Five.

112

Albanian irredentist factions (AIFs)
Albanian factions are the proxy for the DIG. The Kosovar Albanians are not a stateless nation, as
the region of Kosovo is adjacent to the state of Albania. They are, instead, a group that does not
comprise the core population of the Federal Republic of Yugoslavia, later Serbia. AIFs were
primarily involved in conflicts with Serbian and Macedonian forces. The state of Albania assisted
AIFs composed of their ethnic kin in Kosovo, adjoining parts of Serbia, and Macedonia. This
assistance was sometimes direct, involving state military forces but mostly indirectly through
training and equipping insurgents. As this case is nested within the Yugoslav Wars, other conflicts
in the broader occurrence of warfare will be considered as needed.
The DIG is useful to compare development, competition, conflict, and cooperation across
borders to DSNs, while isolating the effect of statelessness. This is in contrast to SSSs, in which
statelessness is shared by the SSS and the DSN, but division by an IRB varies. The AIFs are the
subject of Chapter Six.
Korean state regimes (KSRs)
The two Koreas are a surviving TSON case that represent an ethnic group divided into two states
by a border. The history of conflict between the two, including the Korean War, will be examined
in this work, but outbreaks of border violence and cross-border political interference after the end
of the Cold War will be the major focus for this case.
The TSON is useful for comparison to the DSN because both involve a single group
divided by borders. The difference lies in the multiple conditions of statelessness experienced by
the DSN factions, which stands in contrast to the double statehood of the TSON. The Korean dyad
is illustrated in Chapter Seven.

113

The remainder of this work proceeds as follows: Chapters Four through Seven contain the
case studies; Chapter Eight presents the overall findings of this work. Implications for the literature
are discussed in Chapter Nine, along with suggestions for future research and this work’s
conclusion.

114

Chapter Four
Divided Stateless Nations: Intraethnic Conflict and Cooperation among Rival Kurdish
Political Factions

This chapter illustrates why and how rival factions claiming to represent DSNs may engage in
conflict or cooperation with each other. It examines the experiences of Kurdish political factions
(KPFs) in two conflicts and one cooperative period. Chapter Four also presents evidence for why
these dynamics may occur.
Regional and system-level factors are argued to be the most salient with regards to the
onset of conflict between DSN factions. Test results for the effects of distance yield surprising
results; DSN rivals appear unaffected by proximity to the state center. Possible explanations are
offered, which lead to new propositions for future research. Regional centers and borders are
demonstrated to be important objects of contention between DSN factions. Interestingly, national
capitals are not seen as deterrents for KPF expansion. Rival factions within DSNs are not always
in conflict, however. The indiscriminate killing of unaligned co-ethnics in institutionally
competitive territory by out-group forces incentivizes rivals to cooperate in the short-term.
Cooperation is shown to be unlikely if out-group forces methodically eliminate rivals instead.
Chapter Four assesses the three propositions and their associated hypotheses as a basis of
comparison to SSSs, DIGs, and TSONs in the chapters to follow. The variation between the DSN
and the other types is demonstrated to be significant enough to warrant the classification of DSN
violence as a separate ethnically related form of conflict. The variation across types is holistically
presented in Chapter Eight.

115

This chapter consists of six sections. Section 4.1 is a summation of Kurdish history from
prehistory until 2020. It also covers the nature of their division by internationally recognized
borders; how they came into their long-standing position of statelessness; how they developed self-
determination claims in response to their relations with the states in which they live; and the origins
and development of the major KPFs. Section 4.2 then describes the periods of conflict and
cooperation that were selected for inclusion in this work.
Following those introductory sections, statistical and qualitative arguments in support of
the hypotheses for conflict onset in each period are evaluated in Section 4.3. Section 4.4 presents
quantitative and qualitative evidence for the mechanisms of conflict patterns, while Section 4.5
demonstrates conditions for predicting cooperative patterns between rival factions within DSNs,
as well as conditions for when such patterns may not materialize.
Finally, section 4.6 sums up the findings with regards to DSNs, beginning with the Roots
of Rivalry proposition (P1). The Competing Leaders hypothesis (H1a) receives a limited degree
of support, while the State Manipulation hypothesis (H1b) is not well-supported. Divergent
Development (H1C) is the best supported explanation for the onset of conflict among rival DSNs.
Tests for the Effects of Distance (P2) yield some surprising results, as well as some that were
expected. Interestingly, DSN factions appear to be unconcerned with proximity to the National
Capital (H2a). Potential explanations are offered as opportunities for future research. The Regional
Center (H2b) and Unpacific Borders (H2c) hypotheses are supported, while the results of the
Primary Route (H2d) hypothesis are inconclusive at this time but may be studied further. With
regards to the Out-group Incursion proposition (P3), the Who Dies and Where They Die hypothesis
(H3a) receives significant support from both statistical and qualitative evidence. The Heroes and

116

Traitors hypothesis (H3b) is also supported, while the Non-combatant Deaths hypothesis (H3c)
was unsupported.
4.1 Historical overview of the Kurdish people and Kurdish political factions
It is, at this point in time, axiomatic to recite that the Kurds are the largest nation in the world
without a state of their own. Technically, Tamils are a more populous stateless nation, but Kurds
are much more thoroughly divided by state borders. There are approximately 36.4 to 45.6 million
Kurds in the world, depending on how Kurdishness is defined. Excluding the diaspora there are
15 to 20 million in Turkey (mostly in the east and southeast), ten to twelve million in Iran
(predominantly along the western border with Iraq), eight to eight-and-a-half in Iraq (nearly all are
in the far north), and three to three-and-a-half million in Syria (mostly in the northeast and along
the northern border with Turkey). Figures 4.1 and 4.2 display historical distribution of Kurdish
populations since 1910. There are also historical Kurdish populations in Armenia, Georgia, and
Russia, though they are primarily urban populations that do not pursue territorial claims in those
states. There are also one-and-a-half to two million more in the global diaspora, primarily in
Europe and North America.
Kurds speak an Iranian language comprised of several dialects exhibiting varying degrees
of mutual intelligibility. Kurds are predominantly Sunni Muslim, though there are Kurdish
speakers of Êzîdî, Yarsani, Ismaili, Alevi, Twelver Shia, Christian, and Jewish religious
backgrounds. Not all members of these religious communities consider themselves Kurds, though
KPFs consider them to be Kurds.
161
Some members of those communities enthusiastically identify
as Kurds. Some may identify as Kurds due to economic inducement or political gain.
162

117

Regardless of religious identification, the groups of related peoples that are today known
as Kurds had their ethnogenesis in northern Mesopotamia and southern Anatolia. Archaeogenetic
studies identify Neolithic early European farmers and Bronze Age Indo-European pastoralists as
their ancestors.
163
To the Sumerian, Akkadian, Assyrian, and Babylonian civilizations some of
their forebears were known as the Guti, while Greek, Macedonian, and Seleucid empires knew
them as the Korduene. The Roman name for the ancestors of the Kurds was Carduchoi. These
tribal confederations were frequently at odds with the ruling empires of the day and were known
for their desire to maintain their independence.
164
The name “Kurd” likely refers to semi-nomadic
Iranian tribes generally, which is how the word came to be used by the Median and Achaemenid
Persian dynasties.
165

The Arab conquests of the Middle East during the seventh and eight centuries brought
Sunni Islam to the Kurds. Most had been Zoroastrian, Christian, or had practiced other syncretic
indigenous religions, some of which survive to the present day. For instance, some Zoroastrians
and Êzîdî—even some Kurds—believe that all Kurds were once Zoroastrian or Êzîdî and that
Muslim Kurds betrayed their people by accepting Islam.
166

Kurdish leaders played a pivotal role in the historical formation of the Middle East. One
prominent Kurdish leader, Salah ad-Din Yusuf ibn Ayyub, defeated a Crusader army at the Battle
of Hattin; retook Jerusalem from them; conquered Egypt; established the Ayubbid Sultanate; and
began the process of replacing the Hanafi school of Sunni Islamic jurisprudence with the Shafi’i
school among his people. The Shafi’i rite is the dominant one among Sunni Kurds today, which
sets them apart from their Arab and Turkish Hanafi neighbors on an additional dimension.
167

The political fortunes of the Ayubbid Sultanate waned with the rise of the Khwarezmids
and Mongols east of the Sultanate. The simultaneous rise of the Mamluks in Egypt, as well as the

118

commencement of additional Crusades, further degraded Ayubbid power. The Sultanate itself was
defeated by the forces of Hulagu Khan in 1260, and the remnants of the Ayubbid dynasty were
taken by the Egyptian Mamluks, the Safavid Persian dynasty, and various Turkic groups in
Anatolia.
168
Most of the greater Kurdistan region and its population was annexed into the Ottoman
Empire in 1514, with some in the east remaining under Safavid control.
169

At this time, many Kurdish tribes remained of mixed Islamic and Êzîdî religion.
170
The
Ottoman and Safavid rulers frequently went to war with the Êzîdî, often recruiting Muslim Kurds
to their cause by promising autonomy and Êzîdî lands, which, over time, resulted in the Êzîdî
population being most heavily concentrated in Laliş, some valleys near Sheikhan, and Mount
Şingal. The memories of the many genocides perpetrated from the Muslim conquests, through the
Ottomans, and up to the ISG invasion, are recounted as the 74 ferman, or official orders of mass
murder. Muslim Kurds were themselves also repressed by the Ottoman Empire. Both Muslim
Kurds and Êzîdî have oral histories that include a type of song, a stran, which recounts those lost
in out-group persecution or in-group betrayal.
171

Kurdish autonomy within the Ottoman Empire was sometimes withdrawn by the sovereign.
On those occasions their territory was reduced, populations of Kurds were transferred, and
administrative provinces that were predominantly Kurdish were abolished. A short-lived Kurdish
principality was established following a Kurdish revolt in the aftermath of the Ottoman defeat in
the Russo-Turkish War of 1877-1878. Kurds then revolted against both the Ottoman and Qajar
empires; those revolts were violently put down.
172

Interestingly, the Ottomans supported the same Kurdish tribes it had previously repressed
when those tribes went to war with the Qajar Persian Empire in 1880.
173
The sponsoring of Kurds
against rival states, even if those Kurds were once enemies of the sponsoring state, would

119

foreshadow a recurring dynamic between the future states of Turkey and Iran. Though each state
would depend on Kurds to serve as proxies against the other state, they would cooperate to repress
Kurds if their factions threatened to become too strong.
In the final years of the Ottoman Empire, relations between Kurds and Armenians soured.
Some Kurds were recruited to attack Armenian, Assyrian, and even other Kurdish populations to
seize and occupy the land those groups formerly inhabited.
174
This is significant because Kurds
that cooperate with non-Kurdish forces to fight other Kurds have long been known by the derisive
term caş, or donkey foal.
175
This slur is employed in Kurdish politics to the present day as an
epithet for someone who is viewed as undermining the greater cause of all Kurds. The revulsion
toward caş features prominently in how the KPFs frame conflict and cooperation.
The Ottoman Empire was defeated in the First World War, which eventually resulted in
the 1920 Treaty of Sèvres. This treaty divided the Empire’s territories among the Allied forces
from the First World War. That treaty was renegotiated following the victory of the Turkish
nationalist movement in the Turkish War of Independence. The 1923 Treaty of Lausanne defined
the modern borders of Turkey, Syria, and Iraq, as well as cemented the preexisting border with
Iran. By 1946, and continuing to 2011, Kurdish-inhabited territory had changed somewhat, as
shown in figures 4.1 and 4.2:

120

Figure 4.1
Distribution of Kurdish populations before and after the Treaty of Lausanne (1923)

Note. Modern borders are used for the pre-Lausanne map to assist in visual identification of
change.

121

Figure 4.2
Temporal analysis: Change in distribution of Kurdish populations before and after the Treaty of
Lausanne (1923)

Note. Colors are displayed with a transparency setting to display bodies of water.

The left-hand map in figure 4.1 is the digitized extent of Kurdish-inhabited territory in 1910. On
the right, we see the modified maximum extent of Kurdish population growth from 1946-2017.
The territory no longer occupied by Kurds, the new territory inhabited by Kurds, and the area that
did not change are the focus of figure 4.2. A visual inspection suggests that more land was occupied
than depopulated, especially in Syria and Iraq. Quantifying the approximate areal change in a

122

temporal analysis demonstrates that, despite losses in territory, Kurds occupied about 58,778 more
square kilometers than they did in 1910,
176
or an additional 22% more land than in 1910, as shown
in table 4.1:
Table 4.1
Temporal analysis: Change in Kurdish-inhabited territory, 1910-2011
Year Total area Unchanged area Area lost Area gained Net gain
km
2
km
2
km
2
km
2
km
2

(%) (%) (%) (%)

1910-1918 268,091 NA NA NA NA

1946-2011 326,868 218,280 49,810 108,588 58,778
(81) (19) (41) (22)

Note. Km
2
have been rounded to the nearest whole number.

Division and self-determination movements
The Kurdish population was split mainly across Turkey, Syria, Iraq, and Iran following the
Treaties of Sèvres and Lausanne.
177
Kurdish nationalists refer to the four Kurdistans by their
cardinal directions. Kurdish-inhabited portions of Turkey are called Bakurê Kurdistanê, or
Northern Kurdistan. The Kurdish-populated region of Iran is known as Rojhilatê Kurdistanê,
which corresponds to Eastern Kurdistan. Iraqi Kurdistan, or Başûrê Kurdistanê, is Southern
Kurdistan. Finally, Rojavayê Kurdistanê is Western Kurdistan in Syria. These are known by the
shorthand expressions of Bakur, Rojhilat, Başûr, and Rojava, respectively, which is how they are
generally referred to in factional media.

123

Some of the populational-territorial changes in figure 4.2 were due to violent conflict. The
result of one such conflict is illustrated by looking at the area around Lake Van in eastern Turkey.
While it was mostly Armenian in 1910, Kurds had assumed control following the depopulation of
Anatolian Armenia. Another example is the movement of Kurds into the northeastern Jazira
province of Syria. This was the new state’s response to an uprising against Kemalist Turkey in the
1920s.
178
Note the apparent depopulation of what is now southeast Turkey in figures 4.1 and 4.2;
this was the source population of Turkish-enforced migration into Syria.
These Kurds would retain their ties to the Kurds of the states they came from. Expelled
Kurds simultaneously adjusted to new political conditions in Syria and Turkey with varying
degrees of success. Emergent KPFs developed differing types of self-determination claims and
competing visions for what an idealized Kurdish political entity should look like over time. The
Kurdish experiences in the two older political entities,
179
Turkey and Iran, will be examined first.
Then the Kurds of the two newer states, Syria and Iraq, will be looked at.
Under a policy of social, cultural, and territorial Turkification, many Kurds in Bakur
relocated to urban areas
180
amidst the conflicts between the Turkish state and their Kurdish allies
on side, and Kurdish rebel groups on the other during the latter half of the 20
th
century. The Kurds
who were allied with the Turks were reviled as caş by those aligned with the PKK. That faction is
a transnational Marxist ethnonationalist separatist movement that was founded in 1978 by
Abdullah Öcalan.
181
Despite numerous ceasefires and a three-year peace attempt, the conflict
between the PKK and the Turkish state continues to the present day and has spilled over into Iraq
and Syria. The PKK and its armed forces, the HPG, have also engaged in armed conflict with non-
Marxist Kurdish political factions in Turkey, Iraq, and Syria.

124

The PKK helped found the KCK, an umbrella organization of Marxist Kurdish
ethnonationalist movements. The immediate forerunner to the KCK was known by the unfortunate
initialism of KKK. The KCK’s goal is to convince Kurds across Greater Kurdistan to support a
Marxist-influenced democratic confederalist system of governance.
182
One of its most
consequential allied factions has been the PYD in Syria. There are many other members of the
KCK, such as the PJAK in Iran, and the PÇDK in Iraq, but they do not feature prominently in
intra-Kurdish conflict. Elements of the KCK have been militarily supported by the USSR, the US,
and Iran at various times.
Like the Kurds in Turkey, the Kurds in Iran have met with little successful at achieving
autonomy, though there have been and are several Kurdish separatist movements in Iran. An early
separatist attempt, the Soviet-backed Republic of Mahabad, lasted only for part of 1946. This entity
lacked support from other Kurdish tribes who resented the influential Barzanî family that
established it. Iranian forces retook the Republic’s territory when the Soviets withdrew from
Iran.
183

One active Kurdish separatist movement in Iran is the PJAK, which is a KCK-affiliate that
lacks significant external support.
184
Another movement is Komala, a Maoist Kurdish organization
that is not part of the KCK. Komala was once backed by the Soviets and has recently established
ties with the US.
185
A third major KPF in Iran is the PDKI, which is a leftist organization allied
with the PKK. However, the PDKI pursues independence for Rojhilat, not simply autonomy or
confederalism, which is at odds with the KCK.
186

KPFs in Iraq were frequently at odds with the central government but have had
significantly more success in pursuing goals of autonomy than factions in Turkey and Iran.
Mustafa Barzanî created the KDP in 1946 and sought to achieve autonomy for Kurds in Başûr.
187

125

This goal was at first elusive, and conflict broke out between the KDP and the state of Iraq in 1961.
Iraqi Kurds were granted autonomy in the 1970 Peace Accord, but this was revoked in 1975 when
Iraq and Iran reached an agreement to withdraw support for Kurdish parties on either side of the
border, including Iran’s support of the KDP. This led to a forceful reassertion of control over Iraqi
Kurdistan by Baghdad.
The KDP is primarily rural, tribal, and conservative with regards to their political
worldview. This led to a split with urban intellectuals within the party who wanted to move toward
a center-left political orientation. Jalal Talabani created a new party, the PUK, in 1975.
188
The
PUK’s influence came to be greatest in the east and south of Başûr, while the KDP remained
stronger in the north and west.
The two parties did not cooperate much early on, though there were some exceptions. They
began on opposing sides during part of the Iran-Iraq War. However, by 1984, they cooperated with
each other and with Iranian forces against Iraqi forces when the Hussein regime reneged on
promises of autonomy.
189
During the Iraqi state’s Arabization campaigns of the 1970s and 1980s,
there was little in the way of concerted efforts to unite the two factions until the al-Anfal campaign
of 1986-1989. This particularly brutal incursion by Iraqi forces featured the use of chemical
weapons in numerous Kurdish-populated territories, most notably at Halabja.
190
Non-combatant
Kurds were specifically targeted by state forces, unlike previous conflicts in which pro-state
Kurdish militias were the preferred tool to combat Kurdish guerillas.
The Iraqi government again invaded Başûr following an uprising by Kurds at the end of
the Gulf War in 1991, which led to the imposition of a no-fly zone over what became the KRI by
American, British, and French forces. The KRG was established, with the KDP and the PUK
becoming its two most powerful parties. Disagreements over governance led to guerilla clashes

126

between the two factions, which escalated to warfare in 1994.
191
This conflict is described in detail
in the following section. Two KRGs were created—one for each party—upon the conclusion of
the conflict. That arrangement held until 2003, when both factions joined in the overthrow of the
Hussein regime and once more united under one KRG, though retaining their areas of influence.
They also maintain separate armed forces, though both are known as the Peşmerge (“those who
face death”).
Following the overthrow of the Hussein regime, the KRG began to sell oil independently
from Baghdad. In particular, the KDP has emerged as the more powerful faction within the KRG
and has nurtured an economic relationship with Turkey based on the sale of Kurdish oil. The KDP
developed close political and security ties with Turkey as well. Culturally and socially, KDP-
friendly Kurds in Turkey are relatively more supportive of the Turkish state than Kurds affiliated
with other parties in Turkey.
192
The KDP has also been involved in conflicts with the PKK, as well
as with its offshoots and affiliates in the KCK. The KDP has even created an umbrella organization
of its own, the KNC, to compete with the KCK for influence amongst Kurdish populations in
Syria.
193
The most notable KPF within the KNC is the ENKS (sometimes also called the KNC or
KNC-S) in Syria.
The KDP and PUK have both expanded past the established boundaries of the KRI,
especially when Iraqi forces withdrew from Kurdish-inhabited regions in response to the ISG
invasion in 2014. The KDP took much of northern Nineveh Governorate, particularly the area
around Şingal, while the PUK expanded around Kirkuk. These gains were reversed following Iraqi
reassertion of control over these disputed territories in the wake of the 2017 Kurdistan Region
independence referendum, in which 93% of the votes cast were in favor of independence for
Başûr.
194

127

The other state in which Kurdish autonomy has been established and maintained for a
significant length of time is the Syrian Arab Republic (SAR). The SAR under Hafez al-Assad
frequently repressed Kurds. In particular, Kurds were stripped of their citizenship over suspicion
that much of the Kurdish population were infiltrators sent by the Turkish state instead of having
been forcefully expelled.
195
This designation also applied to Kurds that lived in Syria long before
the partition of the Ottoman Empire. At times, however, the regime courted Kurds by offering
positions of authority to certain Kurdish leaders.
The Syrian state has taken a different position with regards to Kurds in recent years. With
the revolts that accompanied the Arab Spring in Syria, Bashar al-Assad offered most Kurds
citizenship to appease a restive population, thereby reducing the number of fronts he would have
to fight in the Syrian Civil War.
196

The PYD in Syria was created in 2003 by Kurdish democratic confederal socialists in much
the same vein as the PKK. Indeed, the organizations maintain linkages. They share the goal of
autonomy rather than independence, and they are both members of the KCK. The PYD established
its own armed forces, the YPG, in 2011.
197
Individual members are known as Şervan (“warrior”).
The PYD and their armed forces took advantage of the weakness of the SAR during the
civil war and established control over much of north and northeast Syria between 2011 and 2014.
While setting up a canton system the PYD had to contend with another Syrian Kurdish group, the
ENKS, and their paramilitary wing, the Rojava Peşmerge. The ENKS is an offshoot of the KDP
and part of the KNC umbrella. Initially, the ENKS was more popular amongst Syrian Kurds, but
the PYD became the most powerful KPF in Syria after the YPG’s successful defense of Kobanî in
2014.
198

128

Conflict between KNC- and KCK-affiliated groups occurred in Syria and Iraq before and
after the incursion by the ISG into Kurdish-inhabited regions of both states, though they cooperated
to drive the ISG from afflicted areas.
199
The PYD and PKK occupied the area north and west of
Şingal in tandem with the Yazidi Freedom and Democracy Party and their armed forces, the
YBŞ/YJÊ

.
200
Meanwhile, the Rojava Peşmerge and the KDP supported a rival Êzîdî force, the
Êzîdî Peşmerge led by Qasim Şeşo.
201
These groups occupied the area south and east of Şingal.
The KDP has at times co-opted the nominally independent HPÊ.
202
After the ISG receded, some
of these rival forces clashed with each other and a de facto border was established, splitting the
Şingal area in two, as each faction competed over the local Kurdish-speaking population and their
territory.
203
This period of conflict and cooperation is described in the next section.
The PYD has dominated Syrian Kurdish politics in the DFNS and the AA as the most
influential member of the ruling TEV-DEM coalition. However, their success has drawn the
attention of the Turkish state, which has intervened in the Syrian Civil War numerous times,
primarily trying to halt the expansion of the PYD. Operation Euphrates Shield, Operation Olive
Branch, and Operation Peace Spring have resulted in Kurdish displacement and have exacerbated
intra-Kurdish tensions but have not induced cooperation between rivals against out-groups.
204
The
Turkish state has also intervened in northern Iraq, attacking the PKK, the PYD, and their local
allies.
205
This period of intervention is detailed in the subsequent section.
4.2 A depiction of conflicts among rival KPFs
As the work turns to an account of intra-Kurdish conflict and cooperation since the end of
the Cold War, let us first examine an aspect of Kurdish competitive dynamics. Hûnermendê, or
folk musicians, such as Şivan Perwer and İbrahim Tatlıses from the photographs below in figure
4.3, to name just two, are closely aligned with various KPFs.

129

Figure 4.3
Kurdish hûnermendê and factional support

Note. At left, hûnermendê meeting with political leaders. From left to right: İbrahim Tatlıses
(hûnermend); Recep Tayyip Erdoğan, President of Turkey; Mesûd Barzanî, leader of the
KDP; and Şivan Perwer (hûnermend). The men met allegedly to discuss combating KCK-
aligned groups in late 2013 (Çetin, 2013). At right, Şivan Perwer meets with a Brigadier
General in the Zêrevanî (gendarmerie) during the battles to retake Şingal (Miliyet, 2014).

The cultural and political worlds are intertwined in this instance. Performers affiliated with various
KPFs echo the language of politics and they laud the same individuals—the armed forces of the
faction. Friendly fighters are called şêr, which means “lion”. The supporters of rival factions are
instead reviled as caş, which is the “donkey foal” and has been a derogatory term for one who
collaborates with outsiders against the Kurdish nation. These terms are employed regularly in both
music and in politics, whether officially or unofficially.
206
Factional sources sometimes use them
to describe rivals that want to appear to be cooperative for cynical reasons, or for rivals that refuse
to cooperate or withdraw support. This dynamic is significant because this work utilizes
contentious factional media sources to support or refute the hypotheses that pertain to conflict and

130

cooperation. Additionally, interviews would sometimes refer to a stran, which is a type of folk
song of remembrance or mourning of particular şêr or revilement of caş.
Periods of intra-Kurdish conflict and cooperation
There are two periods of conflict between KPFs that were selected for study. The first is
the 1994-1997 war, in which the KDP fought against the PUK and the PKK. The second period is
the 2011-2020 conflict between KNC- and KCK-affiliated groups. Both periods were interrupted
by incursions by out-groups.
There were notable periods that were not selected, though they also featured an invasion
by an out-group. The Iraqi state invaded the KRI from 1986 to 1989 as a means of punishing the
KDP and PUK for assisting Iran during the Iran-Iraq War. The al-Anfal (“spoils”) campaign was
prosecuted mainly against civilians. Chemical warfare and ethnic cleansing were features of this
period of incursion. Limited qualitative evidence suggests there was KDP-PUK cooperation,
207

but the degree of spatial certainty for known cooperative events is quite poor. Existing
georeferenced fatality counts contain levels of confidence that are too low to be incorporated,
208

while manual verification would be prohibitively labor-intensive. This is also why several minor
intra-Kurdish conflicts involving Islamist Kurdish groups in the 2000s, such as the IMK and Ansar
al-Islam against the PUK,
209
were excluded. There simply was not enough data to incorporate into
a systematic analysis.
The second excluded period was the 2017 conflict between the Iraqi state and the KRG
over the latter’s independence referendum. The KRI was invaded, but with very few instances of
violence. This conflict was excluded because total deaths and the number of conflict events were
very low. This is true in absolute terms, and also in comparison to the two selected conflicts.

131

1994-1997 Iraqi Kurdish Civil War
Let us examine the two periods that were selected, beginning with the 1994-1997 Iraqi
Kurdish Civil War. Recall that the initial creation of the KRI took place in the aftermath of the
Hussein regime’s defeat in war by a US-led coalition, which led to an ad hoc partition of Iraq.
210

Factional disagreements over governance of the KRG led to guerilla clashes in 1994. This
escalated into open warfare between the KDP and PUK.
211
The PUK took Hewlȇr, the capital of
the KRI, from the KDP in 1995. Assisting the PUK was the PKK. PKK forces entered Iraq from
surrounding states. They also had a presence in the Qandil mountains and in the area around
Duhok. All factions used the territory of their regional allies to attack their rivals in otherwise
inaccessible territory.
This period of intra-Kurdish conflict experienced out-group invasion. Turkey entered the
KRI to combat PKK forces alongside the KDP, while Iran assisted the PUK in taking territory
from the KDP. Iraqi government forces directly intervened to aid the KDP in retaking Hewlêr from
the PUK. Iraqi forces then briefly entered PUK territory, where they were repulsed by PUK forces
with Iranian assistance.
Despite these invasions by out-groups, the KDP-PUK and the KDP-PKK dyads did not
halt their conflict to eject outsiders. Attempts were made by the US to broker cease-fires, but they
were not lasting until two separate KRGs—one for each faction—were formed. This ended the
war between the KDP and PUK, but the KDP and PKK continued engaging in low intensity
conflict even after the creation of two KRGs.
An analytical map of the conflict is presented below in figure 4.4; a map of out-group
invasions may be found in the section on rival cooperation.

132

Figure 4.4
Extent of the 1994-1997 intra-Kurdish conflict

Note. Some conflict events are too near each other to be displayed individually.

2011-2020 KNC-KCK Conflict
Long-standing KNC-KCK animosity found new grounds for competition in Syria at the
outset of that state’s civil war in 2011. The PYD—a KCK-aligned offshoot in Syria—was formed
in 2003 and established its armed wing, the People’s Protection Units (YPG) in 2011. The PYD
established control over several Kurdish-inhabited areas of northern and northeast Syria when the
Syrian Civil War began.
212

133

The KDP and their Syrian offshoot, the ENKS, competed with the PYD over unaligned
Kurdish-speaking populations in recently claimed territory in Iraq and Syria as state forces
retreated from militant Salafist threats. The factions arrested and killed each other’s sympathizers,
engaged in armed clashes, destroyed rival parties’ headquarters, and demanded that neutral
Kurdish-speaking groups choose between them. The KDP and ENKS were initially more
successful in this conflict in both states. Though they were more popular at first, they lost ground
to the KCK-aligned forces.
213
The latter were viewed as more interested in protecting Kurdish
people from predation by Salafists.
The conflict in Iraq was primarily located around Şingal. The majority of clashes between
KNC- and KCK-affiliated forces were in and around the city, especially between the towns of
Xanasor and Sinunê. Most of the conflictual events in Syria were in key Kurdish-inhabited cities,
without so many instances of violence in rural areas. Acts of arson, enforced disappearances of
rivals, and killings at checkpoints have also been features of this period of conflict.
The analytical map for this period of conflict is displayed in figure 4.5:

134

Figure 4.5
Extent of the 2011-2020 period of intra-Kurdish conflict

Note. Some conflict events are too near each other to be displayed individually.

4.3 The roots of DSN rivalry: Evaluating existing explanations
In this section I engage the Roots of Rivalry proposition (P1). Recall that this proposition says that
DSNs should share a root cause of conflict with other forms of organized intraethnic conflict.
There are also three hypotheses associated with this proposition: Competing Leaders (H1a), a first
image explanation; State Manipulation (H1b), which is a second image explanation; and Divergent

135

Development driven by system-level factors (H1c), a third image explanation. These are the factors
that should lead to factional conflict.
Competing Leaders (H1a)
The history of the KPFs and the results of the temporal analysis suggest that there may be
evidence for each of the hypotheses, though in varying degrees. Let us begin with the Competing
Leaders hypothesis (H1a), which concerns the impact of leadership disputes. The PUK split off
from the KDP over political disagreements and the leaders of each were known to dislike each
other. Even prior to the split, the Barzanî and Talabanî wings within the KDP sporadically used
violence against each other.
214
We may confidently assume that violence most likely caused the
party to split, but, without the involvement of exogenous actors they did not escalate into warfare.
Indeed, violence decreased after the split.
Additionally, the KDP and PKK conflict was not predicated upon individual leaders’
mutual antipathy. Barzani and Öcalan do not appear to dislike each other personally; rather, it is
their factions’ philosophies based on divergent development and ties to non-Kurdish regional
actors that drive the rivalry. The relationship between Barzanî and Öcalan appears to be relatively
cordial. Indeed, they have met and discussed cooperation in the past— Barzanî has given Öcalan’s
faction land in Iraq.
215
Such efforts for compromise seem to become more difficult as time goes
on and development hardens intra-Kurdish divisions. This extends to the 2011-2020 conflict
between the KNC and the KCK. Individual hostility is not a determining factor in this competition
between umbrella groups. While this hypothesis appears supported on the surface, it may not
ultimately explain much without considering other factors.

136

State Manipulation (H1b)
Let us examine the utility of State Manipulation (H1b), the ability of a state to maneuver
ethnonationalist groups into fighting each other. As the first period of conflict took place in Iraq,
a major actor to consider is the Ba’athist regime. The regime had a history of conflict with both
the KDP and PUK. It lost one war with the KDP in the 1960s but won a second war in the 1970s
when Iran withdrew its support of the KDP.
216
The central government unsuccessfully fought a
PUK insurgency in the 1970s. Autonomy agreements between the Ba’ath regimes and Kurdish
groups did not last long.
217
When those broke down and conflict was initiated, both KPFs usually
sided with Iran against Iraq.
218

The Iraqi regime fought alongside the KDP in part of the 1994-1997 conflict, which might
suggest some evidence for this hypothesis. However, Iraqi assistance was heavily limited at the
insistence of the KDP, and Barzanî extracted a pledge from Hussein to withdraw from KDP
territory after Hewlêr was retaken.
219
Furthermore, when the Iraqi forces fought the PUK east of
the KDP-controlled territory, KDP forces did not take part and the Iraqi forces were repelled.
220
It
is apparent that the KDP used the Iraqi state as a means to an end, rather than the other way around.
This is especially important when considering the conflict was internal to the KRI. It began
over political issues, not at the behest of the Iraqi state. Iraq was able to send troops, but they were
supported by limited armor, indirect fire, and artillery assets. Air support and heavy armor support
was not feasible because of coalition threats.
221
The PUK took advantage of this externally
imposed symmetry and proved capable of defeating Iraqi forces. Cease-fires failed because of the
difficulty of resolving this internal dispute, rather than Iraqi manipulation.
222
Thus, State
Manipulation (H1b) is the least likely to be at work.

137

Divergent Development (H1c)
The third hypothesis is Divergent Development (H1c). This posits that system-level factors
are the most likely to lead to factional conflict. Recall once more the temporal analysis; Kurds
expanded their territory in Iraq amidst the weakness of the Hashemite Kingdom following the
imposition of IRBs. Turkey and Iran, especially, but also Syria initially created stronger states than
Iraq and they had radically different international orientations. Turkey was a NATO member.
Pahlavi Iran was backed by the West, while Iran was supported by the Soviet Union before and
after the reign of the Pahlavi dynasty. Syria was also backed by the USSR. Kurdish populations in
these stronger states that became more urbanized also became more sympathetic to left wing
insurgents.
223
Those that remained rural, mostly in parts of Turkey, Iraqi Kurdistan, and Syrian
Kurdistan, were more conservative.
224
Acts of displacement disrupted existing populations but
created transnational linkages that persist to this day.
KPFs developed different political-military, economic, social, and cultural realities based
on their division by borders. This altered the information costs among coethnics. Formerly known
entities frequently changed their position based on relations with host states and, more importantly,
their international backers. Recall that the Republic of Mahabad lived and died by Soviet
patronage. The KRI was created by a UN Resolution and was sustained by an international
coalition more deeply involved in the peace process than Iraq. A similar dynamic sustains the AA
in Syria today.
Social and cultural attributes were influenced by displacement and urbanization. Economic
alliances could be convoluted; the KDP became a Turkish and/or Iranian ally depending on the
time period in question; the PUK became Iranian allies; and the PKK and their allies have been
supported by external states in their insurgencies. Indeed, rivals in both periods have accused their

138

foes of being pawns of non-Kurdish regional actors (sources), thus betraying the Kurdish
identity.
225
Intra-Kurdish conflict ultimately derives from this system-level influence. The Iranian
state and international coalitions have been more successful in getting Iraqi Kurds to act than the
Iraqi state ever was.
Neither Syria nor Iraq convinced these groups to engage in conflict within a single state,
nor are the leaders of KPFs always personally opposed. Rather, these umbrella factions have a
mutually irreconcilable vision of what a united Kurdistan should look like. This vision is
influenced by the divergent development of divided societies. Transnational relations, like the
KNC’s close relationship with Turkey and the KCK’s commitment to democratic confederal
movements in the northern Middle East, are apparently the most powerful explanations for the
roots of KPF rivalry.
This is reflected in the press releases of rival factions engaging in conflict against each
other. In the first period, the PUK and PKK accused the KDP of doing the bidding of Turkey and
the Ba’athists.
226
In response, the KDP blamed the PUK and PKK for being pawns of Iran.
227
In
the second period of conflict, the KCK once more branded the ENKS/KNC as Turkish
underlings.
228
The KNC, for its part, viewed the KCK as an extension of foreign powers that do
not want an independent Kurdistan.
229

Additionally, the KNC-KCK conflict demonstrates how 21
st
century KPFs have become
powerful transnational players in their own right. Consider how a KCK-affiliated group, the PYD,
was able to take and hold territory and sway populations outside of their “home” state. Rival
factional media may accord this development to foreign interference alone, but KCK-aligned
groups instead reference the necessity for unifying all Kurds, regardless of which state they
inhabit.
230

139

Thus, Divergent Development (H1c) is the best supported hypothesis under the Roots of
Rivalry (P1). Analytical eclecticism, with the use of mixed methods, demonstrates the utility of
this intellectual complement when considering a wide set of causal explanations. Without the use
of these seemingly unrelated literary subfields and methodologies, any single explanation would
be incomplete, to the detriment of our theoretical understanding, and also for potential policy
implications.
These results are further interpreted in the final section of this chapter. Chapter Eight will
later compare the DSN to the roots of other varieties of intraethnic conflict discussed in Chapters
Five through Seven. Let us turn now to the study of how conflict among KPFs is carried out.
4.4 How do DSNs fight? A spatial analytic approach to DSN conflict patterns
This section examines conflict patterns. Namely, it looks at what KPFs fight over. It applies and
interprets statistical inference, spatial analytics, and qualitative research. As explained in the
research design, there was a sequence of tests applied to this case to ascertain what was the most
important object of conflict. Recall that the second proposition, the Effects of Distance (P2), asserts
that there should be distance- and capacity-based constraints placed on irregular forces that limit
what can be fought over under the New Wars lens.
This proposition is broken down into the importance of the national capital (H2a), regional
centers (H2b), and the border (H2c), which serve as the EVs of this section. As distance from these
points of interest (POI) increases, conflict events (the DV) should decrease. Distance from primary
routes is also considered as a means of accounting for potential omitted variable bias (OVB).
Including this measure compromises between using Euclidean distance or weighting distance
based on more extensive road networks that were unknown at the time of the conflicts, as highway

140

usage might be important for factions attempting to control territory and populations. Effects were
considered with two models: one for capitals and regional centers, and one for borders. Both
models included primary routes to account for the impact of major highways.
Preliminary analyses
The tests began with exploratory data analysis (EDA). This first step consisted of summary
statistics to look at whether interesting relationships might encourage further exploration of the
data. The results of Pearson’s correlation statistics suggested that there was a strong and significant
negative relationship between distance from regional centers on the likelihood of conflict events
for both periods. However, only the first period suggested a strong significant, positive effect for
the effect of the national capital.
The first period saw a negative, moderately significant relationship between primary routes
and conflict events, while the second period suggested a positive, strongly significant effect. Both
periods suggest a strong, negative relationship in the border model. That relationship was
statistically significant for both periods. In this model, both witnessed strong positive relationships
involving primary routes, but strongly significant for the first period and weakly significant for the
second period. The results of this test can be found in Appendix E, specifically figures E.2a and
E.3b.
The results of the correlation test were interesting enough to move on to OLS regression
analyses, which were similar to the tests for correlation. Support for the National Capital
hypothesis (H2a) only came from the first period, while the Regional Center hypothesis (H2b) was
supported by both models. The Unpacific Borders hypothesis (H2c) was better supported by the
first period than the second. These models were refined with stepwise selection and logarithmic

141

transformation. The resultant effect sizes were certainly weaker, though all three hypotheses were
nevertheless supported with statistically significant results in both periods. The results of these
tests are found in Appendix E, tables E.1a and E.2b, and figures E.3a and E.3b.
OLS models may not be ideal for spatial analysis, but these results were intriguing and
suggested that the data should be tested for spatial autocorrelation and spatial clustering. The
presence of which would indicate that at least second-order effects of spatial dependence were at
work, and possibly first-order effects. The global Moran’s I test indicated strong autocorrelation
with statistically significant pseudo p-values in both periods. Refining this test with EBS resulted
in moderate-to-strong autocorrelation, with slightly less (though still strongly) significant p-values
for both periods. The results of these tests are found in Appendix E, in table E.2 and figure E.4.
Likewise, the Getis-Ord General G statistic revealed that clustering in the first period was
less than one percent likely to be the result of chance, while the odds were under five percent in
the second period. This supports the inference that second-order effects were at work; the data
were not spatially independent, and so more precise methods than OLS were required to
confidently provide evidence in support of the hypotheses. As the clustering was so strong, there
was no need to perform the Gi* test. The results of this test are found in Appendix E, table E.5.
Spatial lag models
The data were autocorrelated and clustered, so a spatially lagged regression model that
incorporated a neighborhood spatial weights matrix was called for. The results of this test are
displayed below in table 4.2:

142

Table 4.2
Spatial analysis and modeling: Spatially lagged regression models of intra-Kurdish conflict
Spatial Lag Model:
1994-1997 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model
Model

Distance from National Capital (km)

-0.075
***
(0.023)

Distance from Primary Route (km)

0.105
(0.096)
0.120
***
(0.038)
Distance from Regional Center (km)

-0.169
***
(0.026)

Distance from Border (km)

-0.091
***
(0.033)
Constant

24.067
***
(7.391)
1.743
(1.500)
Observations 62 45
Log Likelihood -176.193 -115.333
sigma
2
13.054 9.468
Akaike Inf. Crit. 364.385 240.667
Wald Test (df = 1) 6,186.373
***
15.868
***

LR Test (df = 1) 105.933*** 14.279***
Note: *p<0.1; **p<0.05; ***p<0.01

143

Regional Centers Model Borders Model
Rho 0.9578
LR test value 105.93
p-value < 2.22e-16
Asymptotic standard 0.0122
error
z-value 78.654
p-value < 2.22e-16
Wald statistic 6186.4
p-value < 2.22e-16
Log likelihood -176.1926
for lag model
ML residual variance 13.054
(sigma squared)
(sigma 3.613)
AIC 364.3
AIC for lm 468.32
LM test for residual 43.326
autocorrelation
test value
p-value 4.6329e-11
Rho 0.5081
LR test value 14.279
p-value 0.0002
Asymptotic standard 0.1276
error
z-value 3.9834
p-value 6.793e-05
Wald statistic 15.868
p-value 6.793e-05
Log likelihood -115.3334
for lag model
ML residual variance 9.4683
(sigma squared)
(sigma 3.0771)
AIC 240.67
AIC for lm 252.95
LM test for residual 0.3463
autocorrelation
test value
p-value 0.5562

144

Spatial Lag Model:
2011-2020 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

-0.019
***
(0.001)

Distance from Primary Route (km)

0.025
***
(0.005)
0.023
(0.015)
Distance from Regional Center (km)

-0.077
***
(0.001)

Distance from Border (km)

-0.513
*
(0.302)
Constant

11.659
***
(0.800)
0.829
(0.751)
Observations 26 15
Log Likelihood 9.645 -14.749
sigma
2
0.024 0.207
Akaike Inf. Crit. -7.291 39.498
Wald Test (df = 1) 1,968.988
***
2,798.539
***

LR Test (df = 1) 107.159
***
35.578
***

Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

145

Regional Centers Model Borders Model
Rho 0.7224
LR test value 107.16
p-value < 2.22e-16
Asymptotic standard
error 0.0163
z-value 44.373
p-value < 2.22e-16
Wald statistic 1969
p-value < 2.22e-16
Log likelihood
for lag model 9.6453
ML residual variance
(sigma squared) 0.024
(sigma 0.155)
AIC -7.2906
AIC for lm 97.869
LM test for residual 2.4909
autocorrelation
test value
p-value 0.1145
Rho 0.95565
LR test value 35.578
p-value 2.4505e-09
Asymptotic standard
error 0.0181
z-value 52.901
p-value < 2.22e-16
Wald statistic 2798.5
p-value < 2.22e-16
Log likelihood
for lag model -14.7488
ML residual variance
(sigma squared) 0.2071
(sigma 0.4551)
AIC 39.498
AIC for lm 73.076
LM test for residual 10.392
autocorrelation
test value
p-value 0.0013

146

Surprisingly, the National Capital hypothesis (H2a), appears to be refuted. We would logically
expect to see a positive effect conflict. Conflict should increase as distance from the capital
increases, but it appears that there is a weak, though significant, negative effect. It is possible that
powerful DSN factions are less concerned about the ability of their host states to intervene in such
conflicts; this is briefly addressed in the summative section at the end of this chapter. It is also
explored in depth in Chapter Nine.
The Regional Center hypothesis (H2b) is most strongly supported; there is a significant
negative effect. This is to say, as distance from the regional center increases, conflict events
decrease. There is a similar result for the Unpacific Borders hypothesis (H2c). It is negative in
each case. It is much stronger in effect size for the second period, but also much less statistically
significant.
The impact of Primary Routes (H2d) appears to be inconsistent. The results are significant
only in the borders model in the first period, but significant only in the regional centers model in
the second. The role of checkpoints in the second period, which was not as meaningful in the first,
may account for this variation. Likewise, light armored vehicles and technicals were likely to be
more important in the first period of conflict as greater distances from the border were most likely
involved as it involved a larger geographic area.
In all spatially lagged models, the AIC was much lower than the OLS and log-transformed
models, indicating a better fit. Additional robustness checks were applied, including a studentized
Breusch-Pagan test. When rendered as spatially independent and homoscedastic, the spatial lag
models are shown to retain improved fits over the OLS and log-transformed models. These are
found in Appendix E, table E.3.

147

Qualitative support
In all, the statistical evidence suggests that there are discernible patterns in the conflicts of
KPFs. Factional media from both conflicts also support the Regional Center (H2b) and Unpacific
Borders (H2c) hypotheses. In the first period, factional leaders called on their rivals to leave the
regional centers and accepted assistance from non-Kurdish forces when those centers appeared to
be lost to the rival faction. Turkey and Iraq entered the war when Duhok and Hewlêr were taken
by the PKK and PUK, respectively, while Iran joined with the PUK when Slêmanî was at risk of
being taken from the PUK.
Press releases by rival KPFs at the time reflect the caş and şêr distinction. According to
factional media, traitorous Kurds brought in outsiders to wrest control of cities sacred to Kurdish
identity and disrupt the politics of the true caretakers of the Kurdish nation. Additionally, the
importance of the border is also apparent from factional primary sources during the first period of
conflict. Press releases expounded on the traitorous nature of forming alliances with regional
actors.
231
The armed forces of the KPFs traversed the territory of their non-Kurdish allies to strike
at their Kurdish rivals in parts of their territory more distant from the fronts at the regional centers,
thus necessitating control of the border for defense.
Factional media from the KNC-KCK conflict also support the Regional Center (H2b) and
Unpacific Borders (H2c) hypotheses. Media affiliated with groups under each umbrella
organizations attacked the transnational relations of their rival.
232
Both viewed their rivals as
destabilizing the regional center of Şingal for the benefit of non-Kurdish powers. The regional
center was viewed as essential territory by each rival. With regards to the importance of the border,
the factions accused each other of being indistinguishable from TCOs and smugglers of illicit
products,
233
instrumentalizing the border for self-serving purposes. Though these are a potential

148

economic driver of conflict, the actual locations were not known, nor were these economic
incentives recorded with certainty comparable to the other objects of contestation.
Finally, the necessity to include indigenous Êzîdî in their KPFs most likely arose from the
need to appear locally supported and legitimate, in contrast to the transnational rival. Viewed as
the original Kurds, the endorsement from such a group would have done much to cement the
perceived legitimacy of one faction over the rival.
4.5 How do DSNs cooperate? Examining who dies and where they die
In this section I describe the results of tests pertaining to the third proposition; the nature of Out-
group Incursions (P3) should affect whether rivals continue their conflict or cooperate against the
out-group. The first hypothesis states that if unaligned coethnics are killed in institutionally
competitive territory we would expect to see more cooperation. This is the Who Dies and Where
They Die (H3a) hypothesis. The second hypothesis is Heroes and Traitors (H3b); if rival faction
members are killed in institutionally non-competitive territory, we would not expect to see
cooperation. The Non-combatant Deaths hypothesis (H3c) predicts that all deaths of civilian
Kurds, regardless of political affiliation or territorial control by a faction, should lead to
cooperation.
An overview of out-group incursions
Both periods of intraethnic conflict were interrupted by invasion by non-Kurdish forces. In
the first period, the KDP-PUK/PKK conflict, the armed forces of the states of Turkey, Iraq, and
Iran invaded Başur. The general directions of these invasions are displayed in figure 4.6:

149

Figure 4.6
External intervention into intra-Kurdish conflict in the 1990s

Note. Directions of invasion are approximate and do not follow a specified path.

Figure 4.6 displays which states invaded and how many Kurds were estimated to have been killed
by those external forces. We see that relatively few unaligned Kurds were killed relative to
politically aligned Kurds.
The opposite was the case in a different out-group invasion that interrupted intra-Kurdish
conflict, that of the KNC- and KCK-aligned groups. This conflict came to a halt when the ISG
invaded Şingal, Iraq, and Kobanî, Syria, in 2014. These communities were situated in territory that
had been recently acquired by the KDP and PYD, respectively, as the Iraqi and Syrian governments

150

withdrew in the face of the ISG threat, another exogenous shock. The threatened territory is
displayed in figure 4.7:
Figure 4.7
External intervention into intra-Kurdish conflict (KNC- and KCK-affiliated groups)

Note. Threatened territory is an approximation based on reports from various sources.

Kobanî is a primarily Kurdish town in Syria on the Turkish border that holds strategic and symbolic
importance for Kurds. The Êzidî of Şingal speak Kurmanji—some identify as Kurds, though many
do not.
234
Kurdish factions claim the Êzidî to be Kurdish and they are often claimed by those
factions to be representatives of pre-Islamic Kurdish culture.
235

151

In this conflict we see that significantly more unaligned Kurds were killed than politically
aligned Kurds. The ISG massacred thousands of Kurdish-speakers, enslaved thousands more, and
drove survivors into hiding. The loss of population and territory induced cooperation between
factions that were hitherto antipathetic toward each other. The PKK and PYD cooperated with the
KDP to rescue Êzidî civilians trapped on the mountains around Şingal that were at risk of being
killed by the out-group, thus adding to the number of deaths, as well as to drive ISG forces from
the area.
236
Furthermore, the KCK-aligned groups were permitted to establish bases in Iraq to help
clear the ISG from other recently claimed areas.
237
Several units of Peşmerge from the KRG also
assisted the PYD’s YPG Şervan in the battle to retake Kobanî with heavy weaponry that the YPG
lacked.
238

This out-group incursion is the only one that resulted in cooperative action. We shall return
to the salience of this momentarily, but first let us examine one last period of significant out-group
invasion. Turkish state forces (TSK) and Turkish-backed Free Syrian Army (TFSA) forces have
intervened in Iraq and Syria against KCK-aligned groups on numerous occasions. The ENKS in
Syria and the KDP in Iraq have possibly assisted TSK and TFSA forces against other Kurdish
factions during these interventions.
239
They did not assist their rivals against the outsider. Deaths
in this conflict are displayed in Figure 4.8:

152

Figure 4.8
Turkish intervention into intra-Kurdish conflict

Note. Territorial control was fluid throughout this time period. It is collated from numerous sources
to display such control at maximum extents.

In this period of intervention, we see much more aligned deaths than unaligned deaths.
Cooperation did not occur in this period.
Statistical evidence for the impact of unaligned deaths on rival factional cooperative action
Let us return to the period of cooperation that occurred between KNC and KCK groups
against the ISG to explore evidence for the hypotheses under the Out-group Incursion proposition
(P3). The KDE displayed in the analytical map was reproduced in a statistical program and

153

subjected to further analysis. The intensity of cooperative action appeared strongest in Şingal and
Kobanî, where the greatest loss of unaligned coethnic life occurred. Şingal also remains the most
contested region following the end of cooperation, in which direct armed confrontation between
KCK- and KNC-aligned forces has taken place. It potentially offers powerful support to the Who
Dies and Where They Die hypothesis (H3a) because this is an area that neither faction could afford
to lose to the ISG if they wished to pursue populational-territorial state-making activities.
Considered pagans and devil-worshipers, collaboration with the ISG was not an option for the
Êzidî, thus negating the logic of using selective violence on the part of ISG to win support among
the populace.
240
For KPFs, holding the area—even if it meant competing with rivals later—was
preferable to the permanent loss of the population and territory to the indiscriminately violent out-
group.
However, there is a possibility that at least some of this phenomenon was the product of
CSR, though it was not likely. To assess the possibility that some clusters of cooperation were
unrelated to each other (that is to say cooperation was not deliberate), the KDE was subjected to
K- and L-function analyses. The results of the K- and L-functions suggested that clusters of conflict
whose events were located less than approximately 36 kilometers apart were not likely to be the
results of CSR. It was virtually impossible for clusters composed of events less than five kilometers
apart to be considered random. The KDE, as well as the K- and L-functions are displayed in
Appendix E, Figures E.6 and E.7.
An ANN analysis with a Monte Carlo simulation to exact a pseudo p-value was then
performed to compare the results so far with random, independent data as part of one more test for
randomness. The observed values occur much lower in frequency than we would expect to see if
cooperative events were independent from each other. The ANN and K- and L- functions addressed

154

the second-order effect of independence. The evidence supported the assertion that these events
are related and not the product of chance. The ANN analysis is the subject of figure E.8 in
Appendix E.
Poisson point process modeling
Next, a technique was used to assess whether first-order effects were at work, meaning that
cooperative action was a product of underlying processes. The non-stationary PPP with LRT
compared the effects of covariates on cooperative action across four models. The first model
contained the null hypothesis; the phenomenon of factional cooperation was unexplained by any
underlying process. The second model was the Who Dies and Where They Die hypothesis (H3a),
in which the intensity of factional cooperation action was explained by the deliberate killing of
unaligned coethnics. The third model posited that targeting rival coethnics could explain
cooperation or the lack thereof, which tests the Heroes or Traitors hypothesis (H3b). Finally, the
fourth model was a form of control to ensure that a simpler dynamic was not at work. It posited
that all Non-combatant Deaths (H3c)—regardless of affiliation—drove cooperative action. The
results of these models are shown in table 4.3:

155

Table 4.3
Poisson point process modeling of Kurdish rival cooperation, 2014-2017

Analysis of Deviance Table – Likelihood Ratio Test
_______________________________________________________
Model 1: ~ 1 (Null) Poisson
Model 2: ~ Unaligned Deaths Poisson
Model 3: ~ Rival Deaths Poisson
Model 4: ~ Non-Combatant Deaths Poisson
_______________________________________________________
Model Deviance Pr(>Chi) Zval
1 0
2 18.9599 1.335e-05 *** 5.517566 ***
3 16.765 4.232e-05 *** -5.251897 ***
4 17.204 3.358e-05 *** -5.339364 ***
_______________________________________________________
AIC: Start = 529.23
- Unaligned Deaths = 561.24
- Rival Deaths = 549.62
- Non-Combatant Deaths = 560.41
_______________________________________________________
Note. Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
________________________________________________________

The deviation from the null in all models was strongly positive and statistically significant. Thus,
we may reject the null hypothesis (Model 1) of no effect. However, the only positive and
statistically significant cooperative probability (Zval) is from the second model, which tested the
effects that the deaths of unaligned coethnics may have on cooperative action.

156

The results of this test strongly support both the Who Dies and Where They Die (H3a)
(Model 2) and the Heroes and Traitors (H3b) (Model 3) hypotheses. As expected, the killing of
rivals yielded a significant, negative probability of cooperative action. We may confidently reject
both the null hypothesis and the contention that who dies and where they die does not play a role
in factional cooperation.
Interestingly, Model 4 (Non-combatant Deaths, H3c) suggests that including all civilian
deaths appears to provide a significant, negative probability of cooperative action. This result
suggests that affiliation must be accounted for when considering whether cooperation between
rivals may occur. The first generation of literature regarding rival cooperation was perhaps
mistaken in its emphasis on emotional affect. However, with modifications to our methods, we
can still use this generation if it accounts for cynical reasons rivals may cooperate. This finding
will be thoroughly explained in Chapter Eight and its implications for future research is addressed
in Chapter Nine.
Factional evidence of rival cooperation
Factional sources support the statistical interpretation of the effects of deaths on rival
cooperative action. In contrast to the encroachment by the Iraqi government in the 1990s, the
second period witnessed the greatest numbers of out-group incursion events in areas of new
factional competition, rather than in areas of established factional control. Importantly, rival
factions justified their cooperation with rivals and praised the fighters with whom they had hitherto
been in conflict. Groups affiliated with both the KNC and KCK invoked Kurdish identity as
something that needed to be safeguarded from invaders.
241

157

One area that is somewhat puzzling is Hesekê, Syria. With roughly the same number of
rival and unaligned Kurds killed, some cooperation perhaps could have been expected. Indeed, the
ENKS and KDP requested permission to deploy Rojava Peşmerge forces there on multiple
occasions,
242
but were rebuffed by the PYD, who repulsed the ISG successfully. Through
qualitative research into this seemingly puzzling incident, we gain more support for the Heroes
and Traitors hypothesis (H3b), as we would not expect a faction that can repulse an outside force
to allow their opponent to alter the institutional equilibrium or the populational-territorial status
quo through cooperation under those circumstances.
This period of cooperation did not endure. Once the ISG receded as a threat, each faction
divided up the disputed territory and established proxy forces in each other’s new zones of
influence. The factions and their proxies resumed the arrests of sympathizers, killing of supporters
at checkpoints, and border skirmishes. These activities culminated in the Şingal Clashes of March
2017, resulting in dozens of casualties on each side and a renewed humanitarian crisis as yet more
Êzidî fled.
243

The KDP demanded that KCK groups leave Iraq. Their armed forces also continued to
clash.
244
As intra-Kurdish conflict resumed, external intervention occurred again. The TSK
commenced air strikes, taking pains to avoid killing civilians in successive military operations
targeting PKK, PYD, and YBŞ bases and leaders in a campaign that continues as of October
2021,
245
leading to the conflict displayed above in figure 4.8. The PKK and PYD remain
institutionally competitive through their proxy force, the Şingal Resistance Units, or YBŞ.
246

In Syria, numerous TSK operations in support of TFSA and SNA forces wrested territory
from the PYD-dominated AA of North and East Syria in 2018 and 2019, which was resisted by
the PYD and its armed forces.
247
While there were some instances in which unaffiliated Kurdish

158

civilians were killed by either airstrikes or through TFSA abuses, most of the violence in Syria
was directed at YPG fighters, PYD politicians, and activists, all of whom are counted as rivals in
the data due to their political affiliation.
The KDP and their ENKS affiliate were indirect beneficiaries of TSK intervention in Iraq.
They were once more able to compete with the PYD in Syria. The result of no cooperation was
expected. As with other periods in which cooperation did not materialize, each side accused the
other of being pawns of external powers and traitors to Kurdish identity; the “lion” and “donkey”
distinction was made by the KCK and the KNC groups in proclamation of their armed forces as
true Kurdish warriors and leaders, while their opponents were depicted as slaves to foreign
powers.
248

4.6 Summing up: Tentative findings from DSN conflict and cooperation
This section sums up the findings of the three propositions and their hypotheses for the Kurdish
case. These results will be briefly distilled into a description of DSN factional conflict and
cooperation. Then, potential implications for the literature based on the results will be discussed.
This is expanded upon in Chapters Eight and Nine once all cases have been covered.
The results for this case are presented below in table 4.4:

159

Table 4.4
Summary of evidence for DSN conflict based on KPFs

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

P1 Roots of Rivalry
H1a Competing Leaders KDP-PUK/PKK Y Factional Weak
material
KNC-KCK N Factional Moderate
material
H1b State Manipulation KDP-PUK/PKK N Statistical Moderate
analysis,
factional
material
KNC-KCK N Statistical Strong
analysis,
factional
material
H1c Divergent KDP-PUK/PKK Y Temporal Strong
Development analysis,
factional
material
KNC-KCK Y Temporal Strong
analysis,
factional
material
P2 Effects of Distance
H2a National Capital KDP-PUK/PKK N Statistical Strong
analysis,
factional
material
KNC-KCK N Statistical Strong
analysis

160

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H2b Regional Center KDP-PUK/PKK Y Statistical Strong
analysis,
factional
material
KNC-KCK Y Statistical Strong
analysis,
factional
material

H2c Unpacific Borders KDP-PUK/PKK Y Statistical Strong
analysis,
factional
material
KNC-KCK Y Statistical Moderate
analysis,
factional
material
H2d Primary Routes KDP-PUK/PKK N Statistical Moderate
analysis
KNC-KCK N Statistical Moderate
analysis
P3 Out-group Incursion
H3a Who Dies and KNC-KCK Y Statistical Strong
Where They Die analysis,
factional
material
H3b Heroes and Traitors KDP-PUK/PKK Y Statistical Moderate
analysis,
factional
material Moderate
KNC-KCK Y Statistical
analysis,
factional
material

161

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H3c Non-combatant KDP-PUK/PKK N Statistical Strong
Deaths analysis,
factional
material
KNC-KCK N Statistical Strong
analysis,
factional
material

Note. Weak support: evidence was either statistical (*) or qualitative. Moderate support: either
statistical (*) and qualitative evidence or statistical (**). Strong support: both statistical
(** or ***) and qualitative evidence.

For the Roots of Rivalry proposition (P1), it is apparent that there is support for some of the
hypotheses with regards to why factions within DSNs may fight. The impact of the Competing
Leaders hypothesis (H1a) receives weak support in favor of its salience. Though the split between
the KDP and the PUK contributed to their mutual hostility splits played no discernible role in the
other dyads. The split between the KDP and the PUK did not lead to war, which took nearly two
decades to materialize, and when it did break out it came about through other motivations.
The role of State Manipulation (H1b) does not appear to be a compelling explanation in
either of the causes for conflict onset between the two dyads. Though the Hussein regime tried to
take advantage of the war between KPFs, it did not instigate the conflict. There is also no evidence
that suggests that Iraq or Syria encouraged KPFs aligned with the KNC or KCK to engage in

162

conflict with each other. Such absence is not evidence in and of itself, of course, so if such evidence
were to surface, it would call for revisiting this hypothesis.
Divergent Development (H1c) appears to be the best explanation for the growth of rivalry
within dyads of DSNs. Interactions within the region and with the wider international system had
the strongest impact. KPFs created incompatible institutions that came into conflict when they
expanded. These developments originated in the aftermath of the imposition of IRBs. Recall the
temporal analysis and its provision of clues that led to the identification of displacement and
expansion. Those KPFs adjusted to new political, economic, social, and cultural realities, but they
also maintained some transnational linkages that provided opportunities to later seek more
populations to influence once they were able to. These opportunities followed exogenous shocks.
DSNs seem to be most similar to TSONs with regards to the Roots of Rivalry. System-level factors
are the strongest predictors of rivalry, just as they are for the division of a people into two sovereign
states.
249

The second proposition states that the Effects of Distance (P2) should have an impact on
what DSNs might fight over. Interestingly, the distance from the National Capital (H2a) appears
to be the opposite of what we would expect if DSNs are most similar to SSSs as it relates to the
opportunity for conflict. KPFs appeared to be unconcerned about how close or far Baghdad and
Damascus are from their zones of contention. As we shall see, this is distinct from any other variety
of organized intraethnic conflict.
The effect of distance from the Regional Center on conflict events, however, supports the
notion that this POI is salient for DSNs. Conflict events among KPFs were clustered around these
centers, with strategic locations—dams, highways, legal and illicit cropland, natural resources—

163

being somewhat less important. This potentially makes DNSs most like SSSs in their objects of
contention.
Unpacific Borders (H2c) appear to also affect intraethnic conflict among factions within
DSNs. The border can be both an obstacle and an opportunity for DSNs. KPFs took advantage of
state weakness to traverse IRBs, and they were also granted permission to use foreign territory by
their regional allies. Conflict events seem to have a decay impact from the IRB that could be the
subject of future study. In this regard, DSNs may be most similar to DIGs in the manner by which
they are capable of projecting power beyond a base of support. Chapter Six will demonstrate how
the DIG is similarly limited by the border. Chapter Seven will show that TSONs are potentially
similarly limited, but also how they can overcome these limitations in ways DIGs and DSNs
cannot.
Out-Group Incursion (P3) seems to have a powerful impact on whether DSNs might seek
to form an alliance to eject an outside force. Who Dies and Where They Die (H3a) suggests that
cooperative action is driven by the deaths of coethnics who are not firmly aligned with a faction.
This is most likely for cynical reasons, as the factions wish to remain institutionally competitive
with that population in order to access the affected community’s territory. However, this
consideration is at present unique to the DSN.
The Heroes and Traitors (H3b) hypothesis, which posits that cooperation will not occur
when rivals are selectively targeted, is likewise supported in this work. When unaligned
noncombatants were not indiscriminately killed by out-group forces, rival factions withheld
assistance from each other; even if non-combatants were killed (H3c). Ethnic solidarity did not
occur, and rivals instead branded each other as traitors to Kurdish identity.

164

H3b may set DSNs apart from the other types completely. SSSs do not seem to care about
unaligned deaths in the midst of ethnic defection; DIGs cooperate regardless of who is killed;
TSONs in the 21
st
century will apparently never cooperate without absorption or unification.
Interestingly, DSNs cooperating to protect territory and populations for reasons of competition are
the only type of organized intraethnic conflict which exhibit this pattern.
The following three chapters will apply the same logic and series of analyses to cases
representing an SSS, a DIG, and a TSON. Now, let us turn to the SSS and discuss intraethnic
conflict among Chechen Separatist Factions (CSFs).

165

Chapter Five
Single State Separatists: Intraethnic Conflict among Rival Chechen Factions

This chapter addresses a comparative archetype for the DSN with regards to intraethnic conflict,
namely the single-state separatist (SSS). Individual- and, even more so, state-level factors are
argued to be the most significant drivers of intraethnic conflict for the SSS. Support is offered in
the form of statistical and qualitative analysis. Unsurprisingly, nearer proximity to the national
center is negatively correlated with conflict events. This finding presents a point of difference
between this intraethnic conflict variety and that of the DSN. Regional centers are demonstrated
to be the main objects of contestation, like the DSN. Borders, even those internal to the state, are
not as impactful as they are for DSNs as objects of contention. Most significantly, unlike the DSN,
rival factions are not motivated to band together against out-group invaders—even when large
numbers of neutral coethnics are killed by outsiders.
Chapter Five consists of six sections. Section 5.1 briefly describes Chechen history from
ethnogenesis until present. The two periods of conflict will be examined in Section 5.2. Arguments
in support of this work’s propositions and hypotheses will be assessed in Sections 5.3 and 5.4,
while evidence for the mechanisms of the best supported pathways will be presented in Section
5.5. Chapter Five concludes by summarizing the findings for this archetype and comparing them
to the DSN in 5.6.

166

5.1 Historical overview of the Chechen people: Resistance to outsiders and internal
cleavages
The word “Chechen”, like many ethnonyms, is an outsider’s usage of an indigenous self-
appellation. The people we now identify as Chechens call themselves “Nokhchii” and speak a
language that is part of the Nakh family. The phrases for their territory include “Nokhchichoe”,
“Nokhchiin mokhk”, and “Daimokhk.” These terms mean “Chechen home”, “Chechen country”,
and “Fatherland”, respectively.
250
“Chechen” originated as a Nakh expression meaning “within
territory” and has occurred with its modern meaning since the 16
th
century, though the Chechen
people themselves are much more ancient. There are approximately two million of them
worldwide, with the majority (1,200,000) being found in the Chechen Republic of the North
Caucasus district of Russia. Significant numbers also live in the neighboring republics of Dagestan
(90,000), Ingushetia (19,000), Stavropol Krai (12,000), and throughout Russia (50,000). Diaspora
communities are found in the European Union (130,000), Turkey (100,000), and elsewhere in the
Anglosphere. There are also several thousand Chechen refugees in Georgia that fled from violence
during the intra-Chechen conflict continuum.
The Nakh-speaking peoples of the North Caucasus include the Chechens, Dagestanis,
Ingushetians, Kists,
251
and others, though Chechens are specifically a Vainakh people, which is a
linguistic subcategorization.
252
The Vainakh are genetically and linguistically diverse in addition
to being ancient. Though predominantly of Neolithic farming origin, making them most closely
related to populations of the northern Near East, Chechens also exhibit Indo-European admixture
from prehistory.
253

From prehistory to present, the people whom we now call Chechens have been known for
their resistance to outsiders. They were known to ancient Georgian and Armenian historians as

167

Durdzuk or Dourt.
254
In the medieval chronicles of the Georgian and Armenian royal annals,
ancestral Chechens were known for having fought off expansion by many groups, including
Cimmerians, Scytho-Sarmatians, Avars, Huns, Khazars, Arabs, and Mongols.
Throughout this period of development, the ancestors of the Chechens had a complex
system of organization. The tsa, or house, was the smallest unit of social organization for Nakh
peoples. Multiple tsa would form a gar (branch) or nekh (road) of a teip. A teip has a council of
elders that make decisions pertaining to political, economic, and socio-political realms of life.
More than one teip could come together to form a union called tukkhum. Though not perfect
analogues, Chechen organization is roughly similar to the following levels of organization; tsa,
sub-clan; gar/nekh, clan; teip, tribe; tukkhum, confederacy. Though 21
st
century wars have “eroded
the traditional role” of the clan-based structure,
255
the starosty, or elders, retain some influence
over Chechen society. The Chechen tribal system allowed for a measure of democratic
involvement by free males that reinforced their reputation as fiercely independent. It was also
heavily influenced by blood feuds and an honor system in which insults and attacks against kin
would be repaid with violence.
The Vainakh religion had its own cosmology and pantheon of gods, led by the supreme
god, Deela. Some Chechens still refer to Allah as Deela.
256
Between the sixth and 13
th
centuries,
Georgian, Armenian, and Byzantine efforts to Christianize the Vainakh peoples met with varying
levels of success. Mongol invasions in the 13
th
century were followed by the Tartars and the
Golden Horde, and finally those by the Timurids. The chaos introduced by some invasions allowed
for a resurgence in paganism.
257

A Vainakh principality, Simsir, formed by the 14
th
century. It allied itself with the Golden
Horde, and as such adopted Islam as its official state religion. Most Vainakh did not follow suit at

168

that time, however. Simsir was destroyed by Timur the Lame in 1395. Makhama, son of the last
ruler of Simsir, Gayur Khan (also called Khour Ela), was appointed by Timur to serve as Beg
(prince) but was assassinated by rival Chechens who reinstalled Gayur. Gayur was then killed
during negotiations.
258
A vacuum formed in the wake of the Golden Horde’s destruction and the
eventual decline of the Timurid Empire.
Ottoman, Safavid, and Russian competition over influence in the Caucasus followed the
collapse of Simsir. Sufi Islam spread gradually during this period, with an uptick in conversions
in the 19
th
century.
259
Most Chechens came to belong to the Qadiri brotherhood, though
Naqshbandiya maintained a strong presence for a time. The inward focus of Sufi Islam appealed
to the independent nature of the Chechen people.
Chechens, Ingush, Avars, Alans, Dagestanis, and others formed the Caucasian Imamate to
combat Russian expansion into the Caucasus in the early 19
th
century. This union was fraught by
blood feuds and internecine fighting. The Chechen people were mostly subdued by the Russian
Empire after the capture of Sheik Shamil (himself an Avar), the last leader of the Caucasian
Imamate, in 1859.
260
The Murid War was the final portion of the conquest of Chechnya, which
was completed in 1864. Many prominent figures of the Caucasus Imamate were killed as a result
of blood feuds, which likely hindered effective resistance to the invading Russian forces. As we
will see, this dynamic would play out again and again when Chechen unity in the face of Russian
invasions was concerned.
The period of Tsarist control brought several changes to the Chechen people. The majority
of Chechens were Sufi Muslims by the Tsarist period, and defection from the Naqshbandi tariqa
(brotherhood) to the Qadiri rite accelerated.
261
Chechens lost much of their territory in the northern
lowlands of what is now the Chechen Republic, and they only began to resettle there after their

169

return from deportation in the 1950s. Chechens were also treated unfairly under the law; rather
than being tried in civilian courts like Russians and Cossacks, they were under the jurisdiction of
military courts.
262

European-style nationalism among some Chechen figures arose in the 1870s, which led to
several ineffective uprisings against the Russian Empire. Following the First World War, some
prominent nationalists endorsed communism. The Mensheviks, and later the Bolsheviks, promised
greater degrees of Chechen autonomy than the Whites did during the Russian Civil War. However,
the Bolsheviks—and then the Soviets—quickly came to be viewed as just another type of
imperialist conqueror.
263

Soviet persecution made some elements of the population more amenable to a
fundamentalist variety of Hanbali Sunni Islam. Today, Chechens remain predominantly Sufi, of
mostly Qadiri tariqa with some remaining Naqshbandi (following either Shafi’i or Hanafi schools
of Sunni jurisprudence for legal matters). However, there is a small but growing segment of the
society that has been trending Salafist, or Wahhabi. Fighters in failed uprisings against the Soviets
and later in wars against Russia would sometimes come to be split, in part, by religion.
264
Blood
feuds could result if a warlord received patronage from Salafist backers instead of a Sufi
brotherhood, or vice versa. However, religious disputes between leaders did not necessarily result
in intraethnic violence in the same manner that Russian instigation did. We will see this dynamic
with the religious split between two factions that initiated the second period of conflict examined
in this chapter.
Shortly after annexation by Russia, much of the Chechen population was removed from
the lowland area in the north of what is now Chechnya, as shown in figure 5.1, below:

170

Figure 5.1
Distribution of the Chechen population in 1878

Note. Modern borders are used for illustrative purposes.

Many Chechens were removed from the lowlands area in the north prior to this period. The
Chechen population in its entirety was later deported to Siberia by the Soviets in 1946. Chechens
were considered enemies of the USSR for ties to and sympathy for the Axis powers in the Second
World War.
265
Following the death of Stalin in 1953, Chechens began returning. This was
officially allowed by the Soviet state in 1956. The areas inhabited by Chechens following this
return from exile is displayed below in figure 5.2:

171

Figure 5.2
Distribution of the Chechen population in 1956

Note. Modern borders are used for illustrative purposes.

Modern borders were not yet finalized, though the Chechen population has remained
relatively stable with regards to the territory that it inhabits since this period. And so, changes may
be visualized and quantified, as depicted in figure 5.3 and table 5.1:

172

Figure 5.3
Temporal analysis: Change in distribution of Chechen populations before and after deportation

Note. Modern borders are used for illustrative purposes.

173

Table 5.1
Temporal analysis: Change in Chechen-inhabited territory, 1878-2015
Year Total area Unchanged area Area lost Area gained Net gain
km
2
km
2
km
2
km
2
km
2

(%) (%) (%) (%)

1878 11,162 NA NA NA NA

1956-1991 326,868 12,058 8,814 2,348 3,243
(79) (21) (29) (8)

Note. Km
2
have been rounded to the nearest whole number.

Following ethnic cleansing in the 19
th
century and prior to the dissolution of the Soviet
Union, Russians, Cossacks, and Turkic peoples settled in the lowlands north of the Sunzha river
in the northern portion of Chechnya. When Chechens returned from exile, they found out-group
members living in their actual houses. Chechens have since become the majority population in
some of that area in the 21
st
century as non-Chechens migrate to other parts of the Russian
Federation.
Some Vainakh and Nakh groups that had once been considered Chechen also relocated,
willingly or not, to Ingushetia over the centuries. Others were displaced into Georgia, such as the
Kists and Bats. This process and ensuing Chechen nationalism that advocated for a nation-state
would solidify the differences between Vainakh peoples. Chechens and Ingush would come to
regard themselves as different peoples first.
266
Kists and Bats would later do the same.
The change in population and territory between the return in the 1950s and the end of the
Soviet Union in 1991 are perhaps of the greatest interest here because it sets the stage for the

174

conflicts to come. As figure 5.3 shows, many territories inhabited by Chechens in what is now
Dagestan, Ingushetia, Georgia, and North Ossetia-Alania were not resettled by Chechens when
they were allowed to return. However, much of the area in central Chechnya and in Grozny was
resettled, as was some of northern Chechnya. though this would be depopulated again during the
conflict continuum. The original Soviet intention was to resettle Chechens in the northern plains
and away from the mountains in the south to manage future rebellion.
267
This was not achieved,
however, as we see in figure 5.3. With these losses in mind, table 5.1 tells us that there was a very
modest net gain in territory in which Chechens formed the majority population.
5.2 A depiction of intra-Chechen conflict
What is now Chechnya and Ingushetia were combined during Soviet rule to form the Chechen-
Ingush Autonomous Soviet Socialist Republic (ASSR), but the region was far from autonomous.
Russification policies continued, while Chechen nationalist movements in opposition to these
policies also grew.
268
The Soviet Union collapsed in 1991, which meant that the failed sovereign
that was the USSR left a security vacuum that was filled by ethnonationalist forces, just as it was
in other former communist states.
269

Post-Cold War nationalist movements and the Chechen conflict continuum
Numerous nationalist movements set out to replace the Checheno-Ingush ASSR with an
independent state constructed around the Chechen identity.
270
One of the most important of these
movements was the All-National Congress of the Chechen People (NCChP/OKChN), which
would come to establish the Chechen Republic of Ichkeria (ChRI), which would be the most
prominent Chechen separatist faction (CSF) for most of the 1990s.

175

The NCChP was led by Dzhokar Dudayev, a former Major General in the Soviet Air Force
and the first general officer of Chechen origin in the Soviet Union. He was born days before his
family was deported to Siberia in February of 1944. Despite his loyal service in Afghanistan, he
disobeyed orders to suppress Estonian uprisings in 1990.
271

Dudayev, with the backing of major political parties that included the Vainakh Democratic
Party, the Islamic Path, and the Green Movement, abolished the Checheno-Ingush ASSR by force
on September 6
th
, 1991, and declared the Chechen-Ingush Republic’s independence from the
Soviet Union in November of that year.
272
Boris Yeltsin proclaimed a national emergency in
Chechnya. Internal turmoil followed, with the ChRI splitting with Ingushetia and declaring
independence anew in 1993.
Dudayev was not universally admired among Chechen nationalist movements and CSFs,
however. The Provisional Council of the Chechen Republic (PCChR) was an opposition
organization that sought to rule Chechnya instead of the NCChP.
273
Some Chechen elites who had
an interest in the Soviet-era economic sector of oil exploration wanted to retain closer ties with
Moscow. Others were members of TCOs. The PCChR group was led by Umar Avturkhanov, who
was backed by Moscow. Other prominent figures were former Dudayev backers. These figures,
such as Ruslan Khasbulatov and Ruslan Labazanov, wished to retain independence from Russia
but nevertheless joined or supported the PCChR.
For Ruslan Labazanov, his motivation for turning on Dudayev was personal. A prominent
member of the Noxçiyn mafi, or Chechen mafia,
274
Labazanov was a captain in the armed forces
of the ChRI and the chief of the Presidential Guard that protected Dudayev.
275
However, after
Dudayev’s forces killed one of Labazanov’s family members, as well as his fellow mobster

176

Nikolay Suleimanov, Labazanov swore out a blood feud on Dudayev and promptly joined the
opposition.
276

Instances of armed intra-Chechen conflict were sporadic from 1991. Russian involvement
accelerated intra-Chechen conflict in 1994. The First Chechen War between the ChRI and Russian
forces along with the PCChR and Labazanov’s forces (which split from the PCChR) lasted from
December of 1994 until August of 1996. Dudayev was killed in April of 1996, but the ChRI was
victorious. They resisted the Russians and the PCChR; they also defeated Labazanov’s troops in
Groznyy and in Argun. The new president, Aslan Maskhadov, signed the Khasavyurt Accord with
the Russian Federation. The fledgling republic was essentially independent, but it lacked
international recognition and faced difficulties trying to obtain acknowledgement even from
Muslim-majority states.
277

Peace did not last long, however. Just a few months after the end of the war, Chechen
nationalists blew up a Russian barracks. This was followed by attacks on other Russian targets in
the Caucasus and in Moscow.
278
Chechen Islamist forces led by Shamil Basayev then invaded
Dagestan in 1999. Intra-Chechen divisions were fanned by the Federal Security Service (FSB),
which supported the Special Purpose Islamic Regiment (SPIR) as a means of driving a wedge
between Chechen nationalists. The SPIR consisted of a temporary alliance between oil smugglers,
religious extremists allied to Salafist newcomers, and Chechen mafiosi.
279
Numerous assassination
attempts made upon Maskhadov failed and Chechen nationalists kidnapped General Gennady
Shpigun, who served as the Kremlin's envoy to the ChRI. Following these developments, the
Russian Federation once again invaded.
280

The Second Chechen war featured a much more powerful Russian military, a more
nationalistic Russian public, and even more profoundly divided Chechen elites. The FSB and the

177

Main Directorate of the General Staff of the Armed Forces of the Russian Federation (GRU)
proved adept at manipulating Chechens with or without preexisting blood feuds into fighting each
other.
281
All-Chechen units were recruited by Russian forces to perform sweep-and-clear
operations to dislodge Chechen separatist fighters from towns and villages. These all-Chechen
units performed more effectively than Russian ones.
282
Village by village, town by town, and
finally city by city, Russian and Chechen opposition forces defeated those of the ChRI in May of
2000.
An insurgency conducted by the ChRI continued until 2009. This insurgency was fueled
by the increasing Salafist influence among some CSFs within the ChRI. Sufi brotherhoods played
a large part in the First Chechen War, but by the Second Chechen War they were no longer as
important; Salafists began to play a greater role.
283
Russian intelligence agencies would
successfully induce a split, which we will return to in the following sections.
The Qadiri Sufi Grand Mufti of Chechnya, Akhmad Kadyrov, had fought for independence
in the First Chechen War. Alarmed by the rise of Salafist influence and aided by Russia, Kadyrov
fought against the ChRI by late 1999.
284
He was appointed to lead the Chechen Republic by then-
prime minister Putin following the territorial defeat of the ChRI in 2000. After surviving a dozen
or so assassination attempts, his rivals succeeded in killing him with a remote-detonated bomb at
a parade in 2004.
285
Akhmad Kadyrov’s son, Ramzan, assumed control of the Chechen Republic
in 2007 and continued fighting ChRI insurgents until that particular conflict died down in 2009.
The short interlude of peace, the continuity of conflict goals, and the protracted nature of
the conflict itself suggest that the Chechen wars are perhaps best viewed as a continuum.
286
Thus,
figure 5.4 below displays the extent of these conflicts together:

178

Figure 5.4
Extent of the 1994-2009 intra-Chechen conflict

Note. Some conflict events are too near each other to be displayed individually. Territorial control
was too fluid to be mapped. The PCChR briefly controlled Urus-Martan and Ruslan
Labazanov controlled Argun-Shali in the First Chechen War. Conflict events involving
only Russian forces against ChRI forces are excluded.

Throughout the continuum, a Salafist insurgent movement formed with the goal of
establishing an Islamist state in the broader Caucasus region. Some ChRI fighters and leaders, such
as the Chechen first emir of the Caucasus Emirate, Dokka Umarov, were veterans of Salafist-
backed armed conflicts in Chechnya, Ingushetia, Georgia, and Dagestan. Umarov had become the
last leader of the ChRI before declaring the establishment of the Emirate. In general, the bulk of

179

ChRI members had taken a Salafist turn by 2007 (likely backed by Russian intelligence), becoming
unrecognizable to the identifiably Sufi character of the organization in the First Chechen War and
the beginning of the Second.
287
The religiously moderate members of the ChRI still claim to
represent the Chechen people, but no longer control a military force capable of engaging in armed
conflict.
Other key Salafist figures, such as Shamil Basayev, also joined the Emirate. Basayev had
long fought against Russian forces in both Chechen Wars. He led Chechen invasions into Dagestan
and Ingushetia, was responsible for many acts of terrorism—including the infamous Beslan school
siege that ended with the deaths of over one hundred children—and claimed responsibility for the
assassination of Akhmad Kadyrov. Basayev also swore out a blood feud on Ramzan Kadyrov,
even offering a bounty for his severed head.
288

The Chechen wing of the Emirate was known as the Vilayat Nokkhchicho (VN), or
“Chechen governorate.” The VN forces of the Emirate fought against Chechen Republic forces,
the Kadyrovtsy militia, and their Russian backers. Chechen Republic forces proved instrumental
in the degradation and defeat of coethnic Chechen factions that, ostensibly, shared a goal of greater
autonomy and adherence to the fundamental precepts of Islam in a globalizing world.
As security in the population centers improved, VN insurgents had to retreat into the
mountains of the south. Many VN fighters defected from the Emirate to join the ISG in Syria and
Iraq where they were the most in-demand elite troops.
289
By December of 2017 those remaining
in Chechnya had been severely reduced in strength and the insurgency was considered over. The
conflict is depicted in figure 5.5 below:

180

Figure 5.5
Extent of the conflict with the VN of the Caucasus Emirate

Note. Some conflict events are too near each other to be displayed individually. The Emirate
claimed control over all of the Chechen Republic. They physically controlled a handful of
rural locales of uncertain spatial boundaries. Conflict events involving only Russian forces
or no Chechen Republic forces are excluded.

5.3 The roots of SSS rivalry: Evaluating existing explanations
Let us now examine the origins of these conflicts in light of the Roots of Rivalry Proposition (P1)
and its associated hypotheses of Competing Leaders (H1a), State Manipulation (H1b), and
Divergent Development (H1c). In brief, H1a and H1b are best supported, with little support for
H1c.

181

Competing Leaders (H1a)
The periods of conflict make clear that elite competition was an element of intra-Chechen
conflict. The blood feud was central to elite customs and norms; numerous leaders died as a result
of this cultural institution.
290
Even former allies would engage in blood feuds after a falling out, as
Ruslan Labazanov did with Dzhokar Dudayev. Blood feuds going back to the Murid War, if not
earlier, hampered effective resistance against foreign invasions.
There was also a degree of religious disagreement between Chechen elites. From the
Russian conquest up to and during the first Chechen War, Naqshbandi elites lost ground to Qadiri
ones. After the First Chechen War, insurgents were increasingly likely to be Salafists. Elites still
compete on religious grounds. Consider Chechen President Ramzan Kadyrov; a devout Muslim
fundamentalist, he nevertheless cracks down on Salafists and supports Qadiri revival
movements.
291

State Manipulation (H1b)
It appears that H1a influences of intra-Chechen conflict. However, falling outs and the
resultant blood feuds are not the only factor that does so. The GRU and FSB were adept at pulling
away Chechen warlords and inducing them to turn on their coethnics even when a blood feud was
not sworn out on a foe. This was most evident in the Second Chechen War, but even in the First,
warlords defected from the ChRI when they received Russian backing. Labazanov, for example,
was given weapons and even main battle tanks by Russian backers to fight the ChRI.
292

Obviously, both individual- and state-level forces are key drivers here. However,
individual-level tensions seem to provide the foundation for conflict but are not necessarily the
immediate cause of it. This role goes to the host state in the case of intra-Chechen conflict. By the

182

Second Chechen War, Russian forces and intelligence services became increasingly skilled at
turning Chechens into counterinsurgents.
293
They were able to recruit Chechen counterinsurgents
even when there was no blood feud declared between leaders. State Manipulation receives a strong
degree of support.
Furthermore, the formal stateless status of the ChRI meant that the strength of the host state
heavily influenced the ability to maintain a degree of de facto sovereignty. The weakness of the
Russian military in the First Chechen War permitted the ChRI to essentially remain independent
following the Khasavyurt Accord. However, the ChRI was unsuccessful in its attempt to gain
recognition by key Muslim-majority states, other than Taliban-led Afghanistan.
294
The Second
Chechen war then witnessed a much stronger Russian state with a more nationalist prime minister
and populace. Without a recognized Chechen state, there came to be a wide array of Chechen
nationalists fighting each other to essentially assume command of a Russian-sponsored semi-
autonomous region.
Divergent Development (H1c)
Thus far, H1b appears to be the best-supported hypothesis, but system-level factors have
yet to be addressed. As with the Kurdish case, let us examine the effects of division by borders,
and political, economic, and socio-cultural linkages.
The Chechen nation is more or less geographically restricted to Russia. A few thousand
Chechen refugees fled to Georgia during the intra-Chechen conflict continuum. They settled in the
Pankisi Gorge, where some factions there contributed to unrest. Some of these refugees were
recruited by factions to fight in Chechnya but did not lay claim to parts of Georgia. In other words,
CSFs in Georgia did not develop self-determination claims in other states.
295
This is significant

183

because CSFs were not attempting to create multiple states or state-like entities across borders like
KPFs did and continue to do.
There were also Kists and Bats in Georgia. These groups are ethnic Chechens but have
diverged to such a degree that many Kists, Bats, and Chechens no longer consider themselves and
the others as the same group. There were some part-Kist mercenaries that fought against Russian
forces, like Tarkhan Batirashvilli,
296
but most fighters from the Pankisi Gorge are motivated by
Islam, rather than nationalism, and they now tend to fight in Syria and Iraq rather than Chechnya.
Furthermore, Kists and Bats did not join Chechen forces even in small numbers until after a refugee
population from Chechnya was established in the Pankisi Gorge. And so, division appears to be
relatively unimportant for intra-Chechen conflict. As far as religion and cultural ties went, episodes
of interethnic conflict, such as the Abkhaz-Georgia war, saw Islam working as more of a uniting
force than strictly ethnicity.
What role did transnational political, economic, and socio-cultural linkages play in the
origins of intra-Chechen conflict? International support for Chechen fighters was relatively limited
in the continuum. Chechen elites were divided between hardliners who wanted full sovereignty
and slightly less ardent nationalists who were more willing to accommodate Russian influence yet
wanted separation regardless. In terms of economics, elites who profited from the oil trade with
Moscow were only partially at odds with those whose revenue came from TOC and local criminal
activity.
297
Indeed, the SPIR was a temporary alliance between many of these divergent interests
that was held together by Russian support. Neither politics nor economics were very transnational
in scope. By the end of the continuum there was more diaspora support, but even this was relatively
minor for both periods of conflict.

184

With regards to socio-cultural linkages, the introduction of Salafism had an impact on elite
division, to be sure. However, this did not initiate conflict between Chechen groups in the
continuum. Salafists and Qadiris fought side-by-side unless blood feuds or, more likely, Russian
inducement to split came between them.
298
Some Chechens tend to see the Salafist-Sufi divide as
unimportant. Conflict against the Emirate came to be dominated by religious actors, but initially it
began as disputes between nationalists. The Salafists simply came to take over an existing
nationalist organization, and there was no violence between the Emirate and the remaining ChRI
moderates even after the split. So, while transnational religious linkages can prolong or morph
intraethnic conflict, it did not initiate it without some other intervention in this case. Additionally,
the Emirate still geographically ordered itself by ethnicity, apparently recognizing that it was
necessary to compete as separatists and nationalists as well as holy warriors, instead of as jihadis
alone.
One other factor to consider as far as the role religion played in the conflict with the Emirate
is Russia, the host state to the autochthonous Chechen nation. Russian forces sought out the same
Islamist forces that they fought within the Chechen Republic to fight against Russian foes
elsewhere. For instance, in the Abkhaz-Georgian Conflict, fighters such as Shamil Basayev were
supported by Russian intelligence to fight against Georgian forces. This occurred despite the fact
that he was a terrorist who fought Russian forces in the Second Chechen war as a ChRI fighter,
and later as an Emirati fighter. The Russian state saw transnationally supported Salafists as threats,
to be sure, but easily containable ones that could be subverted with sufficient effort. In addition to
not being responsible for initiating intra-Chechen conflict, the relative strength of transnational
religious movements has since waned in Chechnya, thus making this less significant than it appears
on the surface.

185

5.4 How do SSSs fight? A spatial analytic approach to SSS conflict patterns
What do the conflicts of SSSs look like? What objects are most contested? This section argues that
regional centers are—in a manner similar to DSNs. Unlike DSNs, however, the border is less
contested, while expectations for more conflict occurring further from the national capital hold.
The importance of highways is also unclear at this time.
Let us apply the Effects of Distance proposition (P2) and its associated hypotheses:
National Capital (H2a), Regional Center (H2b), and Unpacific Border (H2c) and Primary Route
(H2d). Recall that conflict should increase as distance increases from the national capital but
decrease as distance increases from the other POI.
Preliminary analyses
Pearson’s correlation statistics for the conflict continuum suggested that there were
strongly significant, negative effects for H2a, H2b, and H2d, and positive, strongly significant
effects for H2c. The opposite was expected for H2a (National Capital) and H2c (Unpacific Border).
Very similar trends were found for the conflict with the Emirate. These results are displayed in
figures F.2a and F.2b in Appendix F.
OLS regressions were the next step in the analysis. The similarity of the results across both
periods of conflict was striking. As expected, there were positive, though weakly significant effects
for both the conflict continuum and the conflict with the Emirate with regards to distance from the
national capital. So, conflict was more likely to occur further from ethnic Russian-inhabited
territory than closer to it. Regional centers had a strongly significant negative effect as expected
in both periods, so conflict occurred closer to the center. The border appeared to have a positive
effect that was strongly significant in both periods, which could be surprising if these effects hold.

186

We would expect the border to be fought over, yet it was not. Finally, primary routes were the only
POI that did not have consistent effects. These results are displayed in table F.1a in Appendix F.
Logarithmically transforming the continuous distance data to conform to a normal
distribution revealed similar results. The major difference between these models and the OLS
models was that the primary route results were more similar across periods, though they suggested
that there was a positive effect for the Regional Centers (1) model and a negative effect for the
Borders model (2). Cumulative results for H2d will be explored in detail in section 5.6. Results for
the log-transformed models are found in table F.1b in Appendix F.
At this point, there was enough evidence to probe for spatial autocorrelation and clustering
to explore this variation further. Global Moran’s I demonstrated very strong and very significant
autocorrelation. EBS results remained significant at 99% confidence, but with low-to-moderate
autocorrelation. The Getis-Ord G statistic resulted in very strong clustering for the first period and
weaker, though still strong clustering in the second. These test results are displayed in Appendix
F in table F.2.
Spatial lag models
With autocorrelation and clustering established, the use of a spatial model was justified. A
spatial weights matrix was constructed from the Moran’s I statistic and incorporated into the
models, which are displayed in table 5.2 below:

187

Table 5.2
Spatial analysis and modeling: Spatially lagged regression models of intra-Chechen conflict
Spatial Lag Model:
1994-2009 Intra-Chechen Conflict Continuum

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model
Model
Distance from National Capital (km)

0.401***

(0.107)

Distance from Primary Route (km)

-0.077
(0.166)
-0.133

(0.169)
Distance from Regional Center (km)

-0.346***
(0.092)

Distance from Border (km)

0.532***
(0.111)
Constant

-593.037***
(159.252)
-5.807*
(3.268)
Log Likelihood -1,159.818 -26.942
sigma
2
370.942 7.851
Akaike Inf. Crit. 2,331.636 63.884
Wald Test (df = 1) 4,449.189*** 0.001
LR Test (df = 1) 315.229*** 0.001
Note: *p<0.1; **p<0.05; ***p<0.01

188

Regional Centers Model Borders Model
Rho 0.9463
LR test value 315.23
p-value < 2.22e-16
Asymptotic standard 0.014187
error

z-value 66.702
p-value < 2.22e-16
Wald statistic 4449.2
p-value < 2.22e-16
Log likelihood -1159.818
for lag model
ML residual variance 370.94
(sigma squared)
(sigma 19.26)
AIC 2331.6
AIC for lm 2644.9
LM test for residual 450.82
autocorrelation
test value
p-value < 2.22e-16
Rho 0.010084
LR test value 0.0010641
p-value 0.97398
Asymptotic standard
error 0.28898
z-value 0.034894
p-value 0.97216
Wald statistic 0.0012176
p-value 0.97216
Log likelihood -26.94223
for lag model
ML residual variance 7.8514
(sigma squared)
(sigma 2.802)
AIC 63.884
AIC for lm 61.886
LM test for residual 1.4234
autocorrelation
test value
p-value 0.23284

189

Spatial Lag Model:
2009-2017 Conflict with the Caucasus Emirate

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.162***
(0.051)

Distance from Primary Route (km)

0.005
(0.140)
-0.060
(0.088)
Distance from Regional Center (km)

-0.270***
(0.052)

Distance from Border (km)

0.664***
(0.092)
Constant

-232.262***
(75.202)
-19.231***
(2.659)
Observations 102 13
Log Likelihood -338.137 -18.751
sigma2 41.925 0.961
Akaike Inf. Crit. 688.274 47.501
Wald Test (df = 1) 156.387*** 15.681***
LR Test (df = 1) 81.768*** 5.721**
Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

190

Regional Centers Model Borders Model
Rho 0.77786
LR test value 81.768
p-value < 2.22e-16
Asymptotic standard 0.062201
error
z-value 12.505
p-value < 2.22e-16
Wald statistic 156.39
p-value < 2.22e-16
Log likelihood -338.1369
for lag model
ML residual variance
(sigma squared) 41.925
(sigma 6.4749)
AIC 688.27
AIC for lm 768.04
LM test for residual 36.887
autocorrelation
test value
p-value 1.2516e-09
Rho 0.65245
LR test value 5.7207
p-value 0.016766
Asymptotic standard
error 0.16476
z-value 3.9599
p-value 7.4982e-05
Wald statistic 15.681
p-value 7.4982e-05
Log likelihood
for lag model -18.75071
ML residual variance
(sigma squared) 0.9613
(sigma 0.98046)
AIC 47.501
AIC for lm 51.222
LM test for residual 3.7058
autocorrelation
test value
p-value 0.054222

These results generally confirm the non-spatial models, but with greater confidence in
their utility. The AIC is slightly lower in the first period’s Regional Centers model and the second
period’s Borders model. It is significantly better in the second period’s Regional Centers model,
though ever slightly less useful for the first period’s Borders model.
It is apparent that both periods witnessed expected results for H2a (more conflict further
from the national capital) and H2b (more conflict closer to regional centers), though unexpected

191

results for H2c (conflict appeared further from the border—even internal borders with the other
krai). Once more, H2d results were inconclusive.
5.5 Do SSSs cooperate? Examining who dies and where they die
Unlike in the case of the KPFs, there are no statistical results of rival cooperation for CSFs in this
section. This is because intra-Chechen conflict did not cease once started. However, we can still
apply the logic of the Out-Group Incursions proposition (P3) and its hypotheses. There should be
cooperation where unaligned coethnics are killed and there should not be cooperation where
primarily rivals are killed. These predictions correspond to the Who Dies and Where They Die
(H3a), Heroes and Traitors (H3b) and Non-combatant Deaths (H3c) hypotheses, respectively.
Rivals were counted as the armed forces of rival factions, as well as civilians with a known
affiliation, such as politicians. However, any instance in which non-combatants were killed,
regardless of political affiliation, was also included for a separate category. Only those killed by
out-group forces (not as a result of intra-Chechen conflict, which were the subjects of P2) are
included in this section.

An overview of out-group incursions
Both periods of intra-Chechen conflict were interrupted by Russian intervention. In the
intra-Chechen conflict continuum, a briefly independent ChRI fought against Russian forces at the
outset of the First Chechen War in 1994. Some factions, such as that led by Ruslan Labazanov,
split off from the ChRI and fought as pro-Russian forces, but the ChRI still won this conflict and

192

forced Russian forces to sign a cease-fire agreement in 1996. Despite a heavy toll on the civilian
population, rival factions did not cooperate against Russian forces once intraethnic conflict broke
out.
299

Russian forces once again intervened in the Second Chechen War. Their involvement was
much more targeted in this conflict. Air strikes were carried out with greater precision and utilizing
improved intelligence capabilities compared to the First Chechen War. Unaffiliated civilians
represented a comparatively small share of the (verifiable) total killed by Russian forces.
300
The
Kadyrovs, hitherto major separatist leaders, established a pro-Russian regional government after
turning on their former ChRI allies.
After major combat operations ended in 2000, this war turned into an insurgency that lasted
until 2009, when all factions violently opposing the Chechen Republic became Islamist in nature.
The death tolls of rivals and unaffiliated are displayed below in figure 5.6:

193

Figure 5.6
Russian intervention into the intra-Chechen conflict continuum

Note. Deaths are a verified estimate of those killed specifically through out-group action. Thus,
not all deaths in the conflict are counted here.

Figure 5.6 demonstrates that more unaffiliated Chechens were killed than rivals as a result of out-
group action. We would expect to see some rival cooperation in light of this, but it did not occur.
Once a faction turned on the ChRI it became dependent on Russian assistance for survival. This is
likely a facet of separatist movements more or less geographically constrained to a single state.
The nature of competition makes it so that dependence upon the host state precludes the ability to
unite in the face of that host state’s actions.

194

Russian intervention in intra-Chechen conflict between the Chechen Republic and the VN
of the Caucasus Emirate was even less direct than in the Second Chechen War. As a result, very
few casualties of unaffiliated Chechens were sustained. Russian-inflicted casualties are displayed
in figure 5.7:
Figure 5.7
Russian intervention into the conflict with the Caucasus Emirate

Note. Deaths are a verified estimate of those killed specifically through out-group action. Thus,
not all deaths in the conflict are counted here.

We see from Figure 5.X that Heroes and Traitors (H3b) receives support. We would not expect to
see the Chechen Republic join with the Emirate’s VN forces, especially if rivals would not do so

195

when there were heavy casualties among the unaffiliated in the continuum. Civilian deaths,
regardless of affiliation, seem not have induced cooperation as H3c predicts.
Factional evidence
Statements made by factions and their leaders support the findings associated with P2 and
P3. The attitudes of CSFs toward each other supersede any apparent concern for Chechen
individuals. Russian ability to sway militants to fight coethnics is apparent as well from qualitative
inquiry.
In the continuum, both the First and Second Chechen Wars feature similar themes with
regards to factional conflict. The PCChR and Labazanov’s forces claimed Dudayev and the ChRI
were greater threats to the Chechen people than Russian forces (that were conveniently funding,
arming, and training them) in the First.
301
Dudayev for his part led the ChRI in uncompromising
attacks on coethnic foes.
302
In the Second Chechen War the ChRI under Maskhadov called for
Akhmad Kadyrov’s death, while the Kadyrovtsy faction viewed the ChRI as having failed to live
up to the ideals of Chechen nationalism, which could be seen as justification for collaborating with
Moscow.
303

The ascension of the Salafist faction of the ChRI and its subsequent split, thus forming the
Caucasus Emirate, marked the beginning of a new phase of conflict. However, themes remained
the same; civilian lives appeared to have relatively low influence on the inclination for rivals to
cooperate, while Russian influence was pervasive.
The religiously moderate faction of the ChRI bestowed upon the Chechen Republic—and
the Kadyrovs by name—a satirical award for being traitors to the Chechen people in one of the
few resources where civilian deaths were referenced, albeit without overtures for cooperation.
304

196

The ChRI also accused Russia of intentionally building up the Salafist faction to delegitimize the
nationalist cause by painting it as a mere extension of international terrorism.
305

It is worth noting that the post-split, moderate ChRI under Akhmad Zakayev was the
weakest faction in this period of conflict, as well as the only one to routinely invoke the deaths of
innocent Chechens. However, it could not act on those complaints. It no longer maintained major
armed forces. The stronger factions rarely voiced sincere concerns about noncombatants,
regardless of affiliation. The Emirate and the Chechen Republic would publicize atrocities
committed by their rival without acknowledging those committed by their allies. For example, the
Emirate under Umarov accused Kadyrov of targeting civilians, even as Umarov specifically
advocated for the targeting of non-combatants.
306
Civilians were not specifically targeted by
Russian forces by the end of the Second Chechen War—that task having been outsourced to
Chechen fighters. This, combined with Russia’s ability to manipulate both the Chechen Republic
and the Salafists, this should explain the lack of cooperation. The ChRI appears to be too weak to
utilize after the split with the Emirate.
5.6 Summing up: Tentative findings from SSS conflict and cooperation
Let us now examine the case of the SSS represented by CSFs in the aggregate. Table 5.3 displays
the findings for this case:

197

Table 5.3
Summary of evidence for SSS conflict based on CSFs

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

P1 Roots of Rivalry
H1a Competing Leaders ChRI-PCChR/ Y Factional Moderate
Chech. Rep.-ChRI material
Chech. Rep.-VN Y Factional Moderate
material
H1b State Manipulation ChRI-PCChR/ Y Statistical Strong
Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN Y Statistical Strong
analysis,
factional
material
H1c Divergent ChRI-PCChR/ N Temporal Strong
Development Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN N Temporal Strong
analysis,
factional
material
P2 Effects of Distance
H2a National Capital ChRI-PCChR/ Y Statistical Strong
Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN Y Statistical Strong
analysis

198

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H2b Regional Center ChRI-PCChR/ Y Statistical Strong
Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN Y Statistical Strong
analysis,
factional
material

H2c Unpacific Borders ChRI-PCChR/ N Statistical Strong
Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN N Statistical Strong
analysis,
factional
material
H2d Primary Routes ChRI-PCChR/ N Statistical Moderate
Chech. Rep.-ChRI analysis
Chech. Rep.-VN N Statistical Moderate
analysis
P3 Out-group Incursion
H3a Who Dies and ChRI-PCChR/ N Statistical Strong
Where They Die Chech. Rep.-ChRI analysis,
factional
material
H3b Heroes and Traitors ChRI-PCChR/ N Statistical Strong
Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN N Statistical Strong
analysis,
factional
material

199

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H3c Non-combatant ChRI-PCChR/ N Statistical Strong
Deaths Chech. Rep.-ChRI analysis,
factional
material
Chech. Rep.-VN N Statistical Strong
analysis,
factional
material

Note. Weak support: evidence was either statistical (*) or qualitative. Moderate support: either
statistical (*) and qualitative evidence or statistical (**). Strong support: both statistical
(** or ***) and qualitative evidence.

Let us begin with the Roots of Rivalry (P1). The Competing Leaders hypothesis (H1a) was
supported. Blood feuds were a significant hindrance to cooperation between CSFs. However,
intraethnic conflict began after Russian support was offered to the alternative CSF. Meanwhile,
divergent development did not seem to affect conflict much, if at all. The same political, economic,
and socio-cultural interests could be shared by opposing factions. What mattered more was which
faction Russia preferred to support at that time.
307

State Manipulation (H1b) seems to be the best supported explanation for the initiation of
conflict, followed by Competing Leaders (H1a). This stands in contrast to DSNs, in which
intraethnic conflict is driven by system-level factors associated with Divergent Development
(H1c). Recall that KPFs would fight each other even if their leaders experienced good relations.
They would cooperate even if they despised each other. They could also fight or cooperate despite
potentially not even knowing who their counterparts may be. Host states also had relatively little

200

influence on the initiation of conflict. With the SSS, however, the host state is the single greatest
driver of conflict. KPFs may experience external intervention after they initiate conflict, whereas
the CSFs receive state overtures prior to intra-Chechen conflict.
The differences in territoriality likely matter for the distinction between the origins of
conflict for the DSN and the SSS, though both are stateless. The limited, defined territory of the
SSS contrasts with the DSN. The SSS must compete for survival in the context of a single host
state’s goals and strength. The DSN, however, can compete for territory across borders that may
be beyond the ability of the host state to regulate. This is also the case for matters of transnational
religious linkages; Russia was more influential than Al Qaeda or Taliban-led Afghanistan in
driving intra-Chechen conflict.
Let us turn to the objects of contestation and the Effects of Distance (P2). As expected,
CSFs experience much less conflict nearer the area containing the dominant ethnic group—in this
case that group is ethnically Russian. The National Capital (H2a) hypothesis did not hold for KPFs.
The transnational linkages of the DSN
308
likely allow for such factions to acquire greater power
relative to their host states than the SSS, which lacks the same linkages. Thus, DSNs that reach a
certain level of capacity should become unconcerned about host state interventions absent of
system-level support for the host state.
309
This might emerge as a priority for future research as
detailed in Chapter Nine.
What of the other objects of contestation? As we expect, the Regional Center (H2b) was
supported and makes the SSS similar to the DSN in this regard. Control over the Clausewitzian
nerve centers should be vital for both groups.

201

A key difference is related to the Unpacific Border hypothesis (H2c). The simplest
explanation for why borders were so uncontested is because Georgian and Russian state forces
controlled the border. Attacking hard targets is difficult. Also, if CSFs received fighters from the
Pankisi Gorge, they would need them to arrive undetected, likely through mountainous goat trails
or other means that are difficult to track, even with satellite or UAV technology.
We once again have inconsistent results with regards to the Primary Route hypothesis
(H2d). It is likely for the same reason; these conflicts took place during time periods in which
reliable information simply does not exist. Future work involving the conflicts of tomorrow will
likely address this issue with more reliable data.
Finally, let us compare the SSS to the DSN in terms of their respective propensities for
cooperation against Out-group Incursion (P1) after in-group conflict begins. Unexpectedly, the
volume of non-combatant deaths had no effect on rival cooperation in either period of conflict,
regardless of affiliation. Thus, the Who Dies and where They Die (H3a) and Non-combatant
Deaths (H3c) hypotheses do not receive support as a feature of intra-SSS conflict. DSNs will
cooperate if the loss of territory and population is a likely outcome. Unlike the DSN, the SSS is
competing for a defined territory, thus the winner will take the territory and the remaining
population. The DSN, on the other hand, is creating the territory and needs the support of the
existing population to do so. This difference is perhaps the most salient between the two types, as
this is such a prominent departure with regards to what we may predict factions to do once engaged
in intraethnic conflict.
As expected, the Heroes and Traitors hypothesis (H3b) is supported by factional media
which greatly disparages the rival faction. The difference between the SSS and DSN is that SSS

202

leaders are also much more personally invested in rhetoric. While KPF leaders are much more
restrained, CSF leaders engage in blood feuds and attack their rival in very personal ways.
We will now turn to the next variety of conflict: the divided irredentist group (DIG). The
DIG is not stateless, making it unlike both the DSN and the SSS. It is, however, divided by borders,
like the DSN and unlike the SSS. This will make for an excellent comparison to the DSN to
examine different varieties of statehood or statelessness, while keeping division shared between
types.

203

Chapter Six
Divided Irredentist Groups: Rivalry and Conflict among Albanian Irredentist Factions?

The divided irredentist group (DIG) is represented by Albanian irredentist factions (AIFs) in this
chapter. The DIG arrangement requires the existence of a nation-state whose core community is
formed by the ethnic group of interest. In this instance there is Albania, which is mostly comprised
of ethnic Albanians. It also requires geographically contiguous populations of coethnics to be
divided by borders. Factions claiming to represent these enclaves seek to join the nation-state or
pursue separation or autonomy from the host state. In this chapter we are primarily talking about
ethnic Albanian factions in the Republic of Kosovo, which claims to be an independent state (and
functions as such) but is also disputed by Serbia as being integral to Serbian territoriality. We will
also consider Albanian factions in what is now North Macedonia. Major factions are drawn from
the Ushtria Çlirimtare Kombëtare (UÇK) which translates to “National Liberation Army,” and
some others that will be introduced throughout the chapter. For the purposes of this work, NLA
will refer to the UÇK in then-Macedonia, while Kosovo Liberation Army (KLA) will refer to the
UÇK in Kosovo.
Unlike DSNs, their rivalries do not tend to develop into large-scale armed conflict.
However, they also have a single nation-state that can hierarchically coordinate action and punish
transgressions against coethnics across states. This makes them unique among the four types in
this work. Competition is local; violence against coethnics is exceedingly rare and primarily due
to first-level personal grudges.

204

Like DSNs, DIGs cooperate against outsiders. However, its occurrence is vastly different
from DSN cooperation. They will cooperate even if rivals are killed and do not take into account
affiliation when non-combatants are killed.
This chapter consists of five sections. Section 6.1 briefly recounts Albanian history from
its ethnogenesis until just prior to the recent conflicts involving this ethnic group. These conflicts
are the subjects of Section 6.2. The propositions of the origins of conflicts, objects of contestation,
and cooperation are discussed in Sections 6.3 and 6.4. Section 6.5 concludes with a comparison of
the DIG to the DSN.
6.1 Historical overview of the Albanian people: Resistance to outsiders and internal
cleavages
There are at least seven million ethnic Albanians alive today. Roughly three million form the core
community of the nation-state of Albania. Another two million or so geographically contiguous
and autochthonous
310
Albanians also reside in Kosovo (1.7 million), North Macedonia (500,000),
and Montenegro (30,000). A few thousand more are found in Croatia, Slovenia, and Serbia sans
Kosovo. There were once hundreds of thousands in Greece and Turkey, though many were
expelled from Greece (recent Albanian migrants notwithstanding) and many in Turkey have
assimilated into Turkish culture. These latter two groups will be discussed shortly. A diaspora
population of at least 300,000 exists throughout the Western world.
311

The term “Albanian” comes from Greek, Roman, and Byzantine historians’ names for
tribes variously called “Albanoi”, “Arbanitai” “Albanenses”, and others.
312
The name for their
territory was called “Arbëria.” Albanians themselves used this outsider’s term to refer to

205

themselves, adopting it as “Arbërisht.” The indigenous term “Shqiptar” and its related forms,
which mean “those who speak clearly,” replaced these terms in the Middle Ages.
313

Albanians speak the only descendant of the Paleo-Balkan branch of the Indo-European
language family, though there is insufficient evidence to link them exclusively to Thracian,
Dacian, or most likely, Illyrian peoples.
314
These Balkan peoples warred with Greek, Roman,
Celtic, and Byzantine civilizations before being displaced by Slavic migrations to Wallachia,
where they became the Vlachs, and to the Adriatic coast.
315
Those settling along the Adriatic
became who we now call Albanians. Two dialects formed: The larger of the two is Tosk, in what
would become Albania, Greece, Italy, and Turkey. The other is Gheg, in what is now Croatia,
North Macedonia, Montenegro, and Kosovo, as well as some of northern Albania.
Christianity had been adopted by Albanians in antiquity. The northern tribes became
Roman Catholic, while southern tribes opted for Eastern Orthodoxy. Albanians formed numerous
principalities and other feudal entities in the Medieval era. The Principality of Arbanon was
established in the twelfth century. This was led by the Progon family, which was Orthodox by
confession and had tense relations with prominent Catholic families. These tensions did not lead
to significant intragroup conflict, however. Though an intersectarian alliance held up, the
principality was annexed by the significantly more powerful Byzantine Empire in 1257.
316
Other
principalities were created during Byzantine rule. These were led by rising houses, such as the
Kastrioti and Dukagjini, and together with the Byzantines checked Serbian expansion in the
Balkans.
The Kastrioti would supply the most famous Albanian national hero. Gjergj, better known
to history as Skanderbeg (“Lord Alexander”), is celebrated for his resistance to Ottoman invasions.

206

These invasions began when an Albanian lord of the Thopia family was defeated by the Balsha
family. He allied with the Ottomans to seek vengeance for the loss.
317

This would begin a cycle in which Ottoman suzerainty would be recognized temporarily,
until the ruling families would form a new alliance and eject the Ottomans. During times of
Ottoman control, Albanians would convert to Sunni Islam of the Hanafi school of jurisprudence.
It has since become a slim majority among Albanians, though Catholicism and Orthodoxy are yet
practiced by significant minorities.
Skanderbeg had been raised during the early days of this cycle. He had been converted to
Islam in his youth while a hostage to the sultan. His status as a hostage was meant to guarantee his
father’s fealty for the time being. As a son of a noble whose allegiance the Ottomans wished to
retain, he was educated and trained at Ottoman military academies. Skanderbeg returned to
Albania and initiated a rebellion during a successful war against the Ottomans undertaken by John
Hunyadi, Regent-Governor of the Kingdom of Hungary. Though Orthodox by birth and a Muslim
by compulsion, he was rebaptized as a Catholic to gain the backing of European powers in his
rebellion against the Ottomans.
318

Though they would fight among each other during times of peace, conflict with the
Ottomans united the Thopias, Balshas, Kastrioti, Dukagjini, and other important families. They
formed the Lidhja e Lezhës (League of Lezhë) in 1444 to combat Ottoman invasions, though it
fractured over time. Skanderbeg was its leader, and the League was supported primarily by Naples.
This source of support polarized some of the League’s members, and so it dissolved in the years
following Skanderbeg’s death in 1468.
319

207

Albanian principalities never coalesced into a unified kingdom prior to Ottoman conquest.
This aspect of Albanian history—much more so than religious differences—allowed for intragroup
elite competition. Though Skanderbeg held an alliance together for some time, his support from
Naples infuriated some of the other Albanian elites. This cleavage was exploited during those
cycles of Ottoman control following his life. Ottoman rulers would extend special privileges to
converts to Islam and to those elites who cooperated with them. Without a hierarchical system to
maintain an alliance it is logical to expect the Ottomans to have been able to take Albanian
territory, city by city, fortress by fortress, as happened by 1479.
320

Islam gradually became the majority religion under Ottoman suzerainty. No longer
accommodationist, Ottoman rulers levied more burdensome taxes and conscripted Christian
Albanian boys into the Janissary corps of elite slave-soldiers.
321
These realities spurred conversion.
Due to their martial prowess and other skills, Albanians were sought after for military leaders,
diplomats, envoys, bureaucrats, and politicians throughout the Ottoman Empire. As a result, many
Albanians settled in what is now Greece, Turkey, Egypt, and Libya.
322

By the 19
th
century, the Ottoman millet system of allowing limited degrees of autonomy
for ethnoreligious groups became more difficult to hold together in the face of rising
nationalism.
323
Geopolitical setbacks also challenged the Ottoman Empire’s capacity for holding
all its claimed territories. Economic crises and a military whose strength had evolved to rely on
German artillery
324
more so than its own troops set the stage for uprisings by Serbs, Greeks,
Albanian highlanders, and others in the Balkans during the Great Eastern Crisis of 1875-1878. The
Ottomans were also militarily defeated by the Russian Empire in 1878, which encouraged more
nationalist movements to become bolder.

208

Though Albanian nationalism began along confessional lines, religion came to be
unimportant relative to other Balkan nationalist movements as a Pan-Albanian movement quickly
emerged.
325
Ismail Qemali declared independence from the Ottoman Empire in 1912, serving as
Albania’s first prime minister. Albanian-inhabited Chameria and Kosovo were held by Greece and
Serbia, respectively. Italy incorporated it as a protectorate in 1926 and supported the Kingdom of
Albania in the Second World War, which set the borders for the modern state. The distribution of
the Albanian population at this time is displayed below in figure 6.1:
Figure 6.1
Distribution of the Albanian population in 1940

Note. Modern borders are used for illustrative purposes.

209

Italy’s inability to defeat Greece brought in German involvement, which saw the
establishment of an independent and nationalist Albania, run by the Këshilla, or “Council” under
anti-communist Balli Kombëtar. As the Allied Powers refused to grant Kosovo and Chameria to
Albania,
326
many Albanians believed the Axis was a preferable partner for achieving self-
determination. Aside from Kombëtar’s forces, Albanians also joined the 21
st
Waffen-SS Division
Skanderbeg and the 13
th
Waffen-SS Division Handschar.
327

The communist victories following the Second World War throughout the Balkans resulted
in the mass expulsion of ethnic Albanians from Chameria and ethnic cleansing against Albanians
in Yugoslavia.
328
The Yugoslav Wars of 1991-2001 witnessed a resurgence in irredentist ethnic
conflict between Serbs, Slovenes, Croats, Bosniaks, Macedonians, and Albanians protecting their
ethnic kin.
329
We will examine Albanian irredentism in this context shortly, but first let us see how
Albanian populations had changed after from the Second World War until the dissolution of
Yugoslavia and a renewal of the aspiration to create a Greater Albania. This is the subject of the
following figures and table. Figure 6.2 displays the post-Cold War distribution of Albanians.
Changes in territory are the subjects of figure 6.3, and table 6.1 quantifies these changes.

210

Figure 6.2
Distribution of the Albanian population in 1991

Note. Modern borders are used for illustrative purposes.

211

Figure 6.3
Temporal analysis: Change in distribution of Albanian populations after the Second World War
and prior to the Yugoslav Wars

Note. Modern borders are used for illustrative purposes.

212

Table 6.1
Temporal analysis: Change in Albanian-inhabited territory, 1940-1991
Year Total area Unchanged area Area lost Area gained Net gain
km
2
km
2
km
2
km
2
km
2

(%) (%) (%) (%)

1940 11,162 NA NA NA NA

1991 43,186 35,772 6,963 7,414 451
(84) (16) (17) (1)

Note. Km
2
have been rounded to the nearest whole number.

We see that there was a very modest gain of one percent in Albanian-majority territory,
like the Chechens and unlike the Kurds. However, closer examination reveals that a vast amount
of territory was lost by the expulsions of the Cham—though refugee movements and economic
migration would bring hundreds of thousands of Albanians back to Greek urban areas decades
later.
330
Some lands in the north were lost from the Yugoslav Wars, but regions of Kosovo, Serbia,
and North Macedonia became more homogenously Albanian, as figure 6.3 demonstrates.
Appendix G further elaborates on these results. Migrations will also be discussed in the following
section.
6.2 Background to the Yugoslav Wars and Albanian factional development
Now that the historical overview of the Albanians into the 20
th
century has been presented,
let us examine the roots of late 20
th
and early 21
st
century Albanian irredentist movements. The
National Liberation Movement (NLM) established a communist state after the defeat of the Axis

213

Powers, ending for a time the Pan-Albanian desire for a Greater Albania, as ethnic Albanians were
now dispersed throughout Albania and Yugoslavia. Communist Albania under the NLM alternated
in its support for the USSR, the PRC, and Yugoslavia, seeking to remain independent, and in so
doing discouraged irredentist aspirations for a Greater Albania. It sought to achieve autarky under
Enver Hoxha throughout the Cold War.
331

Post-Cold War division and irredentist movements
Following the Cold War, Albania transitioned to a market economy with some difficulty.
Several politicians devised pyramid schemes that profited them enormously but bankrupted a large
proportion of the population.
332
A rebellion broke out, empowering TOC organizations, which in
turn contributed to the rise of the KLA and NLA who were then in their infancy. These two groups
both use the initialism UÇK, though the K holds different meaning for each faction.
333
They were
among the major ethnic Albanian factions that took part in various conflicts of the Yugoslav Wars,
though not the only ones.
The Yugoslav Wars erupted following decades of nationalist suppression of the various
peoples who constituted the Federal People’s Republic of Yugoslavia (FPRY). The FPRY was
called the Socialist Federal Republic of Yugoslavia (SFRY) from 1963 until 1992. Serbian
domination of politics and the military influenced waves of secessionist movements. From 1991
through 1995, the SFRY fought Slovenia, Croatia, and Bosnia and Herzegovina as those states
broke away. The SFRY also had to contend with those states’ associated coethnic militias and
NATO. Only Macedonia
334
seceded without a violent response from the SFRY, though
Montenegro would later secede peacefully from the Federal Republic of Yugoslavia (FRY—the
successor to the SFRY).
335

214

What is most relevant to the case at hand are two conflicts that are part of the Yugoslav
Wars. The first is the 1998-1999 Kosovo War and its aftermath, in which the KLA and other rival
Albanian factions fought the FRY. Examined concurrently will be the 1999-2001 insurgency
involving the Liberation Army of Preševo, Medveđa and Bujanovac (UÇPMB) in southern Serbia
immediately adjacent to Kosovo. The second is the 2001 insurgency in Macedonia, in which the
NLA was the main faction, though rivals pledged to carry on the fight after the NLA sought peace
with Macedonia. Ethnic Albanian factions were supported by the state of Albania to varying
degrees. Each faction began by supporting unification with Albania but stopped pursuing this goal
over time due to pressure from the US.
The nucleus of the three UÇK factions begins with the People’s Movement of Kosovo
(LPK). The LPK formed in the 1980s as a Marxist-Leninist organization in Kosovo that supported
the Hoxha administration in Albania, with which it sought to unify.
336
Its main competition was
the centrist Democratic League of Kosovo (LDK). Like the LPK, it also sought to unify with
Albania until the late 1990s, when such an outcome appeared impossible to achieve.
The primary disagreement between the LPK and the LDK revolved around the use of
violence to pursue self-determination claims. The LDK under Ibrahim Rugova for a time chose to
avoid violence, while Hashim Thaçi of the LPK established the KLA to combat Yugoslav forces
in the early 1990s and achieve Kosovar independence.
337
The KLA was financially and logistically
supported by the state of Albania, the Albanian diaspora, and TOC. The latter took part through
distribution of illicit narcotics, weapons trafficking, and human smuggling.
338
It was trained by the
Albanian military and the Albanian intelligence service (SHIK, later SHISH),
339
the CIA, the
German BND, and the British SAS.
340
NATO forces established the Kosovo Force (KFOR), a
peacekeeping operation that functioned alongside the United Nations Interim Administration

215

Mission in Kosovo (UNMIK). NATO action against FRY forces benefitted the lightly armed
KLA’s operational capabilities, though NATO was not overly enamored of the KLA’s methods.
341

Following the Kosovo War, several of its key members entered politics. Thaçi for instance,
joined the Democratic Party of Kosovo (PDK). He became Prime Minister in 2007 and declared
independence from Serbia in 2008.
342
Others joined the Kosovo Police, or the Kosovo Protection
Corps (KPC/TMK), which is as a paramilitary emergency response service.
343

Other members of the LPK and KLA did not wish to stop pursuing Albanian irredentist
desires in Kosovo, however. Ethnic Albanians in southern Serbia were also repressed by the FRY
government. The UÇPMB was established by former KLA members in the Preševo Valley, which
is an Albanian-majority region adjacent to Kosovo, but not legally considered to be a part of it.
344

As this region was not officially part of Kosovo, the UÇPMB was not supported by NATO and
KFOR. Rather, KFOR actually disarmed the UÇPMB in support of the FRY forces.
Following the Kosovo War and insurgency in Serbia, members of the KLA and UÇPMB
formed the National Liberation Army in the mountainous northwest region of Macedonia inhabited
by ethnic Albanians. As with Kosovo, they first sought to unite with Albania. NLA troops briefly
took and held several cities in a short war with the Macedonian government before terms were
sought. The Ohrid Agreement of 2001 brought a ceasefire agreement and enhanced guarantees of
Albanian autonomy in Macedonia. While the NLA and other factions have sporadically attacked
Macedonian military and police forces since 2001, the state of Albania no longer provides them
with military assistance like they did in 2001.

216

6.3 Origins of Intra-Albanian conflict—or violence?
Simply put, intra-Albanian violence in the Yugoslav Wars was significantly less widespread in
comparison to any period intra-Kurdish conflict, which itself is not on the scale of intra-Chechen
conflict. As is the case with irredentist conflict in general, the most severe violence was
interethnic.
345

There were nevertheless incidents of intra-Albanian violence, if not organized armed
conflict. These appear to be exceedingly rare, however. I was able to find three georeferenced
incidents of it in the data, which are much too few to perform a statistical analysis of intraethnic
conflict. There are reports of KLA fighters abducting and killing political opponents and
individuals who allegedly collaborated with FRY forces in Prizren, Djakovica, and Klina, as well
as in prisons in Albania set up by the SHISH for use by the KLA.
346
A KLA leader named Sami
Lushtaku was tried for killing an ethnic Albanian, but no body or even the name of the alleged
victim could be produced.
347
It is possible that these incidents were motivated by control over the
illicit narcotics and arms trades, as well as personal vendettas and blood feuds, rather than out of
political rivalry.
One event that could be documented included the detention, abduction, torture, and
attempted assassination of LDK member Rudi Fetah. However, it was unclear if this action was
ordered (or even known of) by the LPK or was the work of fighters moonlighting as racketeers
who wanted material goods.
348
There were certainly tensions between the LPK and LDK, but they
and their leaders— Thaçi and Rugova—did also work together following the end of the Kosovo
War to finalize Kosovo’s post-war status. The armed forces of the LDK—Armed Forces of the
Republic of Kosovo (FARK)—did not engage in combat against the KLA.

217

The KLA and FARK both seemed to have experienced intermittent clan-based murders
resulting from blood feuds.
349
However, these did not extend to the organization, as it did with the
Chechen case. Additionally, there was no alliance sought with FRY forces in order to settle scores.
The events that could be identified also were not conclusively categorizable as organized
intraethnic conflict. In sum, there were insufficient intra-Albanian conflict events to perform an
analysis with regards to the objects of contestation (P2).
However, we may still examine the Origins of Conflict (P1) and Rival Cooperation (P3).
Beginning with P1, let us examine why there may have been intra-Albanian violence, limited as it
was. In brief, Competing Leaders (H1a) is the best supported, while State Manipulation (H1b) and
Divergent Development (H1c) appear to be unsupported.
Competing Leaders (H1a)
Violence between Albanians was strikingly limited relative to the other forms of intraethnic
conflict in this work. Motivations could be linked to the settling of scores, blood feuds, and the
desire to control illicit economic activities. The elimination of local rivals was an objective of
leaders that held first-level disputes with one another. However, there is no evidence that this rises
to the level of organized intraethnic conflict; many KLA members have been arrested for their role
in alleged disappearances by their own successor party, while many FARK fighters temporarily
joined the KLA to take part in fighting against FRY forces. Intra-Albanian violence occurred for
personal reasons, without rising to the level of organized conflict.
350

State Manipulation (H1b)
How about State Manipulation (H1b)? Could Serbia or Macedonia have influenced
violence among Albanians? It does not appear so. The LDK supported a peaceful resolution to the

218

Kosovar question—to the displeasure of the LPK—but they were not supported by Serbia. The
LDK established the FARK to combat Serbian forces once it became apparent that a peaceful
resolution was out of the question. The state of Macedonia did not have a pro-Macedonian force
of Albanians to help combat coethnics as the state of Russia did with regards to Chechen separatist
fighters.
Divergent Development (H1c)
Divergent Development (H1c) is the third explanation for the origins of violence between
Albanian factions. The presence of an Albanian nation-state kept ethnic Albanians in neighboring
states closely aligned with each other. The establishment of a Greater Albania remained the goal
for Albanian factions until Western persuasion induced the most prominent factions to drop the
aspiration.
351
However, doing so did not cause organized conflict between them.
Communism was the political and economic system that ethnic Albanians were familiar
with until the end of the Cold War. Deep ideological divides among factions did not arise across
state borders in such a brief period of time. Ethnic Albanian irredenta supported the Albanian state,
especially as they believed that it was the key to joining Albania.
352
Communism also does not
appear to have had lasting effects on religious affiliations; despite declarations by Hoxha that
Albania was an atheist state—and Yugoslavia’s repression of religious institutions—many ethnic
Albanians maintained their beliefs in secret.
353
Regardless, Albanian national identity remained
more salient than any one faith. Though a small majority of Albanians are Muslim, and some have
joined jihadist organizations, these factions were not affected by religious tensions nearly to the
degree that CSPs were.

219

Furthermore, religion was less of a defining factor even prior to communism with regards
to Albanian factions. It appears even less impactful now. Though there were offers of assistance
made by Islamist and Salafist forces, these were not accepted.
354
Albanian leaders decided that
their resistance to Serbian and Macedonian repression had to be nationalist in nature. It was to be
focused on self-determination. Having any part of their movement associated with global jihad
was viewed by AIFs as detrimental to the cause.
The state of Albania offered crucial support to the KLA, NLA, and UÇPMB. The
government during these conflicts preferred these UÇK groups over the LDK and the FARK
(except for an opposition minority
355
) and provided them with arms, training, safe havens,
intelligence, artillery support, armored support, and coordination with KFOR. Though direct
military intervention was limited to a handful of battles,
356
the state of Albania achieved its goals
without entering into a war with Serbia or Macedonia. There would be no incentive for them to
pursue intra-Albanian conflict.
Another system-level consideration is non-Albanian support. UNMIK, KFOR, NATO, the
US, and the UK wished to preserve the Kumanovo Treaty, which ended the Kosovo War but
excluded the Preševo Valley from being part of Kosovo. It was instead being part of a buffer zone
between Kosovo and the FRY.
357
As such, they were willing to support the FRY against the
UÇPMB in the Preševo Valley, even though they supported the same fighters when they were in
the KLA and NLA. Again, there is no incentive for intraethnic conflict on the part of UNMIK or
NATO.
The final non-Albanian actor of importance here is Germany. That state had a long history
of involvement with ethnic Albanians and alliances with Albanian nationalists. Indeed, the Serbian
government and groups in opposition to ethnic Albanian self-determination aspirations were quick

220

to remind the world of Albanian involvement in the Këshilla, the Skanderbeg, and the Handschar,
accusing many senior leaders of independence movements of being veterans of those units.
358
The
BND was primarily responsible for hand-picking the leadership of the KLA and NLA. However,
the German state appears to have supported cohesion among factions claiming to represent ethnic
Albanians. This brings us to the next section, which deals with cooperative action between AIFs.
6.4 Cooperation among rival(?) Albanian irredentist factions
Before we discuss the results of the statistical analyses on cooperative action, let us examine what
rivalry looks like with these factions. We are primarily talking about the LPK and its armed forces,
the KLA, as well as the LDK and its armed forces, the FARK. These factions were politically
opposed but did not engage in organized armed conflict with each other. There were clan-based
acts of violence between them. They did not officially work together against the forces of the FRY,
but many KLA fighters were actually FARK members that decided to fight the FRY under the
KLA banner to defend their islands of coethnic irredenta. There is no indication that FARK
members were encouraged to do this by the LDK, but they were not forbidden either. Thus, both
rivalry and cooperative action could be considered established.
The other aspect of rivalry here could be between the KLA, the NLA, and the UÇPMB.
Though technically all of these factions are UÇK offshoots, they competed with each other for
available personnel, for resources, and for control of illicit trade such as black-market cigarettes.
359

On the other hand, because there was so much overlap, the armed forces of these factions might
not always be considered separate groups. The NLA and the UÇPMB were established and led by
KLA fighters. KLA and FARK fighters also fought under NLA and UÇPMB banners. For these
reasons, and because the GED does not distinguish between UÇK offshoots, conflict events
involving UÇK-affiliates were considered cooperative.

221

Additionally, in some GED records, FARK forces were also considered part of the UÇK.
This was presumably because deceased FARK fighters were only identified following the conflict
event. They had been temporarily fighting as KLA, NLA, or UÇPMB fighters. The UÇK was
preferred by the majority of the legislature of the nation-state of Albania (as well as its intelligence
services), so FARK fighters could be conceivably expected to fight for UÇK groups as a means of
receiving material support, as well as for ideational motivations, such as nationalistic attachment
and emotional affect.
Let us now examine the results of tests for the third proposition. What effect did Out-group
Incursions (P3) have on cooperation between Albanian factions? We will examine the Who Dies
and Where They Die hypothesis (H3a), the Heroes and Rivals hypothesis (H3b), and the Non-
combatant Deaths hypothesis (H3c).
An overview of out-group incursions
The Serbian Assembly voted to revoke Kosovo’s autonomy in 1989. President Slobodan
Milošević presided over the repression of non-Serbs in Yugoslavia’s final days. The dissolution
of Yugoslavia saw the breakaway of Slovenia, Croatia, Bosnia, and Macedonia beginning in 1991.
Kosovo Albanians declared the establishment of the Republic of Kosova in 1991, claiming
independence in 1992 with Rugova (head of the LDK), as president. The KLA at first sought to
join Kosovo and other Albanian-inhabited territories in the FRY, Macedonia, and Montenegro to
Greater Albania. They pursued violent means to do so, attacking FRY police and military targets
beginning in 1995. The FRY responded by massacring ethnic Albanians and launching mortar and
artillery attacks on suspected KLA bases. Relatively low-intensity conflict escalated to war in
1998.

222

Incidents like the Gornje Obrinje massacre
360
—in which a family of ethnic Albanians were
tortured, mutilated, and murdered—caused disparate factions to join together and fight the FRY.
Artillery and armor support from the state of Albania, along with NATO bombing and the
deployment of KFOR peacekeeping troops, forced the FRY to sign the Kumanovo Treaty, which
allowed for Kosovo to be autonomous once more.
KLA veterans then joined the UÇPMB in the Preševo Valley eastwardly adjacent to
Kosovo in southern Serbia. Their goal was to append the region to Kosovo.
361
The FRY was this
time aided by NATO and KFOR, and the insurgency ended in 2001 with little loss of life and the
surrender of many UÇPMB fighters to KFOR. Those fighters who surrendered were quickly
released. They (and others who did not surrender) joined the NLA in Macedonia just south of
Kosovo. Repression of Albanians there had increased during the 1990s. In addition to preexisting
tensions, Kosovar refugees from the Kosovo War strained the Macedonian economy; created a
transit point for KLA fighters, which Macedonia did not appreciate; and added to a burgeoning
problem of illicit trade in drugs, weapons, contraband products, and humans.
Following the murder of an Albanian father and son in an incident at a checkpoint that
eerily foreshadowed the death of Saeed in the opening of this work, NLA fighters seized much of
Tetovë and its surroundings. The North Macedonian response was much more targeted than that
of the FRY in Kosovo, yet the Albanian factions remained cohesive, unlike in the cases of the
Kurdish and Chechen factions.
A ceasefire led to the Ohrid Agreement. Albanians in Macedonia would be given a wider
degree of political protections against the cultural repressions that led to the insurgency. Though
the NLA disbanded by 2002, some other factions—namely the Albanian National Army
(ANA/AKSh) whose political wing is the FBKSh—have vowed to continue the struggle of

223

building Greater Albania out of portions of North Macedonia, Kosovo, the Preševo Valley, and
Montenegro.
As these conflicts occurred either simultaneously or immediately following one another
using considerable overlap of personnel, cooperative action and out-group intervention were
calculated for all conflicts. The conflict areas and the sources of out-group intervention are
displayed below in figure 6.4:
Figure 6.4
External intervention into ethnic Albanian irredentist movements, 1998-2001

Note. Territorial control for the UÇK is approximate and collated from numerous sources.

224

Figure 6.4 contains the regions which witnessed intervention by out-groups and cooperative action
among AIFs. The graph bars indicate the number of those killed by out-groups. The kernel
indicates a rough density of cooperative events that will be examined further shortly.
Unsurprisingly, the most intense cooperation occurred closest to the Albanian border.
Assistance to either the UÇK factions or FARK fighters would be limited by the desire to avoid
an interstate war. Contrary to the results from the DSN case, cooperative action also took place in
areas that did not necessarily witness only a large number of unaligned deaths.
Preliminary evidence of the impact of unaligned deaths on rival factional cooperative action
Let us explore statistical results of the hypotheses under the Out-group Incursion
proposition (P3). The KDE displayed in figure 6.4 does not clearly indicate the centrality of
unaligned deaths for rival cooperation, and so the irredentist case potentially offers a degree of
variation that would be ideal for comparison to the DSN with the testing that followed.
The KDE was tested using K- and L-function analyses. The results of the K-function
suggested that random clusters of cooperative events do not become incorporated into the analysis
at all. Much of this is likely because of the comparatively small geographic size of this area relative
to the Kurdish case. It is also highly likely that this is due, in part, to NATO and KFOR assistance.
The use of air power in a relatively small region of interest predicts that cooperative clusters can
occur anywhere. Despite eliminating as many NATO- and KFOR-only incidents as possible during
pre-processing, it is impossible to do so if the source materials do not mention NATO or KFOR
forces, and so some may remain in the analysis. Additionally, K:A fighters may have received air
support not described in the source material.

225

Another important consideration involves the type of conflict at hand. Irredentist conflict
often features the intervention of powerful state actors. Kosovo was also territorially controlled by
the KLA, for the most part, with assistance provided by Albania and other state actors. Thus, ethnic
Albanian factions had much more freedom of movement than KPFs, which is typically much more
proscribed due to material asymmetry.
The L-function yielded broadly similar results, though cooperative clusters begin to trend
toward CSR after 24km. Like the case of the Kurds, 8km represented one point distant from the
confidence interval envelope for CSR, though 24km was the most distant. Unlike the case of the
Kurds, a state actor (Albania) offered artillery and armored support, which likely accounts for the
greater distance The KDE, the K-function, and the L-function are displayed in Appendix G, figures
G.2 and G.3.
Next, the ANN analysis with Monte Carlo simulation—along with its associated pseudo p-
value—yielded the expected result; observed values of cooperative events are significantly
different than spatially independent simulations. The results of this analysis are found in Appendix
G, figure G.4.
Poisson point process models
Finally, a non-stationary PPP with LRT tested whether cooperative events could be the
result of an underlying process. One model was the null hypothesis of there being no underlying
process that could explain cooperative action. The other three involved the covariates related to
the hypotheses in this work. These are the Who Dies and Where They Die hypothesis (H3a; Model
2); the Heroes or Traitors hypothesis (H3b; Model 3); and the Non-combatant Deaths hypothesis
(H3c; Model 4). Table 6.2 displays the results of these models below:

226

Table 6.2
Poisson point process modeling of Albanian rival cooperation, 1998-2001
Analysis of Deviance Table – Likelihood Ratio Test
_______________________________________________________
Model 1: ~ 1 (Null) Poisson
Model 2: ~ Unaligned Deaths Poisson
Model 3: ~ Rival Deaths Poisson
Model 4: ~ Non-Combatant Deaths Poisson
_______________________________________________________
Model Deviance Pr(>Chi) Zval
1 0
2 18.9599 1.335e-05 *** 5.517566 ***
3 16.765 4.232e-05 *** -5.251897 ***
4 17.204 3.358e-05 *** -5.339364 ***
_______________________________________________________
AIC: Start = 2839.36
- Unaligned Deaths = 2837.4
- Rival Deaths = 2910.9
- Non-Combatant Deaths = 2838.4
_______________________________________________________
Stepwise Selection:
AIC: 2837.37
_______________________________________________________
Model Zval
3 9.646138***
4 4.140825***
_______________________________________________________
Note. Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
_______________________________________________________

227

To begin, we see that the deviation from the null is strongly significant in each model. We may
confidently reject the null hypothesis, which states that cooperative action among Albanian
factions was not the result of an underlying spatial process.
Comparing covariate effects reveals an interesting initial result; the targeting of neutral
coethnics yields a very weak, negative effect (Zval) that is not statistically significant. A
significant, positive likelihood is found by targeting rivals, while a non-significant positive effect
appears to be the result of targeting all non-combatants.
Following stepwise selection, however, we develop a model with a slightly lower AIC.
This model drops alignment as a predictor of cooperative action. Targeting rivals and any non-
combatant yields a positive, significant effect on cooperative action. It seems the irredentist case
is most supportive of the Non-Combatant Deaths hypothesis (H3c), while refuting the Who Dies
and Where They Die (H3a) and the Heroes and Traitors (H3b) hypotheses. The expectation for
cooperation when non-combatants is vindicated as cooperative action between all ethnic Albanian
factions was coordinated in areas which could be lost to any faction, rather than selectively so as
we witnessed in the DSN case.
The targeting of all civilians was a salient predictor. It seems as though DIG factions are
less concerned about alignment than DSNs. This is not an unexpected result if we consider that
many coethnics would hold maintain and emotional attachment to group identity, especially as
there is already a nation-state comprised of the ethnic group that can force leaders to support the
wishes of the populace. Albania has an interest in maintaining a degree of unity if its military and
intelligence agencies are to coordinate action between them. By contrast, there is no central state
that can compel such cooperation when it comes to the DSN.

228

Cooperation also appears to be more lasting for the DIG than for the DSN. There were still
some instances of violence between the heirs to the LPK and LDK but those may have occurred
for personal reasons. Furthermore, Rugova and Thaci served in the same government without their
factions going to war.
Factional evidence
Factional sources and press releases from the time provide support to the statistical results
for the drivers of rival cooperative action. Factions of all stripes were motivated to protect the
movement of all non-combatant ethnic Albanians; this was acknowledged by KFOR officers who
viewed some “refugees” as people who would cross back and forth whether there was a conflict
or not.
362
Even after UÇK factions stopped trying to join ethnic Albanian-inhabited regions to
Albania to create Greater Albania,
363
they referenced the necessity to protect all Albanians from
oppression.
364

Political actors in the state of Albania also viewed the struggle of Kosovars as one of the
entire ethnic group, rather than just a region. Prime Minister Pandeli Majko said that “the Albanian
government at this moment feels itself very close to the fighters of the Kosova Liberation Army
[UCK], who are defending the nation” of the Albanian people, rather than just a subset of
Albanians.
365
Though not eager for an interstate war, the Albanian military prepared for such a
possibility by massing troops on the border with Kosovo and the FRY to defend ethnic
Albanians.
366
Though Tirana preferred to support the UÇK over other contenders, they were
willing to also provide assistance to UÇK’s rivals—the FARK—if it led to greater in-group
cohesion.
367

229

UÇK fighters such as Major General Gezim Ostreni referenced the necessity to protect all
citizens, regardless of affiliations such as political background in Kosovo and Macedonia.
368
When
UÇK fighters did commit violence against ethnic Albanians, it was not apparently for clear motives
of control over the ethnonationalist movement, but likely due to first-level reasons. In Macedonia,
the Albanian National Army (AKSH) attacked members of the disbanded NLA, though it seems
to have been due to a personal grudge between local leaders.
369
Once he became Prime Minister
of Kosovo, former KLA leader Thaçi was accused of committing violence against ethnic
Albanians that were believed to be collaborating with the Milosevic FRY government.
370
Another
view is that he was knowingly accusing rival mafiosi of collaborating, knowing they were not in,
fact, collaborating, in order to assume control of illicit trade in drugs, weapons, and illegally
harvested organs.
371
Interestingly, some of the accusations against Thaçi may also have been
motivated by individual-level desires for self-advancement. As described above, one senior KLA
member, Sami Lushtaku, was accused of murdering another Albanian in 1998, but the victim’s
name was unknown, and no corpse could be produced, thus illustrating the propensity to engage
in individual-level disputes in this case.
372

6.5 Summing up: Tentative findings from DIG violence and cooperation
This section sums up the findings as they relate to the Albanian case before it makes a comparison
to other cases. It also provides a brief introduction to potential implications, though these are more
thoroughly explored in Chapters Eight and Nine. The results for this case are presented below in
table 6.3:

230

Table 6.3
Summary of evidence for DIG conflict based on AIFs

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

P1 Roots of Rivalry
H1a Competing Leaders UÇK-AKSh Y Factional Moderate
material
UÇK-FARK Y Factional Moderate
material
H1b State Manipulation All factions N Statistical Strong
analysis,
factional
material
H1c Divergent All factions N Temporal Strong
Development analysis,
factional
material
P2 Effects of Distance
H2a National Capital NA
H2b Regional Center NA
H2c Unpacific Borders NA
H2d Primary Routes NA

P3 Out-group Incursion
H3a Who Dies and All factions N Statistical Moderate
Where They Die analysis,
factional
material

231

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H3b Heroes and Traitors All factions N Statistical Strong
analysis,
factional
material
H3c Non-combatant All factions N Statistical Strong
Deaths analysis,
factional
material

Note. Weak support: evidence was either statistical (*) or qualitative. Moderate support: either
statistical (*) and qualitative evidence or statistical (**). Strong support: both statistical
(** or ***) and qualitative evidence.

Let us begin by discussing the Roots of Rivalry (P1). While there was scant evidence for the
Competing Leaders hypothesis (H1a) being impactful for KPFs, it was somewhat impactful for
CSFs. It is particularly well-supported for AIFs. Though intraethnic violence was rare, it did occur.
Generally, it was due to first-level motivations. There were also several instances that were alleged
to occur but could not be substantiated. The accusations themselves suggest the utility for
individual-level explanations for violence within DIGs.
It is also possible that economic motivations came into play at this stage as many instances
of intra-Albanian conflict could have been due to competition over TOC revenue streams. This
presents an opportunity for further research: at what stage of intraethnic competition (if not
organized conflict or warfare) should we incorporate so-called greed explanations? DIG violence
occurs at a much more advanced stage of state-making than DSN conflict. A nation-state for the
group already exists, which makes the range of possible behaviors exhibited by irredenta more

232

restricted than what could be possible for DSNs. So, we might expect economic motivators to
become more salient for DSN conflict further down the line. Perhaps this is more likely to occur
if there is a single nation-state than two competing nation-states.
State Manipulation (H1b) seems to be uncompelling, similar to DSN conflict and sharply
divergent from SSS conflict. Having populations of coethnics inhabiting neighboring states while
there is already a nation-state of their ethnic kin generates an entirely different bargaining process
for irredenta than for SSSs or DSNs. Furthermore, allegations of collaboration with the FRY seem
instead to be motivated by personal animosity or desires to edge out economic rivals. This facet of
intra-Albanian conflict is better explained by H1a.
Divergent Development (H1c) does not appear to influence intra-Albanian conflict, but
rather cooperative potential. The existence of an Albanian nation-state influenced the maintenance
of political, economic, and socio-cultural institutions among coethnics across state boundaries in
ways that were stronger than for the DSN. Though visions of a Greater Albania faded over time,
the state of Albania nevertheless provided assistance to its irredenta to achieve greater political
rights for ethnic Albanians. Without a DSN nation-state, similar efforts will not garner the same
domestic or international support, thus making intraethnic conflict for in-group dominance the
likelier outcome.
There was insufficient data to perform a statistical analysis of the Effects of Distance
proposition (P2), which in and of itself provides a stark contrast to DSN and SSS conflict. Violence
was very rare—and not quite rising to the level of organized conflict. Once again, it is likely the
presence of a nation-state comprised of the group that influences this. All ethnic Albanian factions
were dependent upon the Albanian military and intelligence services for equipment, training,

233

funding, and support. As a result, intraethnic fighting would have been strongly discouraged in
ways that are simply not possible for DSN and SSS factions.
Finally, let us examine the Out-Group Incursion proposition (P3). Who Dies and Where
They Die (H3a) was not apparently a driving force for DIG cooperation as it was selected out of
the PPP model to make it more parsimonious. Also, The Heroes and Traitors hypothesis (H3b)
may be rejected as deaths of rivals still yielded cooperation, nonetheless. The Non-combatant
Deaths hypothesis (H3c) was supported, in contrast to DSNs and SSSs.
These results are quite interesting indeed. DSNs and DIGs are the only two types of
intraethnic violence examined here that witnessed terminations of conflict in order to cooperate.
However, both types are very different from each other when taking into account the motivations
for doing so. DIGs are much closer to what we would expect from the first generation of literature
which highlights emotional affect, which was explicitly referenced by Prime Minister Majko and
Major General Ostreni, as described earlier. Irredenta were to be protected regardless of affiliation,
unlike the selective extension of support that we saw with KPFs. AIFs also did not cynically offer
assistance to gain institutional leverage over a rival. Instead, they cooperated to protect all ethnic
Albanians. That dynamic was very much unlike KPFs.
So far, we have discovered some clear differences among types of intraethnic conflict and
violence. Let us now examine the last variety, the TSON. KSRs represent this final arrangement
of two states comprised of the same ethnic group divided by borders. Following the seventh chapter
we shall then look at the results collectively in Chapter Eight, before concluding with implications
for our continued study of this phenomenon in Chapter Nine.

234

Chapter Seven
Two States, One Nation: Conflict between Korean State Regimes?

Unlike other arrangements of the two states, one nation (TSON) phenomenon such as Vietnam
and Germany, North Korea and South Korea did not reunite during or after the Cold War. North
Vietnam’s incorporated South Vietnam through conquest. East Germany (DDR) and West
Germany (BRD) reunited. The DDR was absorbed peacefully, though on the terms of the BRD.
Furthermore, North Korea and South Korea do not have questions of sovereignty and statehood,
as we see with the Republic of China (ROC), or Taiwan, which make up the remaining surviving
dyadic rivalry. The ROC and the People’s Republic of China (PRC) are also separated by water.
These reasons make the Korean state regimes (KSRs) of the North Korea-South Korea dyad the
ideal candidate for study here.
Though divided and engaging in state-making activities, TSONs are vastly different from
DSNs simply by having internationally recognized state frameworks. Conflict is much rarer and
smaller in scale in the post-Cold War era (though the scale was enormous when interstate warfare
occurred during the Cold War). The stakes may be higher, however, given the greater levels of
military capacity and conflicting alliances that may be drawn into a possible interstate war. Unlike
DSNs, there is no cooperation meant to eject outsiders from the territory of the TSON in this case.
These differences are results of the same system-level sources, however. Transnational
linkages induce divergent political, economic, and socio-cultural development which make
eventual reunification at least as difficult as unification for DSNs. Like the DSN, individual-level
factors are relatively unimpactful. State-level factors such as regime type and economic system

235

are of course at odds, but there is no single state framework (currently that is—each KSR has tried
to create one) that can manipulate KSRs.
This chapter proceeds in six sections. The history of the Korean people is presented in
Section 7.1. Post-Cold War rivalry is described in Section 7.2. The more recent contributions to
this protracted rivalry are examined in Section 7.3. Conflict and cooperation among KSRs are
discussed in Sections 7.4 and 7.5. Section 7.6 concludes by systematically comparing the
phenomena of the TSON and the DSN.
7.1 Historical overview of the Korean people: Resistance to and adoption of outside
influence, and the precedence of north-south division
There are roughly 83 million Koreans in the world, with about 50 million in South Korea
and 25 million in North Korea. There are two other contiguous and autochthonous populations.
The first is in the Yanbian Korean Autonomous Prefecture (YKAP) of China. This region of
Manchuria is adjacent to North Korea in northeast China and is home to another 2.5 million
Koreans. Nearly 200,000 additional Koreans live in Russia’s Primorsky Krai, which is also
adjacent to North Korea.
There is also a sizeable Korean diaspora. Nearly 300,000 Koreans are descended from
those deported by the Soviet Union to Uzbekistan and Kazakhstan in the Second World War. The
USSR was wary of Koreans in the Soviet Far East possibly assisting a feared Japanese invasion.
373

Nearly 1,000,000 people in Japan are descended from Koreans brought to Japan during the same
period. There are also nearly 3,000,000 in North America, and many more around the world.
The word “Korean” is derived from the name of the Goryu dynasty that ruled in the 10
th

through the 14
th
century. Korean people in Russia and Central Asia use this appellation (“Koryo-

236

saram”), but those in North and South Korea use other terms to refer to themselves. In South Korea,
ethnonyms are related to the state (such as “Hanguk-saram” or people from the Korean country.
Ethnic designators like “Joseon-saram” or ethnic Korean people predominate in North Korea.
374

Korean and Korean-descended peoples in other states, including the diaspora, frequently have
other variations of one of the latter two types of ethnonyms. The “Joseon” variation is more
widespread among Korean populations that were resettled in the first half of the 20
th
century and
the “Hanguk” variation is more widespread among more newly established diasporas.
375

The Korean language has multiple standardized forms. The written alphabet is called
Hangul in South Korea and Chosŏn'gŭl in North Korea. Various Korean dialects may technically
count as separate languages, but this notion is contentious.
376
The Koreanic language family is
likely an isolate, but may be related to the Altaic family, specifically Manchurian Tungusic
languages.
377

Whether or not the Korean language is related to Manchurian, ancient proto-Korean
peoples were genetically related to proto-Manchurian and proto-Mongol peoples. They originated
in what is now Manchuria.
378
As they moved south into the Korean peninsula, proto-Koreans
assimilated groups that were ancestral to the northern Han Chinese and the Yamato Japanese
peoples.
379

Agriculture and metallurgy reached the peninsula in the Neolithic and Bronze Age, which
contributed to the creation of the first organized Korean civilization by 1500 BC.
380
Chinese
records tell of two early Korean kingdoms: Gojoseon in the north of the peninsula, and Jin in the
south.
381
The creational account of these kingdoms is described in the Samguk yusa, which is a
collection of foundational myths and accounts of early Korean history, including the establishment
of Gojoseon by the half-tiger, half-bear demigod Dangun Wanggeom.
382

237

The Samguk yusa also describes the end of Gojoseon and another Korean kingdom called
Gika Joseon. The latter alternatively coexisted with Gojoseon or replaced it. Wi Man, a military
leader from the Yan dynasty, seized power in 194 BC and established the Wiman Joseon dynasty
that lasted nearly a century.
383
This period came to an end when the Han dynasty of China
conquered the peninsula and established a number of commanderies. These commanderies were
themselves succeeded by other small states until the advent of the Three Kingdoms Period lasting
from 57 BC to 668 AD. The kingdoms were Silla and Baekje in the south,
384
and Goguryeo (or
Goryeo) in the north, from which we get the word “Korea.”
It was during this period that ties with China brought both Confucianism and Buddhism to
the peninsula. Rulers sought to replace indigenous Korean shamanic practices with belief systems
that were more conducive to the centralization of authority as a means of legitimizing their claims
to power.
385
Though Confucian thought was more influential in the very early Three Kingdoms
Period, Buddhism gradually became the dominant belief system, though Confucianism did not
disappear entirely.
386

The Three Kingdoms Period ended in the seventh century when Silla and their Chinese
allies conquered the other kingdoms. Silla then promptly drove out the Chinese forces. However,
a new north-south divide occurred almost immediately following this brief period of enforced
unity. The North-South States Period consisted in part of Balhae to the north, which was
established by rebels from Goguryeo and was supported by Turkic tribes in Manchuria.
387
Silla,
the state in the south, maintained cultural ties to China despite earlier warfare between the two
states.
Silla and a significant portion of Balhae were incorporated into the kingdom of Goryeo in
the tenth century. Goryeo lasted until 1392 and functioned primarily as a military dictatorship.
388

238

Invasions by Khitans and Mongols forced the kingdom to swear fealty to Ögedei Khan in the 13
th

century. The Mongol Yuan dynasty of China established by Kublai Khan allowed Goryeo regain
some of its autonomy However, when the Ming dynasty replaced the Yuan, King Gongmin
attempted to excise any Mongol influence in Goryeo and retake Korean-inhabited regions north of
the peninsula.
Following a series of usurpations and assassinations late in the Goryeo period, general Yi
Seong-gye founded the Joseon dynasty in 1392.
389
Neo-Confucian thought would displace
Buddhism relatively early on in the Joseon period. Part of this was due to the power of China and
Joseon’s proximity to the larger state. By participating in a tribute system in which Joseon paid
homage to Chinese cultural influence, Joseon was able to rank higher in the East Asian hierarchy
than more recalcitrant states and tribes in the Chinese sphere of influence.
390
This provided
enhanced prestige and prosperity to Joseon, but also made them targets for Manchu and Japanese
invasions in the 16
th
and 17
th
centuries. However, Joseon successfully defeated the invaders, and
those victories were instrumental to the foundation of the modern Korean nation.
391

Though the aftermath of these invasions saw peace with neighboring states, Joseon
experienced domestic turmoil and factional rivalry. There were religious issues as well,
particularly in the 19
th
century. Christianity had become popular among the lower classes but was
perceived as a threat to the ruling classes because of its egalitarian philosophy.
392
Additionally,
there was a Neo-Confucian reformist movement called the Donghak movement, which was
violently suppressed by Buddhist and Neo-Confucian traditionalists.
393

Joseon also became increasingly isolationist as Western powers entered the region. It was
known as the Hermit Kingdom during the late 19
th
century but broke away from the Chinese
tributary system and declared itself the Korean Empire in 1897. This drive toward self-rule and

239

isolation was one result of the Qing defeat at the hands of the Meiji forces in the First Sino-
Japanese War.
394

Russia and Japan became the primary sources of external influence on the Empire. Japan
followed up its victory in the Russo-Japanese War by claiming Korea as a protectorate in 1905.
Korea was then formally annexed by Japan in 1910.
395
Resistance movements under the
government-in-exile based in Shanghai—the Korean Provisional Government (KPG)—fought
against Japanese occupation. The KPG grew out of the Korean National Association (KNA)
independence movement. Armed forces, namely the Korean Volunteer Corps (KVC) and the
Korean Liberation Army (KLA), were at times supported by the Kuomintang (KMT), the USSR,
the US, and France.
The Japanese were defeated in 1945 and the Korean peninsula was administered by the
USSR and the US, with the latter taking the area south of the 38
th
parallel and the former taking
the area north of it. The distribution of the Korean population at the time of division is displayed
below in figure 7.1 and table 7.1:

240

Figure 7.1
Distribution of the Korean population, 1946-2020

Note. The demilitarized zone is included to give sense of the regional and temporal context.

241

Table 7.1
Temporal analysis: Change in Korean-inhabited territory, 1946-2020

Year Total area Unchanged area Area lost Area gained Net gain
km
2
km
2
km
2
km
2
km
2

(%) (%) (%) (%)

1946 256,416 NA NA NA NA

Note. Km
2
have been rounded to the nearest whole number.

As we see in figure 7.1, Koreans make up the majority of the population in very minute
portions of China and Russia that are immediately adjacent to North Korea. The distribution of
Korean preponderance has not meaningfully changed since slightly before the imposition of
modern borders, as seen in figure 7.1 and table 7.1. There were nevertheless deportations of
Koreans in the 19th and early 20th centuries. However, unlike in the cases of the Kurds and
Albanians, these deportations were not to geographically contiguous regions. For instance, the
Soviet Union in 1937 removed tens of thousands of Koreans from the Soviet Far East, primarily
from what is now Primorsky Krai. These Koreans were sent to Kazakhstan and Uzbekistan, where
they gradually adopted local languages and cultural practices alongside of their own. By contrast,
Albanians and Kurds were deported to areas that were geographically contiguous and already
contained coethnics where they could maintain their identity and customs with greater ease.Of the
populations contiguous to the Korean peninsula, the largest is found mostly in an autonomous
prefecture in China. However, even in the YKAP, ethnic Koreans are actually a minority
population relative to Han Chinese.
396
Migration of ethnic Koreans from China to South Korea has
further decreased their numbers in both relative and absolute terms.
397

242

Diaspora populations are also located in regions that are in the same region but are not
immediately contiguous. The nearest diaspora of any size is in Japan. However, it is difficult for
Koreans living in Japan to obtain Japanese citizenship, and they are not viewed as an overly
influential population in that state.
398

There are desires for a Greater Korea that would unite all connected Korean-inhabited
regions, which include the entirety of the peninsula, the bulk of Manchuria, the YKAP, and parts
of Primorsky Krai. There are no movements with considerable mass appeal, however. Unlike those
of divided irredentists with a single nation-state representative that seriously contemplate the
creation of an ethnically contiguous superstate, Korean irredentism is at present more of a fringe
belief.
399

7.2 A depiction of rivalry between Korean state regimes
In the final days of the Second World War, Soviet and American forces occupied Korea in haste,
racing to take as much territory as possible. Without consulting Koreans or anyone else, Colonels
Bonesteel and Rusk used a National Geographic map to arrive at the 38
th
parallel as a demarcating
line between a proposed division of Koreans into two states. This was accepted by the Soviets,
though subsequent proposals by the newly fractured Allied Powers for eventual Korean unification
and independence were not as successful.
400

Division and the establishment of rival regimes
The Soviets supported Kim Il-Sung, a communist guerrilla commander who fought the
Japanese as a member of the Chinese Communist Party (CCP/CPC) before joining the Soviet Red
Army. They then placed him in charge of the Korean Communist Party (KCP) at the conclusion
of the war.
401
Though there were other competitors for the leadership of the nascent Soviet-

243

sponsored state,
402
Kim was able to consolidate his position through leadership of the Korean
People’s Army (KPA) and his cult of personality. The KCP also merged with other communist
parties to form the Workers’ Party of Korea (WPK), of which he was the chairman.
403

The Democratic People's Republic of Korea (DPRK) was established in 1948. The North
Korean state is a single-party totalitarian system. Its state ideology was implemented by Kim Il-
sung (1948-1994). It was further developed by his son Kim Jong-il (1994-2011), and grandson
Kim Jong-un (2011-present). The state ideology is known as Juche. A policy corollary of sorts,
Songun, has also arisen. The former refers to the pursuit of socialistic autarky, with a sustained
revolutionary independence movement guided by hereditary Great Leaders. The latter is the
prioritization of military power to guarantee independence from outside influences.
404

Though referred to as the Hermit Kingdom in 2009 by then-US Secretary of State Hillary
Clinton, the DPRK has developed bilateral diplomatic relations with some 170 states, state-like
entities, and supranational organizations. During the Cold War, however, it maintained relations
almost entirely with other communist states. It did not side with either the USSR or China during
the Sino-Soviet split of the 1950s and 1960s.
405
Both states were essential military backers of North
Korea. East Germany was its primary economic partner.
South Korea, or the Republic of Korea (ROK), was also founded in 1948. The US and
NATO supported Syngman Rhee, a Korean nationalist and anti-communist who had once led the
KPG. Rhee was a convert to Christianity and had earned a PhD from the Department of History,
Politics, and Economics at Princeton University.
406
Rhee was elected the first president of the ROK
under elections overseen by the United Nations Temporary Commission on Korea (UNTCOK) in
1948.

244

The nominally democratic
407
ROK experienced an economic boom beginning in the 1960s,
which is regarded as a miracle of state capitalism following extreme poverty that persisted for
some time after its founding.
408
With its emphasis on export-based trade, rigorous educational
system, and strong external support, the economy of the ROK has far outstripped that of the DPRK
since the 1970s. Though once wealthier than the ROK, the poor incentive structure inherent to the
economy of the DPRK—along with the cumbersome central planning system and decline of its
closest allies—was unable to match the advantages that the ROK developed in the long term.
Furthermore, famine engulfed the North when food aid was disrupted by the Soviet Union’s
collapse. Though an economically growing China began exporting more food to the North, it also
sought to cultivate economic ties to the South.
409

Let us discuss the international relations of the ROK. The ROK reached economic
rapprochement with China, which is now its largest trading partner. However, the ROK has not
returned to China’s geopolitical orbit as was the case during the hierarchical tribute system in East
Asian history. The ROK and Japan share a tense history over the legacy of colonialism and wartime
atrocities.
410
Japan is viewed negatively by many Koreans; indeed, Chinese military buildup is
viewed as less suspicious than Japan’s military development by most Koreans.
411
Despite these
tensions, the two states are military allies because of US encouragement and shared fear of the
DPRK and Chinese expansion. They created the bilateral General Security of Military Information
Agreement (GSOMIA) to monitor North Korean nuclear aspirations. The GSOMIA is an
intelligence-sharing organization that has nearly been abandoned over the tensions regarding the
two states’ histories and opposing stances with regards to the memories of past abuses.
The US is the ROK’s primary ally. Though the perception of the US and Americans has
seen antagonistic periods, the military alliance between the two states is robust.
412
The US has

245

been an influential actor in the ROK’s international relations, more so than any other state. A major
ally throughout the Cold War, the US has long maintained a sizable military presence in the
ROK.
413

Despite the strength of the US-ROK military alliance, there are issues between the two
states. The recent US deployment of the Terminal High Altitude Area Defense (THAAD) anti-
ballistic missile system in the ROK has been a point of tension between the ROK, China, and the
US. Though unpopular with Koreans, the ROK state wished to placate the US by accepting
THAAD. Complaints from the US are derived from the ROK’s economic ties to China and possible
violations of intellectual property rights laws.
With regards to the socio-cultural divide, a major point of contrast is found in religion
within the ROK and DPRK. A majority of the inhabitants of both states are non-religious.
Shamanism and Chondoism are more frequent in the DPRK. In the ROK, Christianity and
Buddhism are more prevalent. ROK society is more religious on the whole than the DPRK.
The rivalry between these states is long-standing. It presents itself in these state-level
distinctions, which are themselves the products of the system-level forces of external division and
dual sponsorship by rival superpowers.
414

The Korean War and attempts at reconciliation
Soon after the founding of both states, the North invaded the South in 1950, starting the
Korean War. This would set off a long history of instances of armed conflict that would occur
throughout the Cold War.
415

The Korean War was fought by North Korea’s KPA and the Chinese People’s Liberation
Army (PLA) and People’s Volunteer Army (PVA) on their side, with the UN on the side of the

246

South’s forces. Seoul was the first and most consequential goal for the KPA, which it seized three
days into the conflict. US intervention kept the South from being completely conquered in the
following week.
For the first few battles, the combined US-ROK forces were unable to dislodge the KPA.
Their forces were numerically small and ill-equipped due to budget cuts implemented by the
Truman administration prior to the war. However, the subsequent battles at Incheon, Pusan, and
the Chosin Reservoir saw US-ROK forces obliterate vastly numerically superior KPA-PVA
forces.
416
They retook Seoul in September of 1950. The KPA presence in the rest of the South was
then cleared.
Fighting centered around the 38
th
parallel for the remainder of the war, as it became the de
facto demarcation line between the two states, and the war ground to a stalemate. The South had
been devastated from the initial invasion. The North was then devastated by UN bombing. China
lost an enormous number of men. Neither the DPRK nor the ROK was capable of seizing the
capitals of their rivals after external intervention took place, nor could they dislodge their foe from
the border. And so, an armistice was negotiated in 1953.
417

The Korean Demilitarized Zone (DMZ) was established as a buffer between the two states
as part of the armistice. Though not on the 38
th
parallel, it runs near it and can be seen in figure 7.1
above. The DMZ also contains a Joint Security Area (JSA). The JSA is in the western portion of
the DMZ. Most negotiations between the DPRK and the ROK take place at the JSA.
This was not the end of armed conflict between the two factions of this TSON, however.
418

A guerrilla campaign in parts of the South continued for a time. A propaganda war that saw the
use of balloons dropping leaflets also occurred. Clashes between the North and South over the

247

DMZ saw hundreds killed on each side between 1966 and 1969. A DPRK attempt to assassinate
ROK President Park Chung-hee in an incident called the Blue House Raid, as well as the DPRK’s
capture of the USS Pueblo also occurred during this period.
More unsuccessful assassination attempts occurred in the 1970s and 1980s, with each side
trying to kill key figures of the rival regime. The murder of American military officers at the DMZ,
and the DPRK’s kidnapping of South Korean and Japanese citizens also raised tensions between
the two states up until the end of the Cold War.
Despite the occurrence of highly destructive armed conflicts, there were also attempts at
reconciliation, even reunification, during the Cold War. In 1972, both states issued the Joint
Communiqué, in which each pledged to work toward peaceful reunification without external
interference.
419
Negotiation attempts in the 1980s were disrupted by the DPRK’s attempt to kill
ROK President Chun Doo-hwan.
The North once believed it could unite the peninsula under its leadership because of its
economic strength founded on an abundance of resources, as well as its military power.
420
Thus,
the North wished to form a confederation with the South so that it could influence its inhabitants
to prefer the political institutions of the North and see it as the leader of the Korean people.
Since the eclipse of the North by the South, the ROK has viewed the framework approach
as working to its advantage. The end of the Cold War saw the DPRK come to fear that the ROK,
Japan, and the US were planning a takeover of the DPRK by the ROK, in a manner similar to the
absorption of the DDR by the BRD. We will return to these post-Cold War reunification strategies
when examining rival cooperation, but for low let us examine the development of these conflicting
approaches.

248

7.3 Examining the origins of Korean rivalry
The two rival KSRs both view the US as the driver of the peninsula’s division.
421
Since the end of
the Cold War, there have been too few conflictual and cooperative events to perform a statistical
analysis. This is expected, as they are both sovereign states; interstate violence is rare in
comparison to violence involving non-state actors in the post-Cold War world. While there will
not be a statistical analysis, we may still examine the three propositions of this work empirically.
Let us begin with an examination of the hypotheses associated with the Roots of Rivalry
(P1). Briefly, the Divergent Development (H1c) hypothesis is the best-supported explanation.
Competing Leaders (H1a) becomes less compelling across time as the Kims are increasingly
shrouded in mystery and experience less adversarial relations with their counterparts. State
Manipulation (H1b) also fails to satisfy, in part because both rivals are state actors, in contrast to
the other three types, but also because their proposed suprastate frameworks face such hurdles to
implementation.
Competing Leaders (H1a)
Clearly, there were tensions between the Kims and various leaders of the ROK. Numerous
assassination attempts were made or planned throughout the Cold War. These were likely to have
been the result of system-level pressures, though. The ability to kill the leader of the rival would
do much to demonstrate the superior capabilities of the regime.
Since the end of the Cold War, there have been numerous inter-Korean talks in which Kim
Jong-il, Kim Jong-un, and their ROK counterparts have met. Assassination attempts on ROK
leaders
422
have also ceased. That said, very little is known about the Kims, even by the ROK. For

249

instance, the true birthdays of members of the family are uncertain to the ROK and its intelligence
services.
The lack of interaction between the leaders of the two KSRs—aside from occasional
summits to discuss issues of geopolitical stability—makes it highly unlikely that personal disputes
explain the origins of the present rivalry. Even going back to the division of the peninsula, leaders
were selected based on their utility to the US and the USSR, rather than instrumentalized for their
preexisting hatreds, as we saw with the SSS case of Chechen factions.
State Manipulation (H1b)
We turn now to State Manipulation (H1b). There is no singular host state to examine as
manipulating the KSRs. Both the DPRK and the ROK are sovereign states, even if the KSRs
compete for group leadership. However, we may yet examine the attempts to create a confederation
as manipulation—if that confederation can be a type of superstate that could manipulate both
KSRs. This is because the KSRs would conceivably become more akin to factional parties beneath
that framework over time achieved. The window would be brief, however, as final reunification
would morph intraethnic conflict into a standard civil war over control of a state in its current form,
which would then rendering it outside of the state-making conditions of this work.
The North has wished to create a Democratic Federal Republic of Koryo (DFRK) as a
means of preserving its system and perhaps reunify under conditions more favorable to the DPRK.
The ROK views the DFRK as a means of turning the ROK into a communist state over time while
avoiding a direct military confrontation.
423

The ROK is presumed to be the eventual victor of any inter-Korean war,
424
but it too prefers
to reunify peacefully. The framework favored by the ROK is the Korean National Community

250

(KNC) model. The KNC would be a vehicle for eventual reunification of the peninsula that would
gradually standardize the political and economic systems of the two states.
425
The DPRK worries
that the greater wealth of the ROK will subsume the DPRK’s institutions.
Plans for implementing the DFRK or the KNC frameworks have become less ambitious
over time as trust between the two states remains difficult to build. And so, a new proposal for
federation—based on centuries-old views on Korea’s role in the tribute system—has been made:
the neutralization-unification strategy. Essentially, both KSRs would renounce nuclear weapons;
declare neutrality; work together economically while preserving political differences; and receive
the guarantee of non-intervention by the major interested powers of the US, China, Russia, and
Japan. They would then unify once mutual trust has been established. However, that very trust, as
well as political will, are the main impediments to this strategy.
426

As these three strategies have yet to come to fruition, we can consider State Manipulation
to not be a convincing explanation for intra-Korean conflict. Even working toward a superstate
framework has been untenable. Obviously, state-level features like regime type, economic
systems, and social spheres of life are conflictual, but what caused the differences to arise in the
first place? To answer this question, we must look at the system-level explanation offered by
Divergent Development (H1c).
Divergent Development (H1c)
The division of the peninsula had its roots in Japanese colonialism and Japan’s conflicts
with China and Russia. The final official partition was carried out by the US and the USSR with
minimal input from Koreans. The nationalist, state capitalist, and socially conservative vision of
the early ROK held by Rhee was supported by the US. The USSR was the benefactor of the

251

communist DPRK. Though there were rival factions even within these nascent states, their external
backers supported their elimination so that their ideological allies could consolidate their control
over the respective KSR.
427
In a sense, the KSRs became outgrowths of their Cold War
benefactors, with distinctly Korean interpretations of their competing systems. Nevertheless,
leadership of the Korean people remained the goal of both KSRs. This was reflected especially
powerfully in each KSR’s fear of the other using the framework of reunification as a means of
taking over the other.
428

Furthermore, a vast amount of the pressure to participate in inter-Korean talks comes from
regional and international players—so much so that in the absence of the USSR, both the ROK
and the DPRK view the US as the primary reason the peninsula is divided, as well as the primary
actor that should be responsible for its reunification.
429

The US for its part, wishes to unify the peninsula under the ROK; too much compromise
favoring the North results in breakdowns of peace talks.
430
Other system-level actors have their
own preferences for whether the peninsula is unified or not, and on whose terms. Russia and Japan
prefer the status quo, as a unified Korea will force them to recalibrate their respective grand
strategies. China remains politically allied with the DPRK, but favors its economic relationship
with the ROK, thus complicating their stance on unification.
The KSRs are dependent on regional powers for support in any potential conflict to achieve
group leadership. This dependence carries over to their desires to reunify peacefully; regardless of
what the KSRs want, any framework ultimately depends on the acquiescence of external system-
level actors. The contestation of group leadership was created by system-level actors through the
establishment of competing and irreconcilable political, economic, and socio-cultural realms in

252

each KSR. Furthermore, system-level actors help sustain intra-Korean conflict through their
various forms of intervention in the peace process.
System-level factors are clearly the most convincing when it comes to the Roots of Rivalry
(P1). The post-Cold War intra-Korean conflict and complications for intervention will now be
explored further in the following sections.
7.4 Examining post-Cold War conflict and cooperation among KSRs
There have been too few intra-Korean conflict events since the end of the Cold War to perform a
quantitative analysis, but we can still examine these events qualitatively to look at what types of
conflict goals may be important. Since the end of the Cold war, there have been 52 fatalities
incurred across fourteen incidents, according to the UCDP. As expected, many of them occurred
near the border. Though none occurred in Seoul, many of these incidents were linked to one
another and displayed a trajectory that pointed toward Seoul.
An overview of post-Cold War intra-Korean conflict events
Several other incidents recorded at the border involved exchanges of fire with no reported
injuries. A handful of additional incidents outside of the GED were discovered when performing
research for this case. However, some could not definitively be proven to be true conflict events.
They may well involve patrols belong to one KSR or the other accidentally entering a zone they
were not supposed to enter. Transgressing patrols generally retreated when warning shots were
fired. Most of these additional incidents also occurred at geographically uncertain locations, and
thus could not be used to augment the GED due to relatively poor levels of spatial confidence.
Four related incidents were naval battles near the islands of Baengyeongdo and
Yeonpyeong in 1999 and 2010. The First Battle of Yeonpyeong concerned disputes over the

253

maritime border between the two states—the Northern Limit Line (NLL), with North Korean
maritime incursions approaching Seoul, which is accessible by waterways if following the NLL.
The DPRK bombarded Yeonpyeong in 2010. Two incidents involving minor skirmishes between
naval vessels also occurred between these events. Though these incidents arose out of each side
trying to intimidate the other through a show of naval force, both sides characterized them as
unprovoked affronts to the Korean people.
431

The remaining incidents involved submersible crafts belonging to the DPRK entering ROK
territory and the subsequent ROK hunts for these crafts and their crews in 1998. One submersible
was found south of the peninsula, near Geoje Island, while the rest were in the northeastern portion
of the ROK. The final incident involved the discovery and killing of one DPRK spy and the capture
of another in Buyeo, where two ROK police officers were also killed. The conflict clusters are
displayed below in figure 7.2:

254

Figure 7.2
Inter-Korean conflict events, 1991-2020

Note. Credit: Uppsala Conflict Data Program. The numbers are fatalities in conflict clusters.

255

It is possible that the national capital could still be considered the object of contestation
among TSONs as demonstrated by the KSRs. Obviously, Seoul would be attacked in case of war.
However, even the border skirmishes shown in figure 7.2 tend to occur near the portion of the
DMZ in proximity to Seoul. Though a DPRK patrol is not going to storm Seoul, probing the
defenses near the intended object of contestation has utility for a military force. And so, the
National Capital (H2a) is likely to be more important in the event of TSON war than the Regional
Center (H2B). The major center of gravity is found in the capital, and thus, like any interstate war,
it will likely be the first focal point of war.
432

It is apparent from figure 7.2 that the border is the most contested feature among KSR
conflicts, as we would expect given the advanced nature of state militaries; the DMZ and the NLL
provide chokepoints for state militaries to hold off their rivals. The borders are also strictly
controlled. This explains why infiltration through submersible crafts was utilized to maneuver
around the hard targets located at the DMZ and NLL. The Unpacific Border (H2c) hypothesis
appears supported by these incidents.
As there are an insufficient number of incidents, we can only speculate as to the importance
of Primary Routes (H2d). As with any interstate war, it is assumed they would be of significance,
but they appear not to be overly impactful at this stage. Infiltration through waterways and insertion
of spies are more useful for interstate rivalry when the opponent controls its territory at this stage
in the conflict between these TSON factions. Reasons why this may be the case are addressed in
section 7.6 and in Chapters Eight and Nine.

256

7.5 Cooperation against outsiders?
There have been cooperative efforts between the two KSRs. However, these do not involve uniting
to eject outsiders. This section will instead cover these cooperative efforts to demonstrate the
system-level constraints on unity.
Even during the Cold War there were some cooperative events aside from the inter-Korean
talks, though these were few in number and limited in scale. The DPRK sent emergency aid to the
ROK following floods in 1984, though the direction of the flow of aid has since shifted from the
South to the North.
433
At the end of the Cold War, the two Koreas began competing in some
international sporting events and civic competitions with a single team, though attendees could be
jailed if they did not receive approval by their government before participating.
434

Following the end of the Cold War, an anticipated DPRK collapse did not materialize,
though it did experience severe economic hardship. This led the two KSRs to establish the
Agreement on Reconciliation, Non-Aggression, Exchange, and Cooperation in 1991, which was
followed by the US and North Korea signing the Agreed Framework in 1994.
435
These accords
were intended to improve economic and cultural interchange, as well as to reduce North Korea’s
nuclear weapons capacity. A direct military hotline was also created to avoid possible crises
resulting from misinterpreted signals.
A prominent attempt at cooperation was the Sunshine Policy of the ROK under the
administration of Kim Dae-jung. This policy began in 1998 and was intended to foster economic
exchange, discuss reunification, and allow families divided by the DMZ to be reunited. It was
continued by the administration of Roh Moo-hyun. There were two Inter-Korean summits to
establish concrete benchmarks for the Sunshine Policy. However, numerous DPRK ballistic

257

missile tests, pressure from the US and Japan, and the election of Lee Myung-bak saw the Sunshine
Policy end in 2010.
436
The policy was revived in 2017 under the administration of Moon Jae-in.
The Inter-Korean summits of 2018 to discuss denuclearization followed soon after.
Though peace talks in 2019 failed to yield results, and Kim Jong-un has expressed interest
in improving both nuclear and artillery capabilities as deterrents, Kim and Moon appear to be on
friendly terms.
437
However, concrete steps toward reunification, denuclearization, or the removal
of artillery threatening Seoul remain elusive. Despite removing some favorable language
characterizing Japan in their 2019 white paper—reflecting a worsening relationship—the ROK is
still allied with Japan. The ROK also seeks to enhance their security ties with the US while
simultaneously engaging in less provocative military exercises.
438

Despite these apparent movements toward reconciliation, it is apparent that current and
historical proposals to form confederal frameworks have yielded few results as each rival fears
gradual conquest by the other, though that fear has shifted from the South to the North since the
end of the Cold War. Taking over the rival through force is also impossible; the DPRK has
powerful deterrents, while the ROK has remarkably powerful allies that would come to its aid.
Furthermore, despite the tensions between Japan and South Korea, there is no evidence that the
South and North would fight Japan together in the near future; even though there are naval tensions
and near-conflict events between the ROK and Japan,
439
the pervasive influence of the US would
discourage any such arrangement.
Additionally, there is no reason to expect to see cooperative events intended to eject the
US from the Korean peninsula. The US also commits relatively few violent crimes on the peninsula
compared to what would be expected for an occupying force. Ultimately, cooperation between
KSRs does not involve unity against outsiders. Additionally, with the number of unaligned

258

coethnics set to essentially zero by formal division by borders, we would not expect this to be a
motivator for unity to eject outsiders.
7.6 Summarizing the TSON case
Let us examine the overall findings of the TSON case as represented by rival KSRs. These findings
will be compared to the DSN, SSS, and DIG. The following chapter will review the results across
all cases collectively. The results for KSR rivalry are presented below in table 7.2:
Table 7.2
Summary of evidence for TSON conflict based on KSRs
Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

P1 Roots of Rivalry
H1a Competing Leaders DPRK-ROK N Factional Moderate
material
H1b State Manipulation DPRK-ROK N Factional Moderate
material
H1c Divergent DPRK-ROK N Temporal Strong
Development analysis,
factional
material
P2 Effects of Distance
H2a National Capital DPRK-ROK Y Historical Moderate
evidence,
factional
material
H2b Regional Center DPRK-ROK N Historical Moderate
evidence,
factional
material

259

Proposition/ Hypothesis Conflict Supported Source of Degree of
Dyad (Y/N) Support/ Support/
Rejection Rejection

H2c Unpacific Borders DPRK-ROK Y Historical Moderate
evidence,
factional
material
H2d Primary Routes DPRK-ROK N Historical Moderate
evidence,
factional
material
P3 Out-group Incursion
H3a Who Dies and DPRK-ROK N Historical Strong
Where They Die evidence,
factional
material
H3b Heroes and Traitors DPRK-ROK N Historical Moderate
evidence,
factional
material
H3c Non-combatant DPRK-ROK N Historical Strong
Deaths evidence,
factional
material

Note. Weak support: evidence was either statistical (* or **) or qualitative. Moderate support:
both statistical (*) and qualitative evidence. Strong support: both statistical (** or ***)
and qualitative evidence.
\
As with the other cases, let us proceed in order, beginning with findings for the Roots of
Rivalry (P1). The Competing Leaders hypothesis (H1a) was not well-supported. Different ROK
leaders from different political parties experienced varying relations with the Kims of the DPRK.
Yet, results in the post-Cold War world remained the same; summits, agreements, and accords

260

have been generally unsuccessful. Rhetoric may vary, but even ROK hard-liners have not sought
to end conciliatory policies without provocation from the DPRK. However, despite the variation,
violence has not markedly increased either. Individual leaders are thus unlikely to be significant
driving forces in inter-KSR conflict.
State-level factors are certainly the sources of irreconcilable differences. The disparate
political, economic, and socio-cultural features of rival KSRs cannot coexist without a loose
confederation that can accommodate such differences. However, as there is no “host state” in that
there is no superstate that KSRs must contend with, the State Manipulation hypothesis (H1b) holds
little utility for this variety of cases at present.
If there were a formidable irredentist movement in the contiguous Korean-inhabited
regions of China and Russia, it would present a fascinating route of study. The Korean TSON case
would be extremely unique; essentially there would be an irredentist movement with two coethnic
states to select among for support. In the event of such a development in the future, the dual-TSON
would have to balance the desires of their coethnics with those of the host states of Russia and
China, which would allow for a new route of comparison to the DSN represented by KPFs.
Like the DSN, system-level factors explained by Divergent Development (H1c) are the
best explanation of intraethnic conflict for the TSON. KSRs and KPFs developed radically
different systems due to their division by borders. As was evidenced by the KSRs themselves, the
division is viewed as the root cause of the difficulty in achieving reunification.
Moving on to the Effects of Distance (P2), we see some similarities to the DSN, but key
differences were found as well. Like other interstate conflicts, the National Capital (H2a) is a likely
object of contestation. The DPRK’s artillery deterrent is aimed at Seoul. Naval incursions appear

261

to head toward Seoul. The administrative nerve centers of Seoul and Pyongyang would be the
initial objective of any potential inter-Korean war. In this regard, the KSR is different form the
KPF, which was neither concerned with taking the national capital, nor were they concerned with
whether their conflicts might draw in state actors by approaching the national capital and even
occupying territory inhabited by non-coethnics.
Unlike the DSN, this TSON also focuses significantly less on controlling the Regional
Center (H2b). This result is expected; without the administrative control of the capital, regional
centers would provide comparatively less utility. Infiltrating these smaller centers provides little
reward compared to the national capital.
Very much like the DSN is the importance of the Unpacific Border (H2c). Though the
border does serve to dissuade interstate war, the nature of the DMZ provides ample opportunities
for patrols belonging to rival KSRs to engage each other. They may be accidental; they may
conversely be probing attempts passed off as accidents. Like the KPFs, KSRs can derive
considerable utility from infiltrating the borderlands of their respective territories.
The lack of trust between rival KSRs incentivizes intelligence collection in the event of an
escalation of tensions and the alteration of information costs imposed by division. Also, like the
DSN, this lack of trust has its root in system-level factors. Continued divergence, reliance on out-
group allies, and the need to pursue group leadership suggests that externally imposed artificial
borders have similar effects on state and non-state actors whose primary identity is ethnic in
nature—if the division results in the creation of multiple states or no states. By contrast, the DIG
has little to no organized intraethnic conflict as there is a single state actor that can assume group
leadership and help enforce cohesion.

262

Primary Routes (H2d) might be relatively unimportant in post-Cold War TSON conflict,
but they may be salient in certain contexts. It is assumed they would be important should there be
an inter-Korean war. Additionally, this work did not study waterways as primary routes. DPRK
submersibles were used in ROK rivers. Naval clashes at the NLL also suggest that future work
could include waterways. This will be expanded upon in Chapter Nine.
Out-group Incursion (P3) is distinct when compared to the DSN in this work, with only
one similarity. Who Dies and Where They Die (H3a) is relatively unimportant for post-Cold War
conflict. There are no unaligned Koreans anymore within the two states.
440
Both states perform the
state-making activities of defining territory and controlling populations quite well. Unlike DSNs,
where control over territory and populations can be fluid due to comparative differences in the
capabilities of their host states, TSONs do not face that problem.
The findings from the Heroes and Traitors hypothesis (H3b) are also dissimilar to DSNs.
Recall that North and South Korea have sent each other aid before and after the Cold War. KPFs,
on the other hand, reject each other’s offers of assistance as they see them being offered cynically.
Accepting assistance would make the accepting party seem weak and the offering party would be
viewed as being in a better position of leadership. KSRs, on the other hand, do readily accept offers
of assistance, even if the assistance is hoarded, squandered, or furthers North Korean self-reliance,
rather than bringing it into the liberal international economic order.
441

Finally, Non-combatant Deaths (H3c) appears insignificant in generating calls for KSR
unity to expel out-groups, similar to KPFs. While there are nearly 30,000 US troops in South Korea
that commit several hundred crimes against Korean citizens per year, the ROK will clearly not
seek to ally with the DPRK to eject the US. Like the DSN, the TSON in this work values political

263

outcomes for the faction over the emotional solidarity that we would expect from using the first
generation of the literature on rival cooperation.
Thus far, we have examined all four cases of this work, beginning with the Kurdish DSN,
and proceeding with the Chechen SSS, the Albanian DIG, and the Korean TSON. Results and
findings in each of the latter three were compared to those of the DSN. The following chapter shall
examine the findings and results in the aggregate. This will further establish the utility of
considering intraethnic conflict in a categorical manner.

264

Chapter Eight
Findings: Tentatively Supporting a Categorical Approach to Organized Intraethnic
Conflict

The preceding cases demonstrate interesting variation between archetypical cases of conflicts
involving a divided stateless nation (DSN), single-state separatist (SSS), divided irredentist group
(DIG), and an arrangement of two states, one nation (TSON). Tables 4.4,5.3, 6.3, and 7.2 detail
the findings in light of the three propositions of this work for each variety. This chapter will unify
the findings from each case in order to develop new theoretical propositions. These propositions
are the categorical pathways of organized intraethnic conflict that comprise this work’s primary
contribution. This chapter proceeds as following: Section 8.1 very briefly reviews the propositions
and their hypotheses. Sections 8.2, 8.3, 8.4, and 8.5 lay out new hypotheses supporting a
categorical approach to organized intraethnic conflict. Finally, Section 8.6 summarizes all findings
together before this work moves to its concluding chapter.
8.1 Summing up: A review of propositions and hypotheses
For the convenience of the reader, let us examine once more the propositions and hypotheses
previously displayed in table 2.1 that make up the Conflict and Cooperation Theory:

265

Table 8.1
Conflict and Cooperation theory: Review of propositions and hypotheses
Proposition Name Prediction
Hypothesis
P1 Roots of Rivalry The roots of DSN rivalry that develop into
conflict are likely to be similar to one
recognized form of intraethnic conflict than
the others

H1a Competing Leaders DSNs are more likely to engage in conflict
following competition between individual
leaders than other explanations

H1b State Manipulation DSNs are more likely to engage in conflict
at the behest of out-group state actors than
other explanations

H1c Divergent Development DSNs are more likely to engage in conflict if
they develop divergent and irreconcilable
societal systems
P2 Effects of Distance As distance from an object of distance
increases, instances of conflict are more
likely to decrease than increase

H2a National Capital As distance from the national capital
increases, instances of conflict are more
likely to decrease than increase

H2b Regional Center As distance from the regional population
center increases, instances of conflict are
more likely to decrease than increase
H2c Unpacific Border As distance from the IRB increases,
instances of conflict are more likely to
decrease than increase

H2d Primary Route As distance from the primary route
increases, instances of conflict are more
likely to decrease than increase

266

Proposition Name Prediction
Hypothesis
P3 Out-group Incursion Rival ethnonationalist factions are likely to
experience unity or fragmentation during
periods of incursion by an out-group than
during periods of institutional equilibrium

H3a Who Dies and Where Rival factions engaged in conflict with one
They Die another are more likely to cooperate
against an external foe that deliberately kills
unaligned co-ethnics in institutionally
competitive territory than during periods of
institutional equilibrium

H3b Heroes and Traitors Rival factions engaged in conflict with one
another are less likely to cooperate against
an external foe that deliberately kills co-
ethnics aligned with a rival faction in
territory firmly held by the rival
H3c Non-combatant Deaths Rival factions engaged in conflict with one
another are more likely to cooperate against
an external foe that deliberately kills non-
combatants regardless of territorial control
or factional affiliation

Note. The general propositions of P1, P2, and P3 frame the sequence of testing; the hypotheses
correspond to the specific tests explained in the research design.

The propositions and their associated hypotheses were applied to each of the four archetypical
cases. Let us now turn to the findings for each category, beginning with the Kurdish political
factions (KPFs) that represented an ideal DSN.
8.2 DSN conflict and cooperation
In brief, the KPFs exhibited interesting similarities to and differences from the other forms of
organized intraethnic conflict, yet they appear to also display behavior that is unique to this

267

category. System-level factors were most impactful in the origins of these conflicts, like the TSON,
but unlike the SSS and DIG, despite being more similar to the latter two types with regards to their
material capabilities. Though powerful states might prevent DSN conflict from occurring in their
territory, once a DSN reaches a threshold of capacity the host state seems to lose influence in
driving conflict within the group. Establishing this threshold could emerge as a research priority
for future work. Thus, the findings of the Roots of Rivalry proposition (P1) go against what we
would have expected given the way the literature has long treated the phenomena of separatism
and irredentism.
The Effects of Distance proposition (P2) are more in line with expectations with regards to
DSN behavior, but here again there are key differences. Regional centers and border crossings
were demonstrated to be the most salient objects of contestation—somewhat like the SSS for the
former, but like the TSON for the latter—despite their physical geographic arrangement most
closely resembling the DIG.
Finally, in light of Out-group Incursion (P3), the KPFs displayed behavior that was
completely unlike the other forms of conflict. Rival factions representing the DSN appear to
cooperate against outsiders if that external force deliberately kills unaffiliated coethnics. They
cooperate in order to enhance their position within the group. Rivals will not cooperate against
out-group threats that only target rivals. The cooperative aspect is totally unlike any other form of
organized intraethnic conflict.
Organized intra-Kurdish conflict did not occur immediately following their division by
IRBs. Rather, it took decades before large-scale organized conflict and warfare broke out.
442
The
cases of the Iraqi Kurdish Conflict and the conflict between factions associated with the KNC and
KCK involved different parties but featured similar roots in divergent development. The following

268

three figures will depict the findings from the three propositions, beginning with the origins of
conflict for DSN factions based on KPFs is displayed below in figure 8.1:
Figure 8.1
Origins of conflict between rival DSN factions

Note. Created by the author using https://www.diagrams.net/

Figure 8.1 explains the findings exhibited by the theory-building case of the KPFs from Chapter
Four. The division of an ethnic group by IRBs can have an impact on a stateless ethnic group.
Existing and new ethnopolitical organizations may develop self-determination claims that are
intended to apply to the group as a whole, but they may need to adjust to new realities in their host
states that can limit or encourage expansion.

269

These factions would receive a great deal of influence from system-level forces as they are
sought as allies by rival powers. Their economic systems may be influenced by trade with their
allies. Transnational linkages to extraregional allies, host and neighboring states, economic
partners, and coethnic kin in neighboring states can contribute to the development of incompatible
institutions.
However, it is those institutions that are carried forward as those factions pursue self-
determination claims and engage in state-making activities, especially when an exogenous shock
alters the regional status quo. The period of separation and the advent of divergent systems alters
the low information costs we would otherwise expect among coethnics. Those institutions form
the basis of a proposed state or statelike entity as competing factions seek to create a new polity
of some kind.
Convincing coethnics of the superiority of those institutions may come at the expense of
rivals as factions pursue in-group leadership. With altered information costs brought on by system-
induced incompatibility, such rivals may be more likely to engage in organized intraethnic conflict
in order to be victorious as they engage in state-making activities.
What does conflict between DSN factions look like once it begins? Figure 8.2 displays the
objects of contestation:

270

Figure 8.2
Objects of contestation for DSN factions in conflict

Note. Created by the author using https://www.diagrams.net/

Once organized conflict between DSN factions begin, such factions must consider the risks of
expansion. The proximity of the national capital of all potential states in the group’s
neighborhood—as well as the territory inhabited by non-coethnics—can predict the likelihood of
intervention.
443
Based on the experiences of KPFs, it is currently expected that a DSN confident
enough to engage in organized conflict will be relatively unconcerned with the power of host states
to intervene, and thus will expand toward territory we would otherwise expect them to avoid.

271

However, this prediction remains falsifiable as there may be variation among DSNs, of which there
are many.
This work does not exclude economic motivators as objects of contestation, but it appears
at this early stage of conflict that DSN factions prioritize strategic concerns.
444
Physical control of
territory—as evidenced by the material and ideational significance of the major regional
population center—will likely be the most fiercely contested object.
Additionally, rival factions will seek to leverage their transnational connections to friendly
coethnics. However, their rival will seek to do the same. The KPFs in all periods of conflict relied
on local militias comprised of coethnics to appear legitimate and to combat their rivals. Borders
will likely be contested as well because each rival would logically seek to bring in their own
transnational allies while degrading the capability of their rival to do the same.
Sometimes, there may be some kind of event that causes rivals to cease fighting each other,
or at least consider a pause. However, we would expect such an event to be extraordinary; some
manner of exogenous shock that would make rivals reappraise the risks and consequences of
continued organized intraethnic conflict. Figure 8.3 displays the drivers of rival cooperation
between DSN factions.

272

Figure 8.3
Cooperation between rival DSN factions

Note. Created by the author using https://www.diagrams.net/

As organized intraethnic conflict occurs, rival factions may occupy new territory inhabited by the
ethnic group that it had not previously administered. Many of these newly accessible communities
will likely be weakly affiliated or totally unaffiliated with either faction. Rivals will seek to
introduce their institutions to that territory as a means of pursuing in-group leadership. One group
may win; alternatively, external mediation or negotiations between the factions may terminate the
conflict.
445
The conflict may also become protracted in nature.

273

However, should some out-group actor invade the same new territory that is claimed by
rival DSN factions, the calculus of the factions could change, depending on the actions of the out-
group forces. If rivals are targeted, even non-combatant civilians that are activists or politicians
affiliated with a rival, the unaffected rival will likely not offer assistance. Alternatively, they may
offer assistance in order to assert their willingness to sacrifice for the group, knowing that their
rival will reject assistance. If the rival were to accept this offer of assistance, they run the risk of
being viewed as a subservient faction that requires help from a superior one in order to protect the
group. The accepting rival would then also provide the offering rival a foothold in their territory
that they may otherwise not have to give up.
If the out-group forces
446
indiscriminately kill unaffiliated coethnics, all rival factions lose
reputation; they cannot defend the group and they risk losing that territory and population
permanently. If that happens, no rival can incorporate that territory or community into their state-
making strategy. Moreover, it will make them look weak to their domestic publics, which may
clamor for in-group cohesion out of feelings of nationalism. This is somewhat like we would
expect from the first generation of research, but different in that the factions themselves are making
political calculations for their continued viability, rather than necessarily feeling any sort of
emotional affect.
The desire to compete for influence with unaffiliated coethnics rather than risk losing the
territory to an out-group is expected to override the desire to see the capabilities of the rival
degraded. Factions will prefer to compromise on introducing rival incompatible systems if it means
they retain a chance of incorporating the territory or community at a later time. Thus, we would
expect rival factions to temporarily assist each other under this condition. They might permit rivals
to establish outposts and systems of governance in their own recently claimed territories. Factions

274

may fight alongside each other in key hotspots to demonstrate their leadership ability and their
devotion to protecting the group.
However, even these cooperative actions will likely be competitive; factions will extol the
virtues of their rival (for the time being), but still portray the rival as subordinate. Cooperative
actions will also likely not last very long should the out-group forces be defeated. Conflict between
factions may resume as the state-making activities of each faction are free to be pursued once
more.
Perhaps other cases will flesh out other avenues following the defeat or withdrawal of an
out-group. Tentative examples of the Pashtun-dominated Taliban and the Salafist Kanuri faction
of Boko Haram suggest that coalescence around a clear victor could occur. The institutional
equilibrium may well shift to favor one faction at the expense of others. This would be different
from the case of Kurdish factions, in which clear winners and losers are not readily distinguishable
from each other.
8.3 SSS conflict and cooperation
To summarize, the CSFs from Chapter Five displayed some similarities to KPFs. However, they
were surprisingly different considering how similar they are with regards to their shared
statelessness and low levels of military capacity. Individual-level factors provided some possible
opportunities for conflict to develop, but state-level factors were the most salient, with system-
level factors being mainly absent.
Regional centers were heavily contested, like the DSN. Borders, however, were not. State
actors will likely control them too thoroughly to be much contested by SSS factions. Also, unlike
the DSN, the national capital and territory inhabited by non-coethnics are likely to be very

275

powerful determinants of conflict, or rather the lack thereof. SSS factions will use terrorism to
target members of the out-group as we saw in the case of Chechen factions, but they do not appear
to seek to occupy their territory during periods of organized intraethnic conflict.
SSS factions are not expected to cooperate once engaged in conflict, regardless of who is
killed and where. Very powerful SSS factions that are able to defeat state forces—like the ChRI
in the first phase of the Chechen conflict continuum—gain nothing from allying with their foes.
If, however, the dominant factions are unable to resist state forces, their weaker rivals can only
gain power, wealth, and influence through the defeat of the dominant faction by receiving
patronage from the host state.
The following figures will follow the same pattern as the preceding ones for the DSN.
Figure 8.4 displays this work’s theory of the origins of conflict among SSS factions:

276

Figure 8.4
Origins of conflict between rival SSS factions

Note. Created by the author using https://www.diagrams.net/

A stateless ethnic group enclosed by an IRB may include more or one faction. Unlike the DSN,
factional cleavages are not likely to be introduced by system-level factors. Rather, these are likely
to be more personal or socio-cultural in nature as within-state contact is maintained in the absence
of division by IRBs. A faction might have a particular religious orientation, or there might be
grudges between leaders that have known each other for some time.
If they rebel against the state, the host state would likely seek to minimize their own
potentially costly involvement, and so would seek to induce coethnic factions to turn on one

277

another. The once-dominant faction may seek to survive against new threats, while the faction
receiving state support could seek to supplant the rival. The preexisting cleavages—even if they
did not lead to conflict previously—can be instrumentalized by the state, which can derail
secessionist desires. Instead, rival factions will engage in organized intraethnic conflict on the
terms of the host state.
Once SSS factions are engaged in conflict, certain objects will be contested more than
others. The conduct of SSS factional conflict is displayed in figure 8.5:
Figure 8.5
Objects of contestation for SSS factions in conflict

Note. Created by the author using https://www.diagrams.net/.

278

Unlike DSNs engaging in state-making activities that cross IRBs, SSS factions will be much more
concerned with proximity to the national capital of the host state. Secessionist violence will likely
avoid territory inhabited by the out-group, as well as territory closer to the capital because they
will likely not have the resources to contest control of such areas. There may be acts of terrorism,
but other objects are more likely to be contested, especially if the state has allies amongst the SSS
faction’s rivals.
447

The regional center’s material and ideational significance makes it a vital object of
contestation for a SSS faction that wishes to retain its position of dominance, as well as a rival
faction that seeks to supplant them. The SSS is unsurprisingly quite similar to the DSN in this
regard.
Unlike conflicts involving very powerful, expansionist DSNs, borders are likely to be
effectively controlled by the host state, as well as by neighboring states that wish to avoid conflict
spillover.
448
This is expected in a case such as that of the CSFs. The power asymmetry and
antagonistic history between Russia and Georgia provide for interesting findings; perhaps a very
weak state would wish to avoid conflict with a much stronger neighbor but assisting ethnic rebels
in the more powerful state might also be in the neighboring state’s interests as Russia accused
Georgia of doing with regards to Chechen rebels. Meanwhile, Russia also supported ethnic
rebellion in Georgia. Future within-SSS comparisons might look for whether this varies across
cases of SSS conflicts.
Cooperation is highly unlikely to occur between rival SSS factions. Regardless of which
coethnics are killed and where they are killed by state forces, rivals are unlikely to compromise.
Doing so would require conflict with the host state on the part of the faction benefitting from
collaboration with that actor. Figure 8.6 displays the findings based on CSF conflict:

279

Figure 8.6
Lack of cooperation between rival SSS factions

Note. Created by the author using https://www.diagrams.net/

The host state would seek to induce rivals to engage in organized intraethnic conflict over group
leadership that could be based on group-level or individual differences. The host state may also
directly intervene
449
and target unaffiliated coethnics, rivals, or non-combatants without
accounting for affiliation. The cleavages being exploited by the state would likely outweigh
concerns over coethnic life because state support is vital for the rival challenging the dominant
one.

280

The dominant SSS faction can attain two outcomes in its conflict with the state and its
rivals. If the pro-state force topples the dominant one, we would expect the incoming regime to
seek to preserve its newfound position relative to the once-dominant rival and new challengers.
Aided by the host state, conflict will likely decrease in intensity. If, however, the dominant rival
defeats the challenger and then at least holds its own against the state, it will seek to pursue
secessionist goals further. The intraethnic conflict may end for a time. However, the host state may
retool its approach to exploiting cleavages between individuals and factions. The host state has the
luxury of time, and it can seek out new proxies and try again if widespread recognition of the
breakaway administration does not occur.
8.4 DIG violence and cooperation
Briefly, the AIFs from Chapter Six function quite differently from DSNs, though both are divided
by IRBs. Statehood status and division isolated from one another do not account adequately for
DSN factional conflict—suggestions for the future study of this key finding will be detailed in the
following chapter.
Rivalry among DIG factions exists, but do not appear to lead to open warfare. Members of
DIGs do engage in violence against one another, but there is no strong proof that it is officially
encouraged by the ethnopolitical organizations that claim to represent their coethnics. The
representative nation-state likely has a great deal of influence among irredentist groups seeking to
join that state. Such a state can enforce a cooperative relationship, whether the factions involved
want it or not.
Interestingly, when there is intraethnic violence, though not necessarily organized conflict,
individual-level factors seem to be the driving force. Also, this is the lone case in which the

281

alternative explanation of economic motivators driving conflict can be supported—though to
repeat, it drives incidents of violence between individuals, not organized intraethnic conflict.
The following figures will depict the origins of violence between DIG factions, as well as
the objects fought over and the reason why cooperation between rivals is maintained. Figure 8.7
displays this work’s theory of the origins of violence between individuals belonging to DIG
factions engaging in violence:
Figure 8.7
Origins of violence between members of rival DIG factions

Note. Created by the author using https://www.diagrams.net/

282

An ethnic group might be divided by IRBs, but one concentration of that group may come to
possess a representative nation-state. Other concentrations—irredenta—remain outside of that
state, but factions claiming to represent them may seek to join the state. Some individuals within
and across factions may have grudges against each other. Other key figures may wish to control
economic sources of wealth at the expense of their individual rivals. However, irredentist claims
require assistance from the coethnic nation-state, and so rival factions are unlikely to go to war
against each other, lest they risk punishment by the state over fragmentation and defection. Thus,
rather than organized intraethnic conflict, we would expect to see sporadic, unsanctioned violence
committed by individuals against others.
The logic for the objects of violence is displayed in figure 8.8:

283

Figure 8.8
Objects of contestation for individual-level DIG violence

Note. Created by the author using https://www.diagrams.net/

284

Should individual members of DIG factions engage in violence with one another, we would
expect the coethnic nation-state to restrain the actions of the antagonists. Those individuals depend
on the coethnic state for support in their irredentist goals. They may even rely on the state for their
very survival against out-group forces. Therefore, they would not seek to overtly embarrass their
benefactors. Thus, major objects of contestation such as regional centers and border crossings
would be proscribed to the antagonists. Open warfare between rivals would damage the credibility
of the state as a protector of the group, which would mean the state would have to take drastic
action against smaller factions that harm the reputation of the state.
However, individual-level violence could still be expected to occur. No ethnic group is
monolith in nature. If the state itself is not affected by a small number of violent actions, absolute
regulation of individual-level grudges will likely be beyond the capabilities or concern of the state.
Likewise, individuals might contest economic resources, as was seen with individual members of
AIFs and their desire to control TOC revenue streams. Even unsubstantiated accusations of such
actions might be spurred by individual-level grudges.
As expected based on what we know about the pacific effects of coethnicity and the
predictions from the first generation of literature regarding rival cooperation, DIG factions
primarily work together, albeit imperfectly. Figure 8.9 displays the mechanisms through which
they do so:

285

Figure 8.9
DIG cooperation

Note. Created by the author using https://www.diagrams.net/

Irredentist movements within the DIG might seek to join their population and territory to the
nation-state to form a “Greater X.” There may be fissures, however. The hosts state or states that
host the territory inhabited by populations of irredenta may also view some factions as being more
imminent of threats than others. Should the irredentist movement gain sufficient strength, out-
group forces
450
may violently intervene in the territory claimed by the irredenta.
451

Regardless of who is targeted by the out-group force,
452
the coethnic nation-state can
manage cleavages that could otherwise be exploited by one or more host states combating

286

irredenta. The overall strategic management provided by the state means that cooperation between
rivals—at the operational level at any rate—will likely be maintained.
453

8.5 TSON conflict and cooperation
To briefly review what was learned from the experiences of KSRs in Chapter Seven, the origins
of TSON conflict are driven by system-level factors. Externally imposed IRBs divide a nation-
state into two. Opposing regimes develop institutions along the lines encouraged or permitted by
their benefactors. Rival benefactors engaging in proxy conflict or competition will contribute to
the alteration of information costs, similar to the DSN.
The objects of contestation share an important similarity to the DSN, but more differences.
National capitals are contested if open warfare erupts, while regional centers would not be primary
targets, which is the opposite of what was found in the case of the DSN. Borders, like for DSN
factions, would be heavily contested.
Cooperation against out-groups cannot occur for the same system-level linkages that
contributed to the development of the rivalry in the first place. The external benefactor
nonsensically would need to acquiesce to its proxy allying with their rival’s proxy to eject the
benefactor’s own forces. Rather, cooperation would focus not on expelling outsiders by force, but
by effecting a beneficial framework of (re)unification with their rival along lines favorable to the
proposing regime. Successful (re)unification could only lead to the departure of out-group forces
if external benefactors give their approval to the framework. As long as those external benefactors
are rivals, this is less likely to occur.
As noted earlier, the departure of out-group forces typically would require one of two
outcomes. The first is conquest, which is what happened to South Vietnam at the hands of the

287

North. The second is absorption, as West Germany did with East Germany at the end of the Cold
War. Considering that neither the US nor China do not seem to be on the precipice of collapse at
this point in time, and neither appears willing to leave their Korean ally to its own devices, we
would not expect either outcome to come to fruition any time soon.
The following three figures lay out the mechanisms of the theoretical contributions
described above for the TSON. We begin with figure 8.10, in which the origins of TSON conflict
are displayed below:
Figure 8.10
Origins of conflict between rival SSS factions

Note. Created by the author using https://www.diagrams.net/

288

IRBs can be imposed upon an ethnic group’s preexisting nation-state,
454
which could result in the
creation of two or more states, each with sovereignty exercised by their respective regimes. Those
regimes might seek to pursue policies of (re)unification with their counterpart. Those policies are
also likely to be influenced by each regime’s regional and extraregional allies. The policies can
help determine what sorts of systems can be developed based on whether such policies can be seen
as superior to those of their rival. The preferences of external sponsors for the form of
(re)unification policies can also carry over to the divergent development of each faction.
Like the DSN, incompatible systems and institutions, as well as altered information costs
contribute to the pursuit of in-group leadership. The change in the nature of contact and the decline
of the pacific effects of coethnicity can lead to organized intraethnic conflict. Unlike DSN factions,
however, rival regimes administer a state with fixed territorial bounds. DSN factions seek to forge
something new, and so they may try to incorporate territory they may not have claimed in the past.
The TSON regime, on the other hand, needs to prevail over another regime with its own predefined
territory.
TSON conflict in the Cold War was devastating. Conflict in the post-Cold War period is
less destructive, though the risk of open warfare remains. It appears, however, less likely to occur
than warfare between DSN factions.
455
The conflict itself will likely be meant to probe the resolve
of the rival regime more so than conquest. The theory of which objects of contestation are likely
to be fought over are displayed in figure 8.11:

289

Figure 8.11
Objects of contestation for TSON regimes in conflict

Note. Created by the author using https://www.diagrams.net/

This cyclical diagram can logically begin with the onset of conflict between rival TSON regimes.
The theory laid out in figure 8.11 relies on the assumption that external sponsors prefer the status
quo in the post-Cold War era; they do not wish to risk direct conflict with their own great power
rival for the sake of violent TSON (re)unification. Instead, they will support their allied regime’s
attempts to pursue in-group leadership through clandestine or low-intensity means that carry an
acceptable risk of discovery but avoid escalation into open warfare.

290

External sponsors will prefer a non-violent method of (re)unifying the TSON under a
framework that is favorable to the institutions of their allied regime.
456
As this externally regulated
strategy does not address the alteration of information costs, this cycle will most likely continue
indefinitely.
However, if violent means are viewed as feasible, rival regimes may seek to test the resolve
of their foe. They would seek information about such resolve and about the capabilities of the other
regime in the event that war is perceived as inevitable in the future. Possible strikes must be
carefully selected; only worthwhile operations would be considered. Regional centers might
provide little utility if opportunities for attacks are limited. DSN factions on the other hand
prioritize control within the group; TSON regimes prioritize attaining a return on expending some
effort on an inherently risky action.
Like DSN factions, TSON regimes will necessarily be somewhat constrained by the border
and other forms of demarcation. These are also primary targets. Unlike DSN factions, however,
the national capital would be probed. Though DSN factions are not expected to seek to capture the
national capital, very powerful DSN factions very likely could do so.
457
Perhaps occupying the
national capital is not worth the trouble, despite the availability of resources that would result. Or,
perhaps, the DSN faction is genuinely committed to its state-making goals. Further investigation
might determine why; regardless, this aspect of organized intraethnic conflict remains a key
difference between the DSN and the TSON.
The nature of proxy conflict in the Cold War era made cooperation to eject outsiders
impossible during that period.
458
TSONs do not readily unite in the post-Cold War period, either.
Figure 8.12 displays the theory for why it does not occur:

291

Figure 8.12
Lack of cooperation against out-group forces between TSON regimes in conflict

Note. Created by the author using https://www.diagrams.net/

Organized intraethnic conflict among TSON regimes will often involve a tangential strategy of
pursuing competitive frameworks rather than open warfare. The rival TSONs’ allies and trade
partners, generally preferring the status quo, will seek to consolidate their influence over their
proxies, thus they prefer the framework implementation strategy over open warfare. Thus, the
organized conflict would become intermittent. However, information costs continue to be adjusted.
Rival regimes will still wish to be prepared for war, even if they do not actively seek to
engage in it. Sponsors of rivals will also wish to know how firm the support of their rival is for

292

their proxy. If there is a massive asymmetry of capabilities and will, an external sponsor may
encourage the resolution of the TSON question through conquest of the rival proxy by the TSON
regime they support.
If rival sponsors support (re)unification, there may be a negotiation over which framework
is preferable, which could result in absorption or compromise. This option and the option of
conquest occurred during the Cold War and its immediate aftermath; they do not seem viable for
post-Cold War TSON conflicts.
The expected outcome for post-Cold War TSON conflict is that of the last pathway
depicted in figure 8.12: interminable, intermittent armed confrontation. External sponsors do not
wish to acquiesce to their own great power rivals or upset the status quo, but they also do not wish
to relinquish influence through a compromise between TSON regimes that excludes the external
sponsor. As issues of trust are not addressed by maintaining the status quo, the pacific effects of
coethnicity are diminished.
Though there are only two TSONs remaining (and one such regime, China, is a great
power), there are many potential TSONs should there be a cascade of successful self-determination
movements. As only one example, there were once two KRGs that were de facto independent.
Should regional and great powers create two Kurdistans—one KNC-dominated and the other
KCK-dominated—there would be a new TSON. There are many independence movements
globally that could potentially be turned into TSONs to resolve (or inflame) regional and great
power disputes over influence.

293

Chapter Nine
Advancing the Study of Organized Intraethnic Conflict

The findings of this work demonstrate that there are potentially significant implications for what
we know about organized intraethnic conflict. At present, the categorical approach offers promise
for pursuing extensions of this research to both confirm or disconfirm the utility of the approach
itself, as well as to further explore potential implications for both literature and for policy. This
chapter will lay out these implications and suggest possible means of studying this phenomenon
for future work. Section 9.1 will be devoted to the literature, Section 9.2 to policy. Section 9.3 will
conclude with a summary of this work.
9.1 The Conflict and Cooperation Theory and its implications for the literature
Pathways of organized intraethnic conflict
The four categories of organized intraethnic conflict described in this work are the divided
stateless nation (DSN), the single-state separatist (SSS), the divided irredentist group (DIG), and
the two-state, one-nation (TSON). Each category exhibited a pathway that is plausibly different
across types, thus offering significant initial support for the Conflict and Cooperation Theory.
Future research could focus on studying multiple cases of each type to explore variation within
each category.
One means of pursuing this research extension would be to use small-n pathway
analysis.
459
The population of cases that are the subject of the case selection graphic in table D.1
located in Appendix D could be a starting point. With modifications based on the findings of this

294

work, a definitive universe of cases could be established. Then, following Weller and Barnes
(2014), the small-n pathway analysis would examine four populations, perhaps as a series of
articles or a book-length project. Each population would correspond to the categories proposed
here. Every category would exhibit a relationship between X1 and Y1. X1 represents the first
propositional pathway involved in this work. There would also be an X2, and an X3 representing
the other propositions.
Y1 is the outcome of the first pathway. The mechanisms linking the outcome to the
purported cause would be tested statistically and explored qualitatively using the population of
cases corresponding to each category. This work provides a blueprint for both testing with an
analytically eclectic strategy, as well as for explaining and illustrating the evidence for the
mechanistic linkages using the systemist technique.
460

Small-n pathway analysis would either support the findings of this work or would
disconfirm both the utility of the approach used here and its findings. This work remains falsifiable;
perhaps the categorical approach would see too much variation with regards to the origins of
conflict within types. What is more likely, however, is that there would be exceptions to certain
cases. These would be considered special but would not mean that this approach is not still useful
for understanding behaviors using these propositions and hypotheses. If, for instance, thresholds
of similarity to other categories were to be exceeded overmuch, the flaws in this approach could
be identified and the approach could be refined for future applications.

295

Limitation or opportunity? Expanding external applications
Examining the causes of organized intraethnic conflict using more cases would go far in
improving this work’s primary limitation: external validity. The small-n pathway analysis would
be one component a strategy that would improve the robustness of this work’s findings.
As this framework is generalizable to a wide population of cases, a global large-n analysis
could also be undertaken. All ethnonationalist movements that meet the scope conditions laid out
in Chapter Two could be statistically analyzed for similarities by type. The goal of such an analysis
would be to identify which factors predict which type of ethnonationalist factions are more likely
to engage in organized intraethnic conflict than others.
Another approach could perhaps take one aspect of the global statistical analysis, such as
which factors predict when a DSN would engage in organized intragroup conflict but illustrate
those findings using a medium-n design. Several representative movements could be explored
through further statistical analysis and qualitative research using vignettes. This approach has been
used in democratization studies to great effect.
461

Any of these approaches would serve to contribute to the production of middle-range
theory. Our understating of the drivers of organized intraethnic conflict would be better
understood. And so, what at first seems like a limitation is also an opportunity for more scholarship
that touches on many areas of interest. This in turn could fuel collaboration across disciplines and
subdisciplines in the field.
The analytically eclectic empirical framework in this work included spatiotemporal
methods to track potential developments within ethnic groups. The strategy that I developed solved
a key limitation of the prevailing dataset of ethnopolitical distribution, the GeoEPR. The map

296

processing techniques provided insight as to where efforts for research could best be focused.
Furthermore, this strategy allows for both visualization and statistical analysis to accommodate
different learning styles. Using only historiographic research or statistical analysis would be
insufficient for a research project like this. The spatiotemporal analysis is much more accurate and
complete in this regard.
This work also offered a suggestion for mitigating the modifiable areal unit problem
(MAUP). The MAUP can be cause for concern when presenting spatial data because of its
introduction of statistical bias and its implied uniformity of areal units. The use of individual event-
level data using vetted sources of data solves this issue.
Improving the study of the objects of organized intraethnic conflict
This work identified conflict intensity as a proxy for the objects of contestation among rival
ethnonationalist factions. Future work could examine other sources of data to see if the findings
hold. For instance, the Armed Conflict Location and Event Data project can be used to locate
conflict avalanches using very similar criteria.
462

As this research program stands, the location of conflict seems to predict what objects are
sought after by rival factions amidst armed conflict. While this work recorded fatalities and fatality
estimates in its source data, it did not test fatalities per se—it tested geographic proximity of events
in which death occurred. Perhaps directly using fatalities as a new proxy would assist future
scholarship with regards to what we know about the motivations for conflict, but also for
cooperation. However, it is far easier to confirm or disconfirm where deaths occurred rather than
how many died.

297

Though primary routes should be vital for controlling population and territory, statistical
analyses did not reveal a correlation with organized intraethnic conflict. Though there is possibly
an implication for the literature based on these findings, I do not believe the current finding will
hold. As more recent instances of organized intraethnic conflict are identified and analyzed, we
will likely have access to better data. As the quality of volunteered geographic information (VGI)
improves, we can include roadways with greater confidence. With applications like
OpenStreetMap (OSM), VGI is becoming ubiquitous. Like fatality data, future work involving the
conflicts of tomorrow will be able to incorporate OSM data effectively, whereas the cases involved
here would have risked unreliable results, given that many of the roadways included in OSM did
not exist at the time. For these reasons, I believe that future work will support the Primary Route
hypothesis (H2d).
There is also an opportunity for methodological out-growth when studying the objects of
contestation. Recall from Chapter Seven that rival KSRs fought over the maritime northern limit
line (NLL). The DPRK also attempted to infiltrate the ROK using inlets and rivers. Are waterways
borders? Are they primary routes? In this case, they are arguably both. Where do waterways fit
into non-state intraethnic conflicts? Including waterways as objects of contestation could
potentially improve our knowledge of this phenomenon. It would also allow for the inclusion of
another key TSON, China-Taiwan.
Another methodological route could be the inclusion of digital elevation models (DEMs)
using some cases that do not involve mountainous conflicts in comparison to others that do.
Elevation could be studied in tandem with primary routes in future work. “Ji çiyan pê ve ne heval”
is a Kurdish axiom that means “no friend but the mountains.” Kurdish fighters of all stripes live
and train in rugged, mountainous territory. They use their terrain to their advantage. Primary routes

298

may well be less significant objects of contestation so much as lines of demarcation, as we saw
with the aftermath of the KNC-KCK cooperation.
463
Kurdish fighters have also used mountainous
terrain to approach targets in war, thus avoiding the necessity of accessing primary routes.
Not only Kurdish fighters, but Albanian and Chechen combatants have referenced the
mountains as integral components of their strategies.
464
Future work might use DEMs to see if
these findings hold. As this work explored initial propositions, I omitted them to attain some
measure of parsimony in models that tested theories that prioritized accuracy and completeness. I
expect such findings to hold, and future work is being planned to explore this avenue.
Additionally, future work could look at cases of conflict that involved relatively flat terrain;
this would predict a greater salience of primary routes. Take, for instance, the Kanuri. Kanuri
factions present an ideal comparison to Kurdish ones given their similar degree of division into
multiple states, conflict with host states, and territorial claims that harken back to a glorious past.
They are different in one regard that is key for this aspect of conflict; they do not inhabit a very
mountainous region. As this case does not involve fighters that exploit mountains for tactical
advantages, primary routes likely factor much more heavily into intra-Kanuri conflict than intra-
Kurdish conflict, while conflicting elevation (which was lacking in these cases) can be likewise
compared. Studying elevation and road access together within most-similar cases should prove
fruitful for improving our knowledge of how these conflicts play out.
The last remaining puzzle for our understanding of organized intraethnic conflict pertains
to the relatively low utility of economic resources as objects of contestation. Despite vast amounts
of natural resources present in the Kurdish, Chechen, and Korean cases, these were not fiercely
fought over by rival factions. Only in the Albanian case did the data reflect possible “greed”
motivators, but even here there were too few incidents to analyze statistically. Perhaps it is the

299

case that survival is indeed more important to these factions in early stages of in-group conflict or
violence, or that greed motivators are only a part of a group’s calculus.
465
Future cases should seek
to prioritize the study of economic motivators separate from other material and, especially,
ideational factors. The possible drawback is that such studies are likely to be more qualitative in
nature given the available data. Perhaps, like with primary route data, this is something that will
improve over time as VGI and other forms of spatial data become more democratized.
466

Improving the study of rival cooperation against out-group forces
The literature on cooperation or the lack of it between rival factions has undergone three
generational phases. The introduction of the DSN that is at the heart of this work provides
important implications for each generation and, potentially, marks the departure point of a fourth
generation of research.
The first generation examined emotional affect on the part of the factions; emotional
solidarity biases factions toward cooperation when faced with conflict with an out-group. Factions
may even initiate conflict with outsiders to force rival in-group factions to join them, whether
smaller rivals wish to or not. This generation works surprisingly well in the case of the DSN in
some regards, though it remains insufficient in others.
Rival DSN factions do indeed cooperate against out-group threats. They also sometimes
cooperate cynically while publicly referencing the importance of defending group identity. Rival
Kurdish factions cooperating against out-groups that posed existential threats to Kurdish people
have always referenced the need to protect their identity from erasure. For these reasons, the first
generation of research remains useful for this variety of cases, though it is even more useful for
the DIG.
467

300

However, rival DSN factions do not always cooperate against outsiders and TSONs seem
to never do so once separate state regimes are established. The first generation of research does
not adequately account for the reasons why a cooperative outcome appears difficult to predict.
There are many conflicts involving out-group forces where rival Kurdish factions did not
cooperate. Though the first generation has many useful insights, it is simply not enough to explain
what we can observe among rival DSN factions.
The second generation looked at reasons why rival factions might not cooperate through
an examination of micro level friction between individuals and groups, as well as the role of state
preferences and pressure. This generation does an excellent job explaining the lack of cooperation
among SSS factions. Rival Chechen factions in particular exhibit all of these frictions. Intra-DIG
violence, likewise, can be explained by micro level disputes between individuals.
However, the second generation of research does comparatively poorly in explaining why
DSNs and TSONs engage in conflict. System-level factors, rather than individual-, group-, and
state-level factors, are much more impactful in creating the conditions necessary for conflict.
468
It
is also insufficient in demonstrating why cooperation is so difficult. The selective use of
violence
469
could potentially explain some aspects of local rival DSN conflict. Rivals that cannot
bear the costs of unity may not cooperate with others, even if they wish to protect coethnics.
However, Chapter Four demonstrated that some KPFs bore the costs of unity anyway, even against
some very powerful out-group actors. The ISG at its peak, for example, was much more powerful
than either the KNC or the KCK, yet factions associated with those rivals cooperated and bitterly
resisted the out-group at great human cost. Islamist forces active in northern Syria were
significantly weaker than the ISG, yet KPF rivals have not cooperated against them. The second

301

generation is insufficient for explaining conflictual and cooperative patterns for the DSN as it
relates to asymmetry of power.
To begin explaining why this variation in outcome is observed, we must look at
implications influenced by the study of the third generation of research, of which there are two
that may comprise the beginning of a fourth generation. The first implication is that territoriality
and political affiliation are key elements of institutional equilibrium that should be included in
future research into rival cooperation amidst organized intraethnic conflict. The state-making
actions of rival factions will necessarily take a turn toward incorporating new physical areas of
control. To maintain legitimacy—both for the populations inhabiting the territory rival factions
seek to gain, as well as their own coethnic domestic publics—rivals will try to preserve their
opportunities to introduce their own institutions as a means of competing with rivals.
Out-group forces that invade such territories and harm unaffiliated populations are
expected to induce cooperation. This is not for genuine emotional affect, but for rather cynical
ones, somewhat similar to what we would expect based on the first generation of research, but
contrary to the second. However, even cynical cooperation is not guaranteed depending on out-
group action, in a departure from the first generation but more in line with the third. For example,
cooperation will not occur if the out-group limits who it seeks to kill, which is counter to
expectations from the first generation where an effect similar to bandwagoning would be expected
to occur regardless.
It is the institutional element introduced by McLauchlin and Pearlman (2012) in the third
generation that is key to predicting whether cooperation will occur or not. However, when it comes
to DSNs, institutional equilibrium must include such factors as exogenous shocks, territorial
control, and the affiliation of populations targeted by out-groups.

302

The second implication for the third generation concerns the state-making activities of rival
DSNs. Rival coethnic factions are rivals for reasons that go far beyond the micro level that was
the focus of the second generation. System-level explanatory factors explain why rivalry develops.
The rivalry is based on competing and, often, irreconcilable differences in visions for an ideal
political entity to represent the group. This is likely to lead to conflict over in-group leadership.
However, competition with rival coethnics is preferred over a loss to out-group forces. Therefore,
the nature of contestation for group leadership should be stressed in future work. Which types of
conflicting visions are more likely to be violent? Does variation in expression this carry over into
SSS, DIG, or TSON conflict? Or is the variation within a category bounded like other elements of
contestation and cooperation, as the findings of this work suggest?
Finally, cooperation could be tested differently. This work laid the groundwork for the
study of who dies and where they die. Future work could look at other proxies. For instance, the
number of deaths can be assessed rather than the locations on their own. Other analyses could also
look at the type of violence. Do atrocities committed by ground forces carry a greater risk of
causing rivals to cooperate than drone strikes or bombings? Perhaps different proxies will reinforce
these findings or provide new insights into what we know about rival cooperation.
The divided stateless nation: State-building, bargaining, and conflict resolution
The possible rise of DSN conflict and cooperation provides an opportunity for expanding
the New Wars literature. The motivations for both conflict and cooperation are partially explained
by a mix of material and ideational factors, but as we saw with the case of the Kurdish factions,
ideology is becoming prevalent once more. The long periods of division can cause adherence to
conflicting systems. However, the ideational, material, territorial, and populational building blocks

303

necessary for multiple factions engaging in state-making activities claiming to represent the same
ethnicity are still compatible with much of the New Wars framework.
DSN factions must contend with many types of challenges when building a state or
statelike entity. Coethnic rival factions, host states, regional and extraregional powers, and even
out-group non-state actors can all be sources of frustration for their goals. Like states engaging in
nation-building, the second image reversed
470
is a powerful force for DSN factions; the KPFs
demonstrated that DSN factions were responsive to the concerns of domestic publics when
unaffiliated coethnics were attacked by outside forces. Unlike state actors creating nationalistic
identities, however, DSN factions represent the inverse wherein actors representing an ethnic
group with a preexisting nationalistic identity seek to create a state. Whether DSN factions
represent a new wave of state-making should be investigated further, given the resurgence of
nationalism and the sheer number of self-determination movements in the world.
The relationship between DSN factions and host states should be examined further to better
understand which DSNs are more likely to witness these patterns of conflict and cooperation than
others. Recall in the case of the KPFs, the most powerful factions expanded in the direction of
their respective national capitals. Which DSNs are powerful enough to engage in state-making
activity to this extent and how do they come to the decision to expand in that regard?
Existing ethnic bargaining literature looks at when minorities rebel against hosts states and
why they do so, typically by using rationalist lenses. However, much of the ethnic bargaining
literature does not significantly distinguish between separatist and irredentist conflict, let alone
those involving DSNs.
471
This work demonstrates a wide degree of variation between conflict
types, in which violence involving DSNs, SSSs, DIGs, and their host state(s) cannot be placed on
the same end of a spectrum running from peaceful bargaining with the center to those forms of

304

violence. The scale varies considerably, as do capabilities, based on which variety of conflict is
being studied.
Taken together, the state-building and bargaining aspects of DSN conflict processes
deserve to be studied further, possibly through a return to the findings of earlier bargaining
literature. Those findings may be confirmed, or they might be updated by studying bargaining
across the types proposed here. Clearly, each type carries different risks. The SSS is most
vulnerable to predation; the DIG risks regional conflagration, while the TSON potentially could
influence the onset of a world war. The DSN exhibits elements of all of these, as well as the
possibility of creating a new state that could upset the territorial status quo of the state system.
Bargaining and state-building approaches should consider these apart from one another, rather than
as interchangeable extremes on a singular continuum.
Methods of conflict resolution follow logically from ethnic bargaining for our theoretical
understanding. Should bargaining with the center lead to violence, means of terminating it will
likely be considered by relevant actors. Conflict resolution mechanisms for solving identity-based
tensions and conflicts are often viewed as a progression from power sharing to consociationalism
to autonomy, and, in only in the most extreme circumstances, partition.
Each of these mechanisms have key drawbacks that limit their utility in some situations.
Power-sharing alone is difficult to achieve without meaningful commitments to effective
accompanying policies and support.
472
Consociationalism, or the pillar system, requires elite
cooperation and preservation of the status quo. It is often argued for as a midpoint between
disrupting the territorial status quo and minor policies unlikely to succeed.
473
When conflict is
recent, it is more likely to happen again, however.
474
Thus, consociationalism might not present a
lasting solution, especially if the demographic proportions change rapidly or elites are unable to

305

form a productive bargaining arrangement. Autonomy must be granted by the state and supported
by external actors, which means it can be revoked by those same powers.
475
Partition might create
a new unstable state with cascading secessions and a risk of future civil war, thus it is considered
only when genocide is involved, though even such an event is not always guaranteed to lead to
partition.
476

How do these measures get applied by the international system to settle intraethnic
conflicts? US mediation has tried and failed to instill power-sharing mechanisms for Kurdish
factions in Syria, but it successfully helped install two autonomous governments in Iraq for a time.
However, these approaches did not sufficiently account for the transnational aspects of these
conflicts; the PKK was not included in either solution, despite their centrality to the issue at hand.
Which mechanisms are more appropriate to solving organized intraethnic conflict?
Furthermore, do they work differently across types? CSF conflict appears to be solved
through the state-sponsored destruction and suppression of the rival by the Chechen Republic.
Intra-AIF violence is handled by the judicial systems of Kosovo. On the other hand, the KSRs
demonstrate the difficulty of achieving unity even when the same nation was once in the same
state framework. Intra-KPF conflict appears nearly interminable; even the peace between the KDP
and PUK is fragile. Perhaps DSN violence is comparatively more difficult to resolve than the other
types. Future research could explore attempts at achieving conflict resolution for organized
intraethnic conflict across the four types.
9.2 Implications for policy
The most vital implication for policy associated with this work concerns US preferences
for managing conflicts that contain an ethnic component. The US has recognized that there are

306

interethnic tensions in states like Iraq and Afghanistan, yet the intraethnic components appear to
be less well-understood. Consociational systems and a top-down instilling of civic nationalist
frameworks are the preferred methods of managing these situations, but those systems do not
always manage to yield satisfactory results.
In the Soviet-Afghan War, the US backed the various Tajik, Uzbek, Turkmen, and Kyrgyz
factions that became the Northern Alliance, as well as Gulbuddin, an ethnic Pashtun mujahideen
leader that followed an interpretation of Islam that was more fundamentalist than that of tribes
which later made up the bulk of the Taliban. The US and especially Pakistan supported
fundamentalist leaders as a means of weakening Pashtun nationalist factions that wished to create
a state out of Pashtun-inhabited areas of Afghanistan and Pakistan.
477

The precursors to the Taliban were ethnically Pashtun, rural, and nationalist in contrast to
Hekmatyar’s fundamentalist faction. Pakistan began supporting pre-Taliban tribes once the
communist government was overthrown. The US chose not to support conservative rural
mujahideen and so lost the chance to influence what would be the most powerful faction involved
in intraethnic Pashtun conflict.
478

Following the overthrow of the Taliban, old factions came back to power in the post-
invasion government of Afghanistan. Among Pashtuns, the US opted to back ex-royalist
mujahideen, urban ex-communists (America’s indirect foe in the Soviet-Afghan War), and
relatively cosmopolitan figures who promised to deliver progress with regards to Western ideals.
However, the US still did not identify the remaining conservative nationalist tribes
479
as viable
proxies. Eventually, Pakistan did, and in so doing steered them toward the Taliban and steered the
Taliban away from nationalist ambitions. The Taliban was even able to make inroads with non-
Pashtuns over time as they became less nationalistic. Following the withdrawal of US forces the

307

Taliban swept the government from power in days and with them all progress toward those
Western ideals that determined which factions received support.
The KDP represents a stark contrast to the Pashtun-dominated Taliban, though both are
key DSN actors. Also rural, conservative, and nationalistic, the KDP has been supported by the
US for decades. Issues like honor killing and female genital mutilation remain, but the KDP has
made some progress toward ending these and implementing Western projects like empowering
women in politics and economics and implementing protection for members of the LGBT
community. Though certainly no Western cosmopolites, the US has long recognized that the KDP
is the faction that provides the most stability and the greatest likelihood of sustainable, if slow,
social progress. In Afghanistan, US policymakers tried to remake that state in ways that proved to
be unrealistic in scope and rate, as evidenced by the Taliban’s staying power, outreach, and
eventual rapid defeat of the government.
Likewise in Nigeria, we can understand why the world order does not want to embrace
nationalistic movements that might redraw the borders of states. However, the alternative has not
been the rise of a cosmopolitan, pro-Western Kanuri movement; it was the creation of Boko
Haram, which combines the worst tendencies of DSN state-making with the worst aspects of
fundamentalist religious legal systems. Additionally, Boko Haram has destabilized the Sahel
anyway.
The most urgent policy implication of this work is to suggest that the international order
should consider more closely the consequences of choosing the wrong side of intraethnic conflicts,
particularly when DSNs are involved. Many of the most pressing crises of our time include DSNs
in some regard. Thus, stateless irredentism must be approached differently than irredentism
involving states or single-state separatism. As subnational identity increases globally and top-

308

down-created civic nationalist identities further weaken, this type of conflict may become more
frequent and destructive.
Realistic winners should be backed in the early stages of such conflicts. Preferred US allies
in Afghanistan, like ex-royalists and urban ex-communists, may be the most amenable to Western
desires for socio-cultural reasons. However, the evidence that they could not win against the
Taliban was there all along; the conservative rural mujahideen did the heaviest fighting and
suffered the worst casualties of the Soviet-Afghan War, while their eventual competitors were
significantly less effective and dedicated.
Though rural conservative nationalists are not viewed as natural allies due to their
sociocultural views, they appear to be the factions that are most prepared to engage in decades-
long struggles, even after others have given up. Other actors, like Pakistan and Al-Qaeda, have
proven able to sway them away from nationalism and toward fundamentalism; perhaps the
international order can guide them toward a more responsible form of nationalism. A minor victory
with potential for advancing Western socio-cultural preferences at a much later time is preferable
to a costly defeat which erases all top-down implementation of progress like what occurred in
Afghanistan.
A second implication for policy concerns states engaging in military intervention against
DSN state-building. Divide et impera is clearly not an easy task when DSN factions are involved.
Though taking advantage of intraethnic tensions certainly occurs, initiating conflict appears
unlikely for state actors to induce reliably. It seems advantageous to avoid targeting unaligned
members of the group in order to avoid drawing in even more foes. Unfortunately, it seems that
non-combatants might be seen as fair game if they are aligned with an enemy faction. The murders

309

of high-profile non-combatant Kurdish activists were recorded on video yet did not lead to
cooperation like the murders of unaffiliated Kurds in Şingal and Kobanî did.
The final implication concerns DSN factions themselves. Turkish-backed factions in Syria
have provided the bulk of the forces that actually engaged in combat with the PYD’s YPG forces.
KNC-aligned factions could not offer assistance without the risk of engaging in conflict against
their Turkish allies. However, it can be domestically damaging to refrain from cooperation. If
factions seek to instill feelings of nationalism in their coethnics, then it is logical for those publics
to expect results. Their domestic audiences will likely not care about the system-level imposed
incompatibility as much as elites; rather, the domestic audience will probably be operating under
the emotional affect proposed by the first generation of literature.
DSN and TSON factions will likely not wish to compromise on these incompatibilities,
and so they may wish to look to the reunification of Germany for inspiration. Unfortunately for
one faction, this means capitulation and absorption. However, if the elites of the capitulating
faction are integrated into the power structure of the absorbing faction and the political, economic,
and socio-cultural systems are preserved and gradually phased out, there might be beneficial
outcomes for the ethnic group as a whole. For instance, the ENKS is not seen as an existential
threat to Turkey. If KNC-aligned groups merge with KCK-aligned groups, there could be a
favorable outcome in Syria and Iraq. The international order would also have an improved position
for steering these factions if there is ideological heterogeneity.
A limitation and opportunity
The primary limitation of this work was its inability to study displacement as an effect of
intraethnic conflict, only as a potential cause of it in the process tracing phase. We saw that

310

displacement had its most consequential impacts on divergent factional development when DSNs
were involved.
However, what does recent displacement mean for intraethnic conflict? We know that
forced displacement and refugee movements can contribute to conflict spillover.
480
How do
receiving factions view coethnic refugees? From the opening vignette involving the Rojave
Peşmerge, we see that some Kurdish factions have successfully created militias out of displaced
coethnics. Is this generalizable to other DSNs?
Furthermore, what does displacement mean more generally for intraethnic conflict? Is it
viewed as a potential source of power, rather than destabilization? Though these questions are
outside of the scope of this work, they present yet more opportunities for research with regards to
factional conflict. There exist a great many fragile, ethnically diverse states whose non-dominant
groups harbor varying levels of self-determination movements. Each one is a potential avenue for
study. There are opportunities for collaboration between scholars whose work involves conflict,
state-building, identity, and migration. Thus, this limitation is also a chance to further our
knowledge of this understudied phenomenon.
9.3 Outlook for organized intraethnic conflict, cooperation, and resolution
This work has systematically studied organized intraethnic conflict as carried out by
ethnonationalist factions claiming to represent entire ethnic groups. It approached this
phenomenon categorically and proposed the Conflict and Cooperation Theory as a means of
predicting outcomes for a large number of potential cases. It examined the roots of such conflict,
the objects that factions actually fight over, and conditions for why they may or may not cooperate
against outsiders. The divided stateless nation was proposed as a configuration of statehood status

311

and territorial division that is distinct from single-state separatists, divided irredentist groups, and
divided societies based on results of tests pertaining to the propositions set forth in this work.
There are immense challenges posed to the international order by divided stateless nations
that do seek to create a state or statelike entity out of existing polities. However, the stakes of
choosing the wrong sides are immeasurably high. Consider the cases that should be compared to
the case of the Kurdish factions representing the DSN in future work, like Pashtun and Kanuri
factions. Humanitarian disasters, mass displacement, and ethnic war can occur when the most
amenable factions are supported yet fail to win the organized intraethnic conflict component.
However, as the case of the Kurdish factions demonstrate, there are viable factions that can be
seen as “good enough.” Rather than losing lives, decades of effort, untold resources, and reputation
as with the case of Afghanistan, there may be a way to achieve diminished, yet sustainable, goals.

312

Chapter 1

1
Author’s interviews with members of the Êzîdî community. Interviews with most individuals may not be
directly referenced, in part to protect their identities. Some are referenced, but may be anonymized to protect
their identity unless otherwise noted.
2
Kataoka (1974).
3
See Buzan (1995) and Buzan & Wæver (2009). Some scholars also make a further distinction between a
bureaucracy and a nation-state as the second and third levels, respectively. For more information, see
Temby (2013).
4
See also Posen (1993).
5
Zhao & Sutter (1991).
6
Gustafson & Hagström (2017).
7
Brown (2010); Dabed (2010).
8
Williams & Guttschuss (2012).
9
Zhao & Sutter (1991).
10
Lilja & Hultman (2011).
11
Kalyvas (2008).
12
Much of the history of intra-Pashtun conflict has historically pitted the more urban and cosmopolitan
Durrani tribal confederation against the predominantly rural Ghilji confederation. Factions within those
confederations have more recently produced opposing strands of communists and nationalists, as well as
opposing mujahideen fighters. Broadly speaking, Pakistan opted to support the factions that would become
the Taliban, while the US and other actors supported Pashtun ex-communist, ex-royalist, and radical
Islamists, as well as non-Pashtun mujahideen fighters. The Taliban, though radical by most standards, is
less so than other factions, like Hezb-i Islami Khalis under Gulbuddin Hekmatyar. Pashtun nationalism was
more central to the Taliban and its predecessors until somewhat recently. For more on the history of Pashtun
conflicts, see Abbas (2014) and the works in Green’s (2017) edited volume.
13
Simmons (2005).
14
See also Kaldor (2013) and Mello (2010). While an influential argument that these varieties of conflict
are unprecedented, there are detractors. Chojnacki (2006), for instance, finds that this variety of warfare is
simply no longer found in wealthy states.
15
See Naím (2005) for an explanation of how organized criminal groups challenge the sovereignty of
nation-states.
16
Tilly (1985).
17
Bieber (2018); Gruffydd-Jones (2017).
18
Axelrod’s (1984) exposition on cooperation describes how reciprocity encourages cooperative behavior
by punishing defection. While reciprocity also regulates and encourages peaceful interaction between
groups (Fearon & Laitin, 1996), the ability to locate co-ethnics with greater facility should logically enforce
the utility of reciprocity within groups to an even greater extent.
19
See also Jinadu (2004).
20
Riphenburg (2005).
21
Skjelderup et al. (2020).
22
The case selection strategy goes into greater detail about how such cases were dealt with.
23
Suhrke (1975).
24
Horowitz (1985).
25
Zhao & Sutter (1991).
26
Asal et al. (2012); Lilja & Hultman (2011).
27
Olson (2002).
28
Warren & Troy (2015).

313

29
Caspersen (2008).
30
Horowitz (1985). This model of outbidding is associated with the primordial view of ethnicity (Geertz,
1963), in which group identity is static, fixed, and ancient. This is in contrast to the constructivist
perspective. The latter view maintains that ethnicity is constructed to serve a particular function, like
industrialization, or cultural and linguistic standardization (Anderson, 2006; Gellner, 1983). Ethnic conflict
may be a modern phenomenon in which ancient symbols are instrumentalized to gain support (Kaufman,
2001). It is beyond the scope of the dissertation to solve this debate, which may contain an unfair treatment
of primordialism. See James & Mohammadian (2017) for a critique of approaches to primordialist and
constructivist approaches to the study of ethnic conflict.
31
Lyall (2010).
32
Warren & Troy (2015).
33
The debate of whether ideational or material factors drive the phenomena of civil war and ethnic conflict
is characterized as being between greed (material) or grievance (ideational) motivators. Collier and Hoeffler
(2004) and Keen (2008; 2012) are seen as some of the most prominent proponents of greed (or opportunity)
models, in which warlords weigh the incentives for self-enrichment against the risks involved in taking
opportunities to seize resources. Gurr (1970; 1993) and Stewart (2008), who are placed in the grievance
camp, claim that violent mobilization occurs due to inequalities across groups. See Taydas et al. (2011) for
a critique of the perpetuation of an exclusionary approach of arguing for the utility of one school of thought
over the other.
34
Kalyvas (2008).
35
Staniland (2012).
36
Gleditsch (2016); Zhao & Sutter (1991).
37
Posen (1993, p. 32).
38
Carment (1994); Carment & James (1995; 1997); Carment et al. (2006; 2009).
39
Lake & Rothchild (1996).
40
Gilpin (1981); Vasquez (1995).
41
Simmons (2005, p. 823).
42
Additionally, war and trade are seen as reciprocally discouraging, thus incentivizing an enduring peace.
See Anderton & Carter (2001).
43
Gibler (2012); Gibler & Tir (2010).
44
Bercovitch et al. (1997, p. 753).
45
Goertz & Diehl (1997).
46
Gu (1995).
47
Tian (2006).
48
Kalyvas (2008); Lyall (2010).
49
Bates (2008).
50
Mueller (1970).
51
Gurr (1970; 1993); Olson (2002).
52
Tarrow (1994).
53
Kalyvas (2003; 2006).
54
Riphenburg (2005).
55
Kalyvas & Kocher (2007).
56
Seymour et al. (2015).
57
Tilly (1985).
58
Gu (1995).
59
McLauchlin & Pearlman (2012).
60
These were identified by Fearon & Laitin (1996) and Habyarimana et al. (2009).
61
For more on humanitarian grand strategies, see Barnett & Snyder (2018).
62
See Messner et al. (2018) for a categorization of confidence in the long-term sustainability for states in
the international system.

314

Chapter 2

63
Sil & Katzenstein (2010).
64
McLauchlin & Pearlman (2012); Seymour et al. (2015).
65
Diehl & Goertz (2012); Valeriano (2012).
66
Chandra (2006).
67
Tajfel & Turner (1979) developed social identity theory in the field of psychology. It has been used in
the discipline of International Relations, particularly for the use of state identity. For more information on
that particular application, see Mercer (1995).
68
Tajfel & Turner (1979).
69
Horowitz (1981).
70
Kalyvas (2008); Lyall (2010).
71
Tilly (1985); Weber (2015).
72
Eliassi (2016).
73
Licklider (1995); Mukherjee (2006).
74
Carment (1993); Posen (1993); Suhrke (1975).
75
Gu (1995); Zhao & Sutter (1991).
76
The roots of New Wars are typically explained through the lenses of greed versus grievance, as briefly
characterized in note 33. Greed-based explanations are often a first image explanation, in which enterprising
individual actors motivate their followers to commit violence against out-groups and in-group rivals as a
means of achieving personal enrichment or leadership goals. These explanations are most similar to the
first image accounts of intraethnic conflict that involve individual leadership disputes. By contrast,
grievance explanations for New Wars focus on the relative deprivation between and among groups (Gurr,
1970). Factions of ethnic groups must contend with rival factions who have different levels of
organizational capacity. The state may also be a source of grievances, or it may inflame grievances against
other factions. This second image explanation is most similar to state manipulation of ethnonationalist
factions for the purposes of divide et impera. For more on this matter, see Buhaug et al. (2014). They argue
that grievance-based explanations should not be discounted out of hand in favor of those based on
assumptions of greed. This work does not take a side in the debate, as doing so is outside of the scope.
77
See notes 33 and 76.
78
Carment et al. (2009); Saideman (2002).
79
This form is not exclusive to New Wars, however. The world wars, for example, also contained
irredentist elements.
80
Simmons (2005).
81
Merton (1968).
82
Sil & Katzenstein (2010, p. 421).
83
Thies (2004).
84
Asal et al. (2012); Caspersen (2008); Lilja & Hultman (2011).
85
Kalyvas (2008); Lyall (2010). We would not expect this under normal conditions, i.e., those conditions
of pacific relations we infer from Fearon & Laitin (1996) and Habyarimana et al. (2009).
86
Gu (1995); Zhao & Sutter (1991).
87
These rivalries did sometimes terminate through unification or, rather reunification. Reunification,
however, did not mean blending systems for the two Germanys or the two Vietnams; it meant absorbing
the vanquished into the victor on the victor’s terms.
88
Cederman et al. (2009).
89
This was tested and empirically supported by Quackenbush (2016). Furthermore, individual regimes may
be toppled by taking control of the capital (Biddle, 2004, p. 7).
90
Maitre (2009); Vinci (2006).
91
Fearon (2004).
92
Cederman et al. (2013).
93
This is not to say that there would be no risk at all.

315

94
Carment et al. (2006).
95
Posen (1993, p. 32).
96
Walzer (2000, p. 107).
97
See also Tufte (2006).
98
Military capacity is not the only limiting factor. Political realities may limit the likelihood of conflict
before it begins. The cost involved might be prohibitive as a society has only so many resources to spare.
See Wohlstetter (1968).
99
These aspects of conflict can be graded categorically. See Boulding (1962).
100
See notes 33 and 76.
101
However, even if coveted by a rival, these targets may simply be too well-guarded by local or
extraregional actors to be directly attacked. This is especially true of oil refineries and other highly
lucrative sources of wealth. See Frynas & Wood (2001).
102
Cederman et al. (2010); Fearon & Laitin (2000); Park (2011).
103
Bercovitch et al. (1997); Gibler & Tir (2010); Goertz & Diehl (1997).

Chapter 3

104
Hayes & James (2016); Sil & Katzenstein (2010).
105
Agnew (1994, p. 68).
106
Buhaug (2006) and Buhaug & Gates (2002) found that the size of a conflict zone and the presence of
natural resources influenced the scope of a conflict. Cederman et al. (2009) utilized GIS to infer that
distance from the national capital and marginalization of ethnic communities increase the probability of the
onset of internal conflict. Buhaug et al. (2014) used GIS to operationalize grievance-based motives, which
were understudied in the literature in favor of greed, or opportunity, factors.
107
Gleditsch & Weidmann (2012).
108
Branch (2016).
109
Bol (2013).
110
Starr (2013).
111
Schultz (2017) applied GIS to the study of interstate war, while Cederman et al. (2009) applied GIS to
the role of ethnic dyads in internal conflict. Ethnic cleavages and identity may be more influential than the
literature that favors opportunity-based motivations recognized. See also Buhaug (2006) and Buhaug et al.
(2014) for more applications of GIS to internal conflict.
112
These approaches were described by Anselin (1998).
113
Pettersson & Magnus (2020); Sundberg & Melander (2013).
114
Raleigh et al. (2010).
115
Dixon & Sarkees (2015).
116
Sarkees & Wayman (2010).
117
See notes 88 and 106.
118
Global Administrative Areas (2018).
119
Yu et al. (2015).
120
Maitre (2009); Vinci (2006).
121
Humanitarian Data Exchange (2020); NextGIS (2021).
122
Wilson (2007).
123
Vogt et al. (2015); Wucherpfennig et al. (2011).
124
For more on the criteria necessary for map processing, see also Chiang & Knoblock (2015); Chiang et
al. (2016); Kerle & de Leeuw (2009).
125
For an overview of various georeferencing definitions and processes, see Hackeloeer et al. (2014).
126
White & Griffin (2013).
127
King et al. (1994).
128
The sources for these data are described earlier in this section.

316

129
Fotheringham & Wong (1991); Openshaw (1984).
130
There are ways to adjust the bins, of course. Additionally, there are other types of choropleth maps to
convey data differently, but each has its own drawbacks that accompany their advantages. For more
information, see Barreto et al. (2018).
131
Robertson & Feick (2018, p. 456).
132
O’Sullivan & Unwin (2010).
133
Tobler (1979). Neighborhoods are also important when considering conflict events; some regions
(neighborhoods) are more or less conflict-prone than others. See Wohlstetter (1968).
134
Moran (1950).
135
Tsai (2012).
136
Getis & Ord (1992); Harris et al. (2017).
137
Anselin et al. (1996).
138
Ibid. See also Breusch & Pagan (1979).
139
Gramacki (2018).
140
Warton & Shepherd (2010).
141
Ibid.
142
O’Sullivan & Perry (2013).
143
The information on Love Canal that follows comes primarily from Colten & Skinner (2010).
144
Lawless (1986); Renner & Wharton (2013).
145
Many, but not all, were conducted remotely due to travel restrictions during the COVID-19 pandemic.
146
Branch (2014).
147
Tufte (2006, p. 43).
148
Barreto et al. (2018).
149
Producing maps at weekly time steps would also be prohibitively labor-intensive, while not adding
enough value to the study of these phenomena to justify the undertaking at this stage of research.
150
They were also chosen for different shades of intensity in grayscale for color-blind readers.
151
Gerring & Cojocaru (2016).
152
Selecting on the dependent variable biases results. See King et al. (1994).
153
Kalyvas (2008); Staniland (2012).
154
Carment & James (1995; 1997); Posen (1993); Suhrke (1975).
155
Gu (1995); Zhao (1991).
156
Marshall (2017; 2020); Marshall & Elzinga-Marshall (2017).
157
Cederman et al. (2010).
158
Fearon and Laitin (2000).
159
Carment et al. (2009); Saideman (2002).
160
Though there was a significant amount of air support from NATO.

Chapter 4

161
Ajansa Nûçeyan a Firatê [ANF] (2018, August 2); Rudaw (2015, August 3).
162
Interviews with members of Kurdish-speaking non-Sunni communities revealed a complex system of
association with the Kurdish identity, as well as distancing from it.
163
Hennerblicher (2012).
164
Bois (1966); van de Mieroop (2015).
165
Ferdowsi (2006).
166
Abrahamian (2006); Spät (2018).
167
van Bruinessen (1991).
168
Burns (2005).
169
Monshi (1979).
170
Bidlisi (2005).

317

171
Interviews with members of both Kurdish and Êzîdî communities would sometimes include mention of
a particular stran or qawal, which is a variety of hymn.
172
Özoğlu (2004).
173
Olson (1989).
174
Bloxham (2007).
175
Yavuz & Gunter (2001). See also McDowall (2004) for more on historical Kurdish militias that have
served out-group forces against other Kurds.
176
Due to the inclusion and exclusion of areas of mixed population, this figure is most likely inexact.
177
Kurds were also located in Armenia and Georgia after these treaties. However, Kurds in those states are
predominantly urban in distribution, rather than regionally based, and do not advocate for violent self-
determination. I also devote significantly more space to the experiences of Kurds in Syria and Iraq than
Turkey and Iran because those are the areas for which the analyses are performed. This is because intra-
Kurdish factional conflict over territory and population since the end of the Cold War has taken place in
those two states. The village guards of Turkey are not the armed forces of a particular Kurdish political
party, and so that is not sufficiently similar to the conflicts of the major KPFs.
178
McDowall (2004). See also Barzani (2003); Gunter (2014).
179
By “new” and “old” I refer to the dominant group in each state. Turks had long been in charge of the
Ottoman Empire, while Arab-speakers were only recently in control of the new states of Iraq and Syria.
Iranians and Turco-Persians had been in charge of earlier Persian empires.
180
Some relocation was forced, but some was due to economic necessity. See Güvenç (2011).
181
van Bruinessen (1992).
182
Öcalan (2017).
183
Jwaideh (2006).
184
For more about PJAK, see Free Life Party of Kurdistan [PJAK] (2020).
185
Komala’s history and goals are outlined in Komala (2020)
186
The PDKI is also known as the KDPI or KDP-I. They describe their founding in Democratic Party of
Iranian Kurdistan [PDKI] (2020).
187
Gunter (2008).
188
Ibid.
189
Wilson (2007).
190
Yildiz (2012, p. 65)
191
Gunter (1996).
192
For more information about the political-military, economic, social, and cultural ties between the KRG,
and specifically the KDP, and the Turkish state, see Ahmed (2012); Yildiz (2012); Zedalis (2012).
193
Kurdistan National Council [KNC] (2020). See also Gunter (2013).
194
Hama et al. (2018).
195
Gunter (2014).
196
Ibid.
197
Knapp et al. (2016). An all-female unit, the YPJ, was established in 2013.
198
The YPG, though outnumbered and outgunned, defeated ISG forces. However, they were assisted by
Peşmerge from the KRG. See Gunes & Lowe (2015).
199
ICG (2018, February 20).
200
Ibid.
201
Ibid.
202
Ibid.
203
Rudaw (2017, March 5); Sinjar Resistance Units [YBŞ/YJŞ] (2017, March 4). The YJŞ is now known
as the YJÊ.
204
For more information about Operation Euphrates Shield, Operation Olive Branch, and Operation Peace
Spring, see Syrian Observatory for Human Rights [SOHR] (2019, May 5; 2020, October 9).
205
Romano (2020).
206
Aydoğan (2020).

318

207
Rogg & Rimscha (2007).
208
To use the example of the GED, it would be as if only a handful of events were included and each one
was georeferenced to the centroid of Iraq (the geographic center of the areal unit) because the source
material only mentioned the state name.
209
For more on the enmity between the PUK and the IMK, as well as Ansar al-Islam, see Human Rights
Watch [HRW] (2004).
210
Recall that this a key external shock according to Fearon (2004).
211
Gunter (1996).
212
Though no longer in control of some of their cantons, those areas are still claimed by the PYD. For more
information, see Democratic Union Party [PYD] (2020).
213
Gunes & Lowe (2015).
214
McDowall (2004).
215
For more on the nature of the personal relationship between these two men, see Ibrahim (2020) and
Kutschera (1994).
216
For background on the Iraqi-Kurdish wars, see Harris (1977).
217
Ibid.
218
Romano (2006).
219
Kurdistan Democratic Party [KDP] (1996, March 20; 1996, September 12).
220
Patriotic Union of Kurdistan [PUK] (1996, October 28).
221
Wheeler (2002).
222
KDP (1997, July 1); PUK (1996, May 26).
223
Sayari & Hoffman (1991).
224
Gunter (1996); Koontz (2019).
225
KDP (1997, July 1); PUK (1996, October 28); Rudaw (2017, March 5); YBŞ/YJŞ (2017, March 4).
226
PUK (1996, October 28).
227
KDP (1997, July 1).
228
ANF (2017, March 13).
229
Chiefly Iran, but also other unspecified states. See Rudaw (2017, March 5).
230
ANF (2020, February 20; Rudaw (2017, March 5).
231
KDP (1997, July 1); PUK (1996, October 28).
232
ANF (2017, March 13); Hawar News Agency [ANHA] (2017, March 2); KDP (2017, March 20); Rudaw
(2017, March 5).
233
ANF (2017, March 13); KDP (2017, March 4)
234
This was frequently a contentious subject in interviews. As an example, an interview with an Êzidî man
named Faisal demonstrated that whether an individual Êzidî identified as a Kurd or not often depended on
whether they later became affiliated with Kurdish and Kurdish-sponsored forces that would come to be
established in the area.
235
Murad (2017).
236
Gunter (2015).
237
KDP (2017, March 4).
238
Glavin (2015).
239
ANF (2020, February 22).
240
Kalyvas (2006).
241
This would not last for long, however, as evidence by the Sinjar Clashes.
242
Arafat (2016).
243
Rudaw (2017, March 3); YBŞ/YJŞ (2017, March 4).
244
KDP (2017, March 4).
245
Aliyev (2020, June 17)
246
In keeping with other KCK-aligned groups there is also an all-female parallel force, the YJÊ. For more
information, see Kurdistan Communities Union [KCK] (2018, March 23).
247
HRW (2020).

319

248
ANF (2020, November 12); KDP (2019, October 15).
249
Salehyan & Gleditsch (2006).

Chapter 5

250
Jaimoukha (2005, p. 13).
251
Georgian Kists are technically Chechens, but many consider themselves Georgian by nationality. Some
Kists have taken part in conflicts involving Chechen factions, however those Kist fighters are more likely
to fight in Syria or elsewhere in the Middle East than they are in Chechnya.
252
Ingush are also a Vainakh people, as are Kists and Bats. The Vainakh peoples were once politically
linked, but the rise of Chechen nationalism in the 19th century cemented the differentiation between the
two peoples.
253
For more on the origins of Nakh people, see Balanovsky et al. (2011) and Nasidze et al. (2004).
254
Early references to the Chechens in the historical record come from the Georgian Chronicles and the
Armenian Chronicles. See translations and treatments of these early medieval works by Bedrosian
(Georgian Chronicle, 1991) and Toumanoff (1963).
255
Sakwa (2005, p. 9).
256
This was learned of in an interview with a member of the Chechen diaspora in Southern California.
257
Jaimoukha (2005).
258
The life of Timur and his conquests of the North Caucasus were recorded in the Zafarama of Yazdi in
1425. See Thackston’s (Yazdi, 1989) translation.
259
Jaimoukha (2005).
260
Hamburg et al. (2004).
261
Gammer (1994).
262
Dunlop (2011).
263
Ibid.
264
Vatchagaev (2014).
265
Gammer (2006).
266
Ibid.
267
Karcha (1959); Wixman (1982).
268
Dunlop (2011).
269
Posen (1993).
270
Gall & de Waal (1998); Hughes (2005).
271
Dunlop (2011).
272
Ibid.
273
Lapidus (1998).
274
Vor v zakone, or thief-in-law, refers to transnational criminal organizations in the former Soviet Union.
There are vor of each ethnicity. Chechen TOC is distinct because they base their TOC organization on the
clan structure, rather than the prevailing vor structure. See Galeotti (2018) for general information with
regards to the vor. See Galeotti (2008) for how the Chechen bratva, or brotherhood, is different from the
vor.
275
Labazanov had served in the Soviet military and was also a convict. Convicts that he freed formed the
core of his forces. See Cornell (2001); German (2003); Hughes (2007).
276
Labazanov himself would die in 1996 as a result of a blood feud, either by a rival Chechen warlord or
the family of a police officer he murdered.
277
See Smith (2014) for Chechnya’s international relations between the First and Second Chechen Wars.
278
Some of the bombings are alleged to have been false flags to give Russian forces a pretense for invading.
See Amnesty International (2006).
279
The Main Directorate of the General Staff of the Armed Forces of the Russian Federation (GRU/GU)
would eventually recruit Chechen fighters to kill the leader of the SPIR, Arbi Barayev. See Center for
Defense Information [CDI] (2003) and Littell (2006).

320

280
Russell (2005).
281
Hughes (2001).
282
Lyall (2010). Some prominent pro-Moscow Chechen units included the Vostok and Sever Battalions.
See Fuller (2007, November 1).
283
Vatchagaev (2014).
284
Hughes (2005).
285
Possibly as the result of a blood feud with Basayev. He was hated by more hardline separatists who
would place bounties on his head. See British Broadcasting Corporation [BBC] (2000, June 12).
286
Hughes (2001).
287
The declaration of the Emirate caused a rift between religious moderates in the ChRI and Salafists. See
Fuller (2007, November 1). Some Chechen factions viewed this development as the FSB splitting the
factions. See RFE/RL (2007, October 22) and Fuller (2010, August 4).
288
Basayev also fought in the First Nagorno-Karabakh War, the Abkhaz-Georgian conflict, and in
Afghanistan. Interestingly, he was supported by the Russian GRU in The Abkhaz-Georgia conflict and the
FSB prior to the First Chechen War. Basayev was eventually killed by the FSB via remote-detonated bomb.
For more on his life and actions, see Civil Georgia (2020, July 14) and Steele (2006, July 10). He also wrote
a book on his philosophy and tactical expertise. See Basayev (2006).
289
Chechen ISG fighters fought fanatically but most were killed off through attrition. See Fuller (2015,
January 2).
290
Souleimanov & Aliyev (2015).
291
He has set out to construct Europe’s largest mosque, as well as engaged in other forms of religious
outbidding. See Whitmeyer (2015).
292
Lieven (1998).
293
Hughes (2001).
294
Grant (2000).
295
See Eckel (2019, August 28) and Maza (2014, July 22) for the impact of the Pankisi Gorge and the
peoples who inhabit it with regards to Chechen conflicts.
296
Batirashvilli was raised a Christian, but eventually converted to Islam following a personal apocalypse.
He was a non-commissioned officer in the US-trained Georgian special operations forces. Batirashvilli was
medically discharged against his will following an injury. He later became the leader of Chechen forces in
the ISG as Omar al-Shishani. He was killed in July 2016. For more on his life and death see Mamon (2015,
July 13) and RFE/RL (2016, July 15).
297
For more on the cleavages that influenced intra-elite divides, see Hughes (2001).
298
Maza (2014, July 22).
299
Labazanov, as an example, was killed. Umarov was the subject of false propaganda claiming he had
surrendered to Kadyrov. At no point was there any attempt at cooperation or reconciliation. The closest
event I could locate was Kadyrov offering medical care to Umarov—who had lost a leg—if he would
prostrate himself and beg for forgiveness before Kadyrov and the Chechen people. See Chechnya Weekly
(2007, November 15).
300
Of the 150,000-250,000 civilians killed throughout the continuum, roughly 40,000 died during the
Second Chechen War. See Kramer (2005).
301
Associated Press [AP] (1994, September 6); Zucchino (1995, February 9).
302
Higgins (1996, January 22).
303
BBC Monitoring (1999, December 23); Koryashkin & Marychev (1999, November 29).
304
State News Agency (2009, December 19).
305
State News Agency (2015, November 23). While the ChRI no longer joined with the Emirate, they did
not fight them either.
306
Kavkaz Center (2017, July 31); RFE/RL (2010, August 4).
307
There is also a difference in the bargaining ability of DSNs relative to SSSs. The SSS is more likely to
in a weaker bargaining position relative to the host state than an ascendant DSN, which can draw upon

321

more sources of support. See Pearlman & Cunningham (2011) for more on bargaining and conflict
processes.
308
Seymour et al. (2016).
309
The Kurdish independence referendum of 2017 is likely an excellent example. Kurdish factions were
unconcerned about Iraqi responses. American, Turkish, and Iranian actions and responses were far more
consequential.

Chapter 6

310
This term itself is contentious due to contemporary Albanian irredentist sentiment. An ethnic Albanian
pop star called Dua Lipa caused controversy over a tweet regarding Albanian nationalism. See Lipa (2020,
July 19) for the original tweet and its reactions.
311
Danver (2013).
312
The Albanians of the Balkans are not to be confused with the Albanians of the Caucasus in what is now
Azerbaijan. There is also a Caucasian Iberia, which is distinct from the Iberian Peninsula.
313
For more on the linguistic evolution of Albanian ethnonyms, see Demiraj (2010) and Lloshi (1999).
314
This work will not delve into the histories of each people, nor their religions, as it is uncertain which is
the progenitor of modern Albanians. For what is known about the post-Roman ethnogenesis of modern
Albanians, see Bowden (2003).
315
For Balkan history in antiquity, see Ptolemy (2000) and Polybius (2010).
316
See Ducellier (1999) and Elsie (2010).
317
Jorga (1908).
318
Skendi (2015).
319
See Frashëri (1964) and Babinger (1992).
320
Albanian irredentist claims for statehood would be inspired by the League of Lezhë, though the League
itself never reached the status of a sovereign state. See Sugar (1977).
321
Mostly comprised of Christians from the Balkans and animist Turkic peoples, the Janissaries were elite
slave-soldiers who fought for the Ottoman Empire. They are similar to—though not interchangeable with—
the mamluks of Egypt. For more, see Goodwin (1997).
322
Albanians in Greece that migrated in the Middle Ages and early modern period are known as Cham. For
more on the Cham see Kretsi (2002).
323
See Masters (2001) for a history of the emergence and failure of this Ottoman system of governance.
324
Allen & Muratoff (1953).
325
Duijzings (2002).
326
Vickers (1998).
327
The Waffen-SS was a multinational military organization that was part of the Schutzstaffel. See Stein
(1984).
328
For more on the reasoning for the expulsion from Chameria, see Baltsiotis (2011).
329
Posen (1993).
330
Iosifides et al. (2007).
331
Hall (2001).
332
Pike (2013).
333
See the entry regarding these groups in the initialisms guide.
334
The parliament of the Republic of Macedonia voted to change its name to the Republic of North
Macedonia. The change was finalized in 2019. As the conflicts studied in this work occurred before the
name change, I refer to the state as Macedonia when appropriate. For more on the use of name changes to
improve international relations and domestic politics, especially as it relates to North Macedonia, see
Dimitrieva & Tiede (2020).
335
For an overview of the Yugoslav Wars, see Baker (2015).
336
For more on the background to conflict in Kosovo and this faction in particular, see Elsie (2011) and
Judah (2002).

322

337
For more on the individual and organizational differences between the LDK and LPK, see Mulaj (2008).
338
See Hedges (1999) and Jonsson (2014).
339
The SHIK is now the SHISH. See Marty (2010).
340
For more on external support to UÇK factions in general and the KLA in particular, see Judah (2002);
Lombardi (2009); Ron (2003).
341
NATO’s primary worry was that the KLA was too polarizing to lead to lasting unity among factions.
The organization was also concerned that Kosovo would become a haven for TOC. See International
Institute for Strategic Studies (1999).
342
Thaçi was later elected president of Kosovo. He was also indicted for war crimes and crimes against
humanity for his role in the KLA. For more on his life, see Boyes & Jagger (2018).
343
Krasniqi (2000, July 18).
344
Reuveny & Thompson (2010).
345
The Yugoslav Wars were replete with atrocities. For more information on the crimes themselves and the
memories of them in Balkan societies, see Obradović (2013).
346
See HRW (2001, October 26) and Haxhiaj (2018, March 6).
347
Morina (2017, July 3).
348
Haxhiaj (2018, March 6).
349
The formal blood feud process is known as Kanun. See Meçe (2017). Though both societies have such
an institution, they are not the same. For similarities and differences between the blood feud systems of
Albanians and Chechens, see ten Dam (2010).
350
See Hislope (2004); Husanović (2000) for descriptions of violence during this period.
351
Other factions arose in the 2000s who still sought to create a Greater Albania. These groups, based in
Kosovo, the Preševo Valley, Montenegro, and North Macedonia would be considered rivals to the UÇK
groups. However, they are unpopular and primarily underground. The concept of Greater Albania is no
longer as prevalent as it once was. See Stojarova (2010).
352
Elsie (2011).
353
Ibid.
354
Perritt (2008).
355
See Bieber & Daskalovski (2004).
356
Direct Albanian intervention involving troops, armor, and artillery occurred during the Battles of Košare
and Paštrik. See Forage (2001).
357
North Atlantic Treaty Organization [NATO] (1999, June 9).
358
Babuna (2000); Hedges (1999). Indeed, some Albanians were enthusiastic about these historical
linkages, however most leaders would have been infants or not even alive yet, clearly unable to serve in the
Waffen SS. See also Koktstidis & ten Dam (2008).
359
Glenny (2012).
360
Henriksen (2013).
361
Reuveny & Thompson (2010).
362
NATO (2001, March 22).
363
Judah (2001, May 3).
364
Naegele (2001, March 21).
365
Majko (1999, March 25).
366
Majko (1999, March 24).
367
ICG (1998, October 1). However, the Albanian government also allegedly allowed the KLA to set up
secret prisons in Albania to detain and torture LDK members. Nevertheless, these incidents appear to have
been classified by the EU’s Special Investigative Task force as isolated incidents attributed to individuals
and not the KLA as a whole. See Haxhiaj (2018, March 6).
368
Naegele (2002, April 17).
369
RFE-RL (2002, April 5). External actors were also concerned about the impact that these individual-
level threats could have on cohesion among Albanian factions. See Naegele (2002, April 16).
370
HRW (2001).

323

371
Bytyci (2010, December 10).
372
Morina (2017, July 3).

Chapter 7

373
See Martin (1998) and Polian (2004) for the deportation of ethnic minorities in the USSR in general and
the Koreans in particular.
374
Chosŏn is another form of this word that is frequently encountered.
375
For more on Korean ethnonyms, see Lee & Lee (2016).
376
Cho & Whitman (2020).
377
For more on the proposed linkages between Koreanic and Altaic languages, see Lee & Ramsey (2011).
378
Goedde et al. (1987).
379
Jin et al. (2009).
380
Nelson (1993).
381
Lee (1984).
382
Ilyon (2008). Ilyon, or Il-yeon, was a Buddhist monk who compiled the stories that were included in the
Samguk Yusa in the 13
th
century.
383
Sima (1993). Sima Tan was a Chinese historian. His works covered the early records of the Han dynasty.
His son, Sima Qian, compiled and printed his father’s work after his death.
384
There was also the confederacy of Gaya in the south, which was eventually absorbed by Silla. See Lee
(1984).
385
Joe (1982).
386
Baker (2008).
387
Some of these Tungusic and Mongol peoples, like the Khitan, would provide the heads of later Korean
dynasties. At times, those same tribes of conquering nomads would also be absorbed by the Korean
population. See Seth (2016).
388
Kim (2012).
389
Yi (1984).
390
Kang (2010).
391
Haboush et al. (2016).
392
This was true for both Protestant and Catholic sects. See Kim (1983).
393
Now called Cheondoism, this form of Neo-Confucianism still exists in both North and South Korea. For
more on Cheondoism, see Yao (2000).
394
For more with regards to this impactful war, see Paine (2003).
395
Caprio (2009).
396
According to the National Bureau of Statistics of China (2019), all non-Han ethnicities in the YKAP
made up a combined 34.49% of the population at the end of 2018. The YKAP is located in Jilin province.
There are also Tungusic and Manchu people in the YKAP, and so the Korean proportion is even lower than
this figure, as it includes all non-Han Chinese peoples.
397
This was once a destination for Koreans fleeing wars between Qing China, Japan, and the Russian
Empire. See Song (2018).
398
Korean Central News Agency [KCNA] (2000, March 22).
399
Yun (2016).
400
Stueck (2002).
401
For more on Kim Il-sung and the beginning of North Korea, see Lankov (2002).
402
Ibid. For instance, Cho Man-sik was originally preferred as a leader by the Soviets, but he was more
nationalistic and not amenable enough to communism, and so was removed and likely killed.
403
Leadership of the WPK is hereditary; the North Korean constitution stipulates that a member of the Kim
family is to be in charge of it. See Isozaki (2017) for more on how succession has been modified over the
decades.

324

404
Juche is more strongly associated with Kim Il-sung, while Songun arose under Kim Jong-il. For more
information of Juche, see North Korea (2012). See the official news organ of the WPK as described in
Young (2021) for more on Songun. These ideologies are largely compatible with each other and are known
as Kimilsungism-Kimjongilism.
405
For more on the doctrinal conflict between these two states, see Li & Xia (2014).
406
See Lew (2013) for more information on Rhee’s life and times. See also Rhee’s (2015) own diary for
insight into his worldview before coming to power.
407
Though founded as a democratic state, democracy was only considered consolidated in 1987. Military
autocracy was instituted through occasional overthrows, though some periods were more democratic than
others. Now in its Sixth Republic, the ROK is considered a mature, stable democracy. For more on the
ROK’s part in the Third Wave of democracy, see Woo (2017).
408
After its establishment by the UNTCOK, the ROK first tried economic policies similar to other post-
colonial states. For instance, import substitution, in which the new state attempts to shore up domestic
industry rather than engage in trade with the wider economic order, was the norm for a time under Rhee.
The early government was also heavily dependent upon US aid. However, Rhee also oversaw the transition
to a model more akin to state capitalism, in which domestic industries were heavily subsidized in exchange
for significant financial support for state policies later. See Ku (2017b) for more on the ROK’s economic
development across time.
409
See Ku (2017a) for more on the relative struggles each state has faced.
410
Japan’s refusal to apologize for abuses like the forced use of “comfort women” factors into the sentiment.
This term refers to enslaved women, in this case Korean, though others were so also taken. They were
forced to serve as prostitutes for Japanese soldiers in the Second World War (Japan claims they were
volunteers). This factors heavily into the negative view of Japan. See Soh (2009).
411
Kang (2009).
412
Chae (2010); Stueck & Yi (2010).
413
The US also kept nuclear weapons in the ROK until 1991.
414
Totten & Jo (1991).
415
This work will not focus on these conflicts because it is concerned primarily with post-Cold War conflict,
and so it presents only a brief depiction of them.
416
The American 5
th
Marine Regiment of the 1
st
Marine Division, as an example, inflicted such severe
losses on the PVA that the entire 9
th
Army Group was too understrength to be used for four months. See
Cohen (2012). For more on the course of the Korean War, see Halberstam (2007).
417
The Korean War also had a major impact in Europe. It likely stopped escalations between the US and
USSR. See Stueck (1995).
418
The remainder of Cold War-era conflict is explained in depth by Buzo (2016).
419
This was done in secret. See Chubb (2014).
420
Harrison (2002).
421
Ibid.
422
Other DPRK figures, such as Kim Jong-nam (Kim Jong-un’s brother), and others have been assassinated.
423
Kwak (2016).
424
O (2015).
425
Kwak (2016).
426
Both the ROK and DPRK would be required to make large concessions based on low trust. See Kwak
(2016) for more information regarding the issues facing the neutralization-unification strategy.
427
Lee (2017).
428
Totten & Jo (1991).
429
Blaming the US for the current predicament is not constrained to the North. US assistance was predicated
upon the ROK normalizing relations with Japan, which angered Rhee. For more on the attitudes among the
Koreans toward the US, see Harrison (2002).
430
These dispositions are drawn from Kim (2016) for the US; Kim (2016) for China; Zhebin (2016) for
Russia; and Kaseda (2016) for Japan.

325

431
See, for example, KCNA (1999; June 19); Ministry of Foreign Affairs [MOFA] (2011).
432
Especially as DPRK artillery is concentrated near the DMZ, with Seoul being the presumed target. For
more on expectations for war between the two states, see O (2016).
433
Bluth (2008); Jung et al. (2021).
434
Particularly if politics were involved. This happened to Lim Su-kyung when she attended the World
Youth Festival in 1989 without the permission of the ROK. See Barraclough (2019).
435
Bluth (2008); O (2016).
436
Bae (2020).
437
Kim, J. (2020, September 25).
438
MOFA (2019).
439
For instance, a Japanese naval patrol aircraft was nearly fired upon by a South Korean naval vessel in
2018. See Lee & Oba (2019).
440
Those living in neighboring states might be considered unaffiliated. More research into the development
of irredentist claims is needed to make this assertion, however.
441
Different donors have different goals, but generally seek to increase the DPRK’s engagement with the
international order. However, it is possible that the DPRK uses violence, such as the naval clashes described
in this chapter, to receive aid that then allows them to become more independent of the system. For more
information, see Jang & Suh (2017).

Chapter 8

442
Though there were occasional skirmishes between members of the KDP and PUK in the 1970s, these
were few in number and likely were not officially encouraged by either party.
443
I include the national capital in the international system given the possible size of a state. Very large
states may technically straddle multiple regions. Indeed, the KPFs are distributed across states variably
considered to be in the Middle East, Central Asia, and the South Caucasus. So, while the term
“neighborhood” might be more accurate, “region” may be imprecise for some DSNs.
444
This finding somewhat vindicates those skeptical of economic motivators being the primary cause of
low-intensity and identity-based conflict, at least in the initial stages of such conflict. See Maitre (2009);
Taydas et al. (2011); Vinci (2006).
445
Early disputes between the KDP and PUK resulted in PUK defeat. Later, the war between these factions
was ended through US mediation. Barzanî and Öcalan—friendly with each other—settled the dispute
between their factions peacefully for a time, until system-level forces drove a renewal of conflict. The KNC-
KCK conflict has not terminated, but for a time it seemed as though US mediation would make some
progress.
446
All variables included in the international system may also be restricted to the region if “region” is
understood to be the geographic “neighborhood.” This would also depend on the case.
447
Such acts would be meant to divert resources from the regional center that the SSS wants to control.
448
Gleditsch et al. (2008).
449
The host state is displayed following conflict onset in figure 8.6 to illustrate that direct intervention can
come later. It is expected to vary by case.
450
The out-group force is displayed in the international system, but it could also be placed in the region.
Like the preceding diagrams this is likely dependent upon the case.
451
The irredentist faction also could initiate violence.
452
Recall the results of the PPP; accounting for affiliation yielded an inferior model compared to the model
that included only non-combatants and rivals.
453
This does not rule out individual-level violence.
454
Such division may even be implemented without input from the group being divided.
455
The overall rarity of the TSON may be the cause of this; only the two Koreas and China-Taiwan remain.
There are many more DSNs in the world.

326

456
Such frameworks have yet to be successfully implemented since the BDR absorbed the DDR. There are
solutions sought for by KSRs, but neither regime-preferred framework has been seriously considered by
the rival. A third, more neutral strategy is impossible to implement without first addressing the altered
information costs and issues of trust that led to the conflict in the first place.
457
The Pashtun-dominated Taliban, for instance, could have captured Kabul much earlier than it did. At
one point, it could have been possible for Boko Haram to conquer more of Nigeria than it wished.
458
Indeed, it seems to have not occurred before the Cold War, either, as the phenomenon of IRB-created
TSONs arose during the Cold War. The Second United Front, for example, was not quite a true TSON.

Chapter 9

459
Weller & Barnes (2014).
460
Particularly as put forth by James (2019); Pfonner & James (2020).
461
Haggard & Kaufman (2016).
462
Lee et al. (2020).
463
Highways were used as ad hoc borders, rather than as objects of contestation.
464
Matloff (2017).
465
Maitre (2009); Taydas et al. (2011); Vinci (2006).
466
However, the democratization of geospatial data carries its own risks. See Robertson & Feick (2018).
467
Recall how the Albanian prime minister viewed the struggle against non-Albanians by irredentist
factions as a struggle for the entire nation. See also note 365.
468
The third generation of research partially incorporates system-level factors but does not focus on them.
469
Kalyvas (2006).
470
Sambanis et al. (2015). See also Saideman & Ayres (2008) for the logic of irredentist state actors.
471
See, for instance, key works addressing the ethnic bargaining literature, at least in part. Cetinyan (2002)
discusses rebellion against the host state but considers separatism and irredentism (termed “those with kin
abroad who control an adjacent state”, p. 659) under the same models of rebellion. Jenne (2015) treats
separatism and irredentism as part of a spectrum, in which they are apparently interchangeably extreme (p.
40; note 59). Jenne et al. (2007) follow the same premise (note 18). Saideman (2003, note 6) also considers
the two concepts to exist in the same vein.
472
See McGarry & O’Leary (2006); O’Leary (2016).
473
Lijphart (1969; 1977); Rossi (2014).
474
Cederman et al. (2009); Fearon & Laitin (2000; 2003).
475
For more on the revocation of autonomy and the relationship between autonomy and secession, see
Siroky & Cuffe (2015).
476
O’Leary (2016); Sambanis & Schulhofer-Wohl (2009).
477
Williams (2011).
478
The US did support the Haqqani Network, which became a Taliban ally much later on. However, that
support stopped long before that faction allied with the Taliban.
479
Many of these tribes came from the Ghilji confederation, which also included some smaller Maoist rural
communists supported by the US during the Soviet-Afghan War, as well some of the fundamentalist
mujahideen factions like Hekmatyar’s. The US, by the end of the Soviet-Afghan War, however, preferred
to support tribes from the Durrani confederation, which included former royalists and urban communists.
See also note 12.
480
Salehyan & Gleditsch (2006).

327

References

Abbas, H. (2014). The Taliban revival: Violence and extremism on the Pakistan-Afghanistan
frontier. Yale University Press.
Abrahamian, L. (2006). Armenian identity in a changing world. Mazda.
Adams, M. (2020). Steppe dreams: Time, mediation, and postsocialist celebrations in Kazakhstan.
University of Pittsburgh Press.
Agnew, J. (1994). The territorial trap: The geographical assumptions of international relations
theory. Review of International Political Economy, 1(1), 53–80.
https://doi.org/10.1080/09692299408434268
Ahmed, M. M. A. (2012). Iraqi Kurds and nation-building. Palgrave Macmillan.
Ajansa Nûçeyan a Firatê. (2017, March 13). YBŞ/YJŞ: Li dijî êrîşan em ê gelê xwe heta dawiyê
biparêzin [Press Release]. Firat News Agency.
https://anfkurdi.com/kurdistan/ybs-yjs-li-diji-erisan-em-e-gele-xwe-heta-dawiye-
biparezin-77440
Ajansa Nûçeyan a Firatê. (2018, August 2). 'Bila statûya Şengalê were naskirin'. Firat News
Agency. https://anfkurdi.com/rojane/bila-statuya-Sengale-were-naskirin-100528
Ajansa Nûçeyan a Firatê. (2020, February 20). PYD: Hevdîtina ENKS’ê xizmetkirina ji bo
dagirkeran e. Firat News Agency.
https://anfkurdi.com/rojava-sUriye/pyd-hevditina-enks-e-xizmetkirina-ji-bo-dagirkeran-
e-124641

328

Ajansa Nûçeyan a Firatê. (2020, November 12). Gelek mîna Mam Wuso hene ku îxaneta PDK’ê
qebûl nakin. Firat News Agency.
https://anfkurdi.com/rojane/gelek-mina-mam-wuso-hene-ku-ixaneta-pdk-e-qebul-nakin-
134792
Aliyev, J. (2020, June 17). Turkey launches Operation Claw-Tiger in northern Iraq. Anadolou
Agency.
https://www.aa.com.tr/en/middle-east/turkey-launches-operation-claw-tiger-in-northern-
iraq/1879479
Allen, W. E. D, & Muratoff, P. (1953). Caucasian Battlefields. Cambridge University Press.
Amnesty International. (2006). Russian Federation: Amnesty International’s concerns and
recommendations in the case of Mikhail Trepashkin.
https://www.amnesty.org/download/Documents/76000/eur460122006en.pdf
Anderson, B. (2006). Imagined communities: Reflections on the origin and spread of nationalism.
Verso.
Anderton, C. R., & Carter, J. R. (2001). The impact of war on trade: An interrupted times-series
study. Journal of Peace Research, 38(4), 445-457.
https://doi.org/10.1177/0022343301038004003
Anselin, L. (1998). Exploratory spatial data analysis in a geocomputational environment. In P. A.
Longley, S. M. Brooks, R. McDonnell, & B. MacMillan, (Eds.), Geocomputation: A
primer (pp. 77-94). Wiley. https://doi.org/10.1016/0166-0462(95)02111-6

329

Anselin, L., Bera, A., Florax, R., & Yoon, M. (1996). Simple diagnostic tests for spatial
dependence. Regional Science and Urban Economics, 26(1), 77–104.
https://doi.org/10.1016/0166-0462(95)02111-6
Arafat, H. (2016, November 22). Thousands of Rojava Peshmerga prevented from battling IS in
Syria. Kurdistan24.
https://www.kurdistan24.net/en/news/4e7394c6-988d-4ffe-b0a5-
c3739d275744/Thousands-of-Rojava-Peshmerga-prevented-from-battling-IS-in-Syria
Asal, V., Brown, M., & Dalton, A. (2012). Why split? Organizational splits among ethnopolitical
organizations in the Middle East. Journal of Conflict Resolution, 56(1), 94-117.
https://doi.org/10.1177/0022002711429680
Associated Press. (1994, September 6). Government troops take opposition stronghold. Times -
Picayune (Pre-1997 Fulltext).
Axelrod, R. (1984). The evolution of cooperation. Basic Books.
Aydoğan, B. (2020, April 15). The Iraqi Kurds’ destructive infighting: Causes and consequences.
London School of Economics.
https://blogs.lse.ac.uk/mec/2020/04/15/the-iraqi-kurds-destructive-infighting-causes-and-
consequences/
Babinger, F. (1992). Mehmed the Conqueror and his time. Princeton University Press.
Babuna, A. (2000). The Albanians of Kosovo and Macedonia: Ethnic identity superseding
religion. Nationalities Papers, 28(1), 67–92. https://doi.org/10.1080/00905990050002461

330

Bae, J. (2020). Limits of engagement? The sunshine policy, nuclear tests, and South Korean views
of North Korea 1995–2013. International Relations of the Asia-Pacific, 20(3), 411–443.
https://doi.org/10.1093/irap/lcz004
Baker, C. (2015). The Yugoslav Wars of the 1990s. Palgrave.
Baker, D. (2008). Korean spirituality. University of Hawaii Press.
Balanovsky, O., Dibirova, K., Dybo, A., Mudrak, O., Frolova, S., Pocheshkhova, E., Haber, M.,
Platt, D., Schurr, T., Haak, W., Kuznetsova, M., Radzhabov, M., Balaganskaya, O.,
Romanov, A., Zakharova, T., Soria Hernanz, D. F., Zalloua, P., Koshel, S., Ruhlen, M.,
…Balanovska, E. (2011). Parallel evolution of genes and languages in the Caucasus
region. Molecular biology and evolution, 28(10), 2905–2920.
https://doi.org/10.1093/molbev/msr126
Baltsiotis, L. (2011). The Muslim Chams of Northwestern Greece: The grounds for the expulsion
of a “non-existent” minority community. European Journal of Turkish Studies, Article 12.
https://doi.org/10.4000/ejts.4444
Barnett, M., & Snyder, J. (2018). The grand strategies of humanitarianism. In M. Barnett & T.
Weiss (Eds.), Humanitarianism in question: Politics, power, ethics (pp. 143-171). Cornell
University Press. https://doi.org/10.7591/9780801461538-008
Barraclough, R. (2019). Political travel at Cold War's end: International student exchanges
between Australia and the two Koreas. In S. Park (Ed.), Revisiting Minjung: New
perspectives on the cultural history of 1980s South Korea (pp. 85-102). University of
Michigan Press. https://doi.org/10.3998/mpub.10027900

331

Barreto, M. C., Kosminsky, D., & Esperança, C. (2018). Hexagonal hierarchical cartogram:
Towards a thematic map of Brazil. InfoDesign-Revista Brasileira de Design da Informação
15(1), 45-62.
Barth, F. (1969). Ethnic groups and boundaries. Little Brown and Company.
Barzani, M. (2003). Mustafa Barzani and the Kurdish liberation movement. Palgrave Macmillan.
Basayev, S. (2006). Book of Mujahiddeen. Kavkaz Center.
Bates, R. H. (2008). The logic of state failure: Learning from late-century Africa. Conflict
Management and Peace Science, 25(4), 297–314.
https://doi.org/10.1080/07388940802397376
Bercovitch, J., Diehl, P. F., & Goertz, G. (1997). The management and termination of protracted
interstate conflicts: Conceptual and empirical considerations. Millennium, 26(3), 751-769.
https://doi.org/10.1177/03058298970260030201
Biddle, S. (2004). Military power: Explaining victory and defeat in modern battle. Princeton
University Press.
Bidlisi, S. K. (2005). The Sharafnama, or, the history of the Kurdish nation (M. R. Izady,
Trans.). Mazda. (Original work published 1597)
Bieber, F. (2018). Is nationalism on the rise? Assessing global trends. Ethnopolitics, 17(3), 519-
540. https://doi.org/10.1080/17449057.2018.1532633
Bieber, F., & Daskalovski, Z. (2004). Understanding the war in Kosovo. Routledge.
Blainey, G. (1973). The causes of war. Free Press.

332

Bloxham, D. (2007). The Great Game of genocide: Imperialism, nationalism, and the destruction
of the Ottoman Armenians. Oxford University Press.
Bluth, C. (2008). Korea. Polity Press.
Bois, T. (1966). The Kurds (M. W. M. Welland, Trans.). Khayat Book & Publishing Company.
(Original work published 1965)
Bol, P. K. (2013). On the cyberinfrastructure for GIS-Enabled Historiography. Annals of the
Association of American Geographers, 103(5), 1087-1092.
https://doi.org/10.1080/00045608.2013.792178
Boulding, K. E. (1962). Conflict and defense: A general theory. Harper.
Bowden, W. (2003). The construction of identities in post-Roman Albania. Late Antique
Archaeology, 1(1), 57-78.
Boyes, R., & Jagger, S. (2018). New state, modern statesman: Hashim Thaçi – A biography.
Biteback Publishing.
Branch, J. (2014). The cartographic state: Maps, territory and the origins of sovereignty.
Cambridge University Press.
Branch, J. (2016). Geographic Information Systems (GIS) in International Relations. International
Organization, 70(4), 845–869. https://doi.org/10.1017/S0020818316000199
Breusch, T., & Pagan, A. (1979). A simple test for heteroscedasticity and random coefficient
variation. Econometrica, 47, 1287-1294. https://doi.org/10.2307/1911963
British Broadcasting Corporation (1999, December 23). Chechen mufti calls for talks with all real
forces in Chechnya to stop the war.

333

British Broadcasting Corporation. (2000, June 12). Russia appoints Chechen leader.
http://news.bbc.co.uk/2/hi/europe/787811.stm
Brown, N. J. (2010). The Hamas-Fatah Conflict: Shallow but wide. The Fletcher Forum of World
Affairs, 34(2), 37-51.
Brubaker, R. (2004). Ethnicity without groups. Harvard University Press.
Buhaug, H. (2006). Relative capability and rebel objective in civil war. Journal of Peace Research,
43(6), 691–708. https://doi.org/10.1177/0022343306069255
Buhaug, H., & Gates, S. (2002). The geography of civil war. Journal of Peace Research, 39(4),
417–433. https://doi.org/10.1177/0022343302039004003
Buhaug, H., Cederman, L.-E., & Gleditsch, K. S. (2014). Square pegs in round holes: Inequalities,
grievances, and civil war. International Studies Quarterly, 58(2), 418-431.
https://doi.org/10.1111/isqu.12068
Bunge, M. (1996). Finding philosophy in social science. Yale University Press.
Burns, R. (2005). Damascus: A history. Routledge.
Buzan, B. (1995). The level of analysis problem in International Relations reconsidered. In K.
Booth & S. Smith (Eds.), International Relations theory today (pp. 198-216). Polity.
Buzan, B., & Wæver, O. (2009). Macrosecuritisation and security constellations: Reconsidering
scale in securitisation theory. Review of International Studies, 35(2), 253–276.
https://doi.org/10.1017/S0260210509008511
Buzo, A. (2016). The making of modern Korea (3
rd
ed.). Taylor & Francis.
https://doi-org.libproxy1.usc.edu/10.4324/9781315688985

334

Bytyci, F. (2010, December 10). Kosovo PM denies "mafia-style" crime allegations. Reuters.
https://reliefweb.int/report/serbia/kosovo-pm-denies-mafia-style-crime-allegations
Caprio, M. E. (2009). Japanese assimilation policies in colonial Korea, 1910-1945. University of
Washington Press.
Carment, D. (1993). The international dimensions of ethnic conflict: Concepts, indicators, and
theory. Journal of Peace Research, 30(2), 137–150.
https://doi.org/10.1177/0022343393030002002
Carment, D. (1994). The ethnic dimension in world politics: Theory, policy, and early warning.
Third World Quarterly, 15(4) 551–582. https://doi.org/10.1080/01436599408420398
Carment, D., & James, P. (1995). Internal constraints and interstate ethnic conflict: Toward a
crisis-based assessment of irredentism. Journal of Conflict Resolution, 39(1), 82–109.
https://doi.org/10.1177/0022002795039001004
Carment, D., & James, P. (1997). Wars in the midst of peace: The international politics of ethnic
conflict. University of Pittsburgh Press.
Carment, D., James, P., & Taydas, Z. (2006). Who intervenes? Ethnic conflict and interstate crisis.
Ohio State University Press. https://doi.org/10.1111/j.1468-2486.2008.01825.x
Carment, D., James, P., & Taydas, Z. (2009). The internationalization of ethnic conflict: State,
society, and synthesis. International Studies Review, 11(1), 63-86.
https://doi.org/10.1111/j.1468-2486.2008.01825.x

335

Caspersen, N. (2008). Intragroup divisions in ethnic conflicts: From popular grievances to power
struggles. Nationalism and Ethnic Politics, 14(2), 239-65.
https://doi.org/10.1080/13537110801985070
Cederman, L.-E., Buhaug, H., & Rød, J. K. (2009). Ethno-nationalist dyads and civil war: A GIS-
based analysis. Journal of Conflict Resolution, 53(4), 496–525.
https://doi.org/10.1177/0022002709336455
Cederman, L.-E., Gleditsch, K.S., Salehyan, I., & Wucherpfennig, J. (2013). Transborder ethnic
kin and civil war. International Organization, 67(2), 389–410.
https://doi.org/10.1017/S0020818313000064
Cederman, L.-E., Wimmer, A., & Min, B. (2010). Why do ethnic groups rebel? New data and
analysis. World Politics, 62(1), 87–119. https://doi.org/10.1017/S0043887109990219
Center for Defense Information. (2003). In the spotlight: The Special Purpose Islamic Regiment.
Center for Defense Information Terrorism Project.
https://www.webcitation.org/603DwNZkm?url=http://www.cdi.org/terrorism/spir.cfm
Çetin, F. (2013, November 18). Devrimine karşı Erdoğan-Barzani işbirliği. The Rojava Report.
https://rojavareport.wordpress.com/2013/11/18/erdogan-and-barzani-cooperating-against-
the-rojava-revolution/
Cetinyan, R. (2002). Ethnic bargaining in the shadow of third-party intervention. International
Organization, 56(3), 645–677. https://doi.org/10.1162/002081802760199917
Chae, H. (2010). South Korean attitudes toward the ROK–U.S. alliance: Group analysis. Political
Science and Politics, 43(3), 493–501. https://doi.org/10.1017/S1049096510000727

336

Chandra, K. (2006). What is ethnic identity and does it matter? Annual Review of Political Science,
9(1), 397-424. https://doi.org/10.1146/annurev.polisci.9.062404.170715
Chechnya Weekly (2007, November 15). Kadyrov offers Umarov medical care. North Caucasus
Weekly, 8(44). https://jamestown.org/program/kadyrov-offers-umarov-medical-care/
Chiang, Y.-Y., & Knoblock, C. A. (2015). Recognizing text in raster maps. GeoInformatica, 19(1),
1–27. https://doi.org/10.1007/s10707-014-0203-9
Chiang, Y.-Y., Leyk, S., & Knoblock, C. A. (2014). A survey of digital map processing techniques.
ACM Computing Surveys, 47(1), 1-44. https://doi.org/10.1145/2557423
Chiang, Y.-Y., Leyk, S., Narges, H. N., Sima, M., & Tian, X. T. (2016). Assessing the impact of
graphical quality on automatic text recognition in digital maps. Computers and
Geosciences, 93(C), 21–35. https://doi.org/10.1016/j.cageo.2016.04.013
Cho, S., & Whitman, J. (2020). Korean: A linguistic introduction. Cambridge University Press.
Chojnacki, S. (2006). Anything new or more of the same? Wars and military interventions in the
international system, 1946–2003. Global Society, 20(1), 25–46.
https://doi.org/10.1080/13600820500405442
Chubb, D. (2014). Contentious activism and inter-Korean relations. Columbia University Press.
Civil Georgia. (2020, July 14). Ex-FSB head says Shamyl Basayev cooperated with Russian
intelligence in Abkhazia. https://civil.ge/archives/359380
Cohen, E. A. (2012). Military misfortunes: The anatomy of failure in war. Simon and Schuster.
Collier, P., & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers,
56(4), 563-95. https://doi.org/10.1093/oep/gpf064

337

Colten, C. E. & Skinner, P. N. (2010). The road to Love Canal: Managing industrial waste before
EPA. University of Texas Press.
Cornell, S. (2001). Small nations and great powers: A study of ethnopolitical conflict in the
Caucasus. Curzon.
Dabed, E. (2010). Decrypting the Palestinian political crisis: Old strategies against new enemies:
Chile 1970-73, Palestine 2006-09. Arab Studies Quarterly, 32(2), 73-91.
Danver, S. L. (2013). Native peoples of the world: An encyclopedia of groups, cultures, and
contemporary issues. Sharpe Reference.
Demiraj, B. (2010). Shqiptar–The generalization of this ethnic name in the XVIII century. In
B. Demiraj (Ed.), Wir sind die Deinen: Studien zur albanischen Sprache, Literatur und
Kulturgeschichte, dem Gedenken an Martin Camaj (1925-1992) gewidmet (pp. 533–565).
Otto Harrassowitz Verlag.
Democratic Party of Iranian Kurdistan. (2020). About. PDKI.
https://pdki.org/english/about/
Democratic Union Party. (2020). Internal System. Partiya Yekîtiya Demokrat.
https://pydrojava.org/english/internal-system/
Diehl, P. F., & Goertz, G. (2012). The rivalry process: How rivalries are sustained and
terminated. In J. A. Vasquez (Ed.), What do we know about war? (2
nd
ed., pp. 83-110).
Rowman & Littlefield.

338

Dimitrieva, J., & Tiede, L. (2019). Global review of constitutional law: North
Macedonia. International Journal of Constitutional Law, 18(2), 605–613.
https://doi.org/10.1093/icon/moaa033
Dixon, J. S., & Sarkees, M. R. (2015). A guide to intra-state wars: An examination of civil,
regional, and intercommunal wars, 1816–2014. SAGE Publications.
https://doi.org/10.4135/9781452234205
Ducellier, A. (1999). Albania, Serbia and Bulgaria. In D. Abulafia (Ed.). The new Cambridge
medieval history, Volume 5, c.1198–c.1300 (pp. 779-795). Cambridge University Press.
Duijzings, G. (2002). Religion and the politics of 'Albanianism': Naim Frasheri's Bektashi
writings. In S. Schwanders-Sievers & B. J. Fischer (Eds.), Albanian identities: Myth and
history (pp. 60-69). Indiana University Press.
Dunlop, J. B. (2011). Russia confronts Chechnya. Cambridge University Press.
Eckel, M. (2019, August 28). Former Chechen commander gunned down In Berlin; Eyes turn to
Moscow (and Grozny). Radio Free Europe/Radio Liberty.
https://www.rferl.org/a/chechen-commander-khangoshvili-berlin-assassination-moscow-
(and-grozny)/30133813.html
Eliassi, B. (2016). Statelessness in a world of nation-states: The cases of Kurdish diasporas in
Sweden and the UK. Journal of Ethnic and Migration Studies, 42(9), 1403-1419.
https://doi.org/10.1080/1369183X.2016.1162091
Elsie, R. (2010). Historical dictionary of Albania (2
nd
ed.) Scarecrow Press.
Elsie, R. (2011). Historical dictionary of Kosova. Scarecrow Press.

339

Fearon, J. (2004). Separatist wars, partition, and world order. Security Studies, 13(4), 394–415.
https://doi.org/10.1080/09636410490945965
Fearon, J. D., & Laitin, D. D. (1996). Explaining interethnic cooperation. American Political
Science Review, 90(4), 715-735. https://doi.org/10.2307/2945838
Fearon, J. D., & Laitin, D. D. (2000). Violence and the social construction of ethnic identity.
International Organization, 54(4), 845-877. https://doi.org/10.1162/002081800551398
Fearon, J. D., & Laitin, D. D. (2003). Ethnicity, insurgency, and civil war. American Political
Science Review, 97(1), 75-90. https://doi.org/10.1017/S0003055403000534
Ferdowsi, A.-Q. (2006). Shahnameh: The Persian Book of Kings (D. Davis, Trans.). Viking.
(Original work published 1010)
Forage, P. C. (2001). The battle for Mount Pastrik: A preliminary study, The Journal of Slavic
Military Studies, 14(4), 57-80. https://doi.org/10.1080/13518040108430498
Fotheringham, A. S., & Wong, D. W. S. (1991). The modifiable areal unit problem in multivariate
statistical analysis. Environment and Planning A, 23(7), 1025–44.
https://doi.org/10.1068/a231025
Frashëri, K. (1964). The history of Albania: a brief survey. Tirana.
Free Life Party of Kurdistan (2020). Policy. https://pjak.eu/en/policy/
Frynas, J., & Wood, G. (2001). Oil and war in Angola. Review of African Political
Economy, 28(90), 587-606. https://doi.org/10.1080/03056240108704568
Fuller, L. (2007, November 1). Russia: Is North Caucasus resistance still serious threat? Radio
Free Europe/Radio Liberty. https://www.rferl.org/a/1079059.html

340

Fuller, L. (2010, August 4). Umarov’s U-turn. Radio Free Europe/Radio Liberty.
https://www.rferl.org/a/Umarovs_UTurn/2118487.html
Fuller, L. (2015, January 2). Six North Caucasus insurgency commanders transfer allegiance to
Islamic State. Radio Free Europe/Radio Liberty.
https://www.rferl.org/a/islamic-state-north-caucasus-insurgency-commanders-
allegiance/26773615.html
Galeotti, M. (2008). Blood brothers: The rise of Chechen organized crime. Jane’s Intelligence
Review.
Galeotti, M. (2018). The vory: Russia's super mafia. Yale University Press.
Gall, C., & de Waal, T. (1998). Chechnya: Calamity in the Caucasus. New York University Press.
Gammer, M. (1994). Muslim resistance to the tsar: Shamil and the conquest of Chechnia and
Daghestan. F. Cass.
Gammer, M. (2006). The lone wolf and the bear: Three centuries of Chechen defiance of Russian
rule. University of Pittsburgh Press. https://doi.org/10.3200/HIST.34.3.96-97
Geertz, C. (1963). Old societies and new states: The quest for modernity in Asia and Africa. Free
Press of Glencoe.
Gellner, E. (1983). Nations and nationalism. Blackwell.
Georgian Chronicle (R. Bedrosian, Trans.) (1991). Sources of the Armenian Tradition.
German, T. C. (2003). Russia’s Chechen war. Routledge.
Gerring, J. (2007). Case study research: Principles and practices. Cambridge University Press.

341

Gerring, J., & Cojocaru, L. (2016). Selecting cases for intensive analysis: A diversity of goals and
methods. Sociological Methods & Research, 45(3), 392–423.
https://doi.org/10.1177/0049124116631692
Getis, A., & Ord, J. K. (1992). The analysis of spatial association by use of distance statistics.
Geographical Analysis, 24(3), 189–206.
https://doi.org/10.1111/j.1538-4632.1992.tb00261.x
Gibler, D. M. (2012). The territorial peace: Borders, state development, and international conflict.
Cambridge University Press. https://doi.org/10.1017/CBO9781139060233
Gibler, D. M., & Tir, J. (2010). Settled borders and regime type: Democratic transitions as
consequences of peaceful territorial transfers. American Journal of Political Science, 54(4),
951–968. https://doi.org/10.1111/j.1540-5907.2010.00473.x
Gilpin, R. (1981). War and change in world politics. Cambridge University Press.
Glavin T. (2015). No friends but the mountains: The fate of the Kurds. World Affairs, 177(6), 57-
66.
Gleditsch, K. S. (2016). Transnational dimensions of civil war. Journal of Peace Research, 44(3),
293–309. https://doi.org/10.1177/0022343307076637
Gleditsch, K. S., Salehyan, I., & Schultz, K. (2008). Fighting at home, fighting abroad: How civil
wars lead to international disputes. Journal of Conflict Resolution, 52(4), 479–506.
https://doi.org/10.1177/0022002707313305

342

Gleditsch, K. S., & Weidmann, N. B. (2012). Richardson in the Information Age: Geographic
Information Systems and spatial data in international studies. Annual Review of Political
Science. 15(1), 461–481. https://doi.org/10.1146/annurev-polisci-031710-112604
Glenny, M. (2012). The Balkans: Nationalism, war, and the great powers, 1804-2011. Penguin
Books.
Global Administrative Areas (2018). GADM Database of Global Administrative Areas (v3.6)
[Data set]. https://gadm.org/data.html
Goedde, H. W., Paik Y. K., Lee, C. C., Benkmann, H. G., Kriese, L., Bogdanski, P., & Winkler,
M. (1987). Red cell and serum protein polymorphisms in three population groups of South
Korea. Gene Geography, 1(3), 177–188.
Goemans, H. E., & Schultz, K. A. (2017). The politics of territorial claims: A geospatial approach
applied to Africa. International Organization, 71(1), 31–64.
https://doi.org/10.1017/S0020818316000254
Goertz, G., & Diehl, P. F. (1997). Linking risky dyads: An evaluation of the relations between
enduring rivalries. In G. Schneider & P.A. Weitsman (Eds.), Enforcing cooperation: Risky
states and intergovernmental management of conflict (pp. 132-160). Macmillan.
Goodwin, G. (1997). The janissaries. Saqi Book Depot.
Gramacki, A. (2018). Nonparametric kernel density estimation and its computational aspects.
Springer International Publishing. https://doi.org/10.1007/978-3-319-71688-6
Grant, T. D. (2000). Current development: Afghanistan recognizes Chechnya. American
University International Law Review, 15(4), 869-894.

343

Green, N. (Ed.). (2017). Afghanistan’s Islam: From conversion to the Taliban. University of
California Press. https://doi.org/10.1525/9780520967373
Gruffydd-Jones, J. (2017). Dangerous days: The impact of nationalism on interstate
conflict. Security Studies, 26(4), 698–728.
https://doi.org/10.1080/09636412.2017.1336393
Gu, W. (1995). Conflicts of divided nations: The cases of China and Korea. Praeger.
Gunes, C., & Lowe, R. (2015). The impact of the Syrian War on Kurdish politics across the Middle
East. Chatham House.
https://syria.chathamhouse.org/assets/documents/20150723SyriaKurdsGunesLowe.pdf
Gunter, M. M. (1996). The KDP-PUK Conflict in Northern Iraq. The Middle East Journal, 50(2),
224-241.
Gunter, M. M. (2008) The Kurds ascending: The evolving solution to the Kurdish Problem in Iraq
and Turkey. Palgrave Macmillan.
Gunter, M. M. (2013). The Kurdish Spring. Third World Quarterly, 34(3), 441-457.
https://doi.org/10.1080/01436597.2013.785339
Gunter, M. M. (2014). Out of nowhere: The Kurds of Syria in peace and war. Hurst and Company.
Gunter, M. M. (2015). Iraq, Syria, ISIS and the Kurds: Geostrategic concerns for the U.S. and
Turkey. Middle East Policy, 22(1), 102–111. https://doi.org/10.1111/mepo.12116
Gurr, T. R. (1970). Why men rebel. Princeton University Press.
Gurr, T. R. (1993). Minorities at risk: A global view of ethnopolitical conflicts. United States
Institute of Peace Press.

344

Gustafsson, K., & Hagström, L. (2018). What is the point? Teaching graduate students how to
construct political science research puzzles. European Political Science, 17(4), 634–648.
https://doi.org/10.1057/s41304-017-0130-y
Güvenç, M. (2011). Constructing narratives of Kurdish nationalism in the urban space of
Diyarbakir, Turkey. Traditional Dwellings and Settlements Review, 23(1), 25–40.
Haboush, J. K., Haboush, W. J., Kim, J. M. Wang, S. Cho, H. & Chizhova-Kim K. (Eds.). (2016).
The Great East Asian War and the birth of the Korean Nation. Columbia University Press.
https://doi.org/10.7312/habo17228
Habyarimana, J., Humphreys, M., Posner, D. N., & Weinstein, J. M. (2009). Coethnicity: Diversity
and the dilemmas of collective action. Russell Sage Foundation.
Hackeloeer, A., Klasing, K., Krisp, J. M., & Meng, L. (2014). Georeferencing: A review of
methods and applications. Annals of GIS, 20(1), 61-69.
Haggard, S. & Kaufman, R. (2016). Dictators and democrats: Masses, elites, and regime change.
Princeton University Press,
Halberstam, D. (2007). The coldest winter: America and the Korean War. Hyperion.
Hall, D. (2001). Albania in Europe: Condemned to the periphery or beyond? Geopolitics, 6(1),
107–118. https://doi.org/10.1080/14650040108407709
Hama, H. H., Abdulla, F. H., & Jasim, D. (2018). One battle, two narratives? Rudaw’s framing
during the 2017 conflict over the disputed territories of the Kurdistan Region and Iraq.
Journal of International Communication, 24(2), 238-261.
https://doi.org/10.1080/13216597.2018.1474125

345

Hamburg, G., Sanders, T., & Tucker, E. (2004). Russian-Muslim confrontation in the Caucasus:
Alternative visions of the conflict between Imam Shamil and the Russians, 1830-1859.
Routledge.
Harris, G. S. (1977). Ethnic conflict and the Kurds. Annals of the American Academy of
Political and Social Science, 433(1), 112–124.
https://doi.org/10.1177%2F000271627743300111
Harris, P., Brunsdon, C., Lu, B., Nakaya, T., & Charlton, M. (2017). Introducing bootstrap
methods to investigate coefficient non-stationarity in spatial regression models. Spatial
Statistics 21(A), 241-261. https://doi.org/10.1016/j.spasta.2017.07.006
Harrison, S. S. (2002). Korean endgame: A strategy for reunification and U.S. disengagement.
Princeton University Press.
Haxhiaj, S. (2018, March 6). Kosovo’s political murders: Unpunished but not forgotten. Balkan
Transitional Justice.
https://balkaninsight.com/2018/03/06/kosovo-s-political-murders-unpunished-but-not-
forgotten-03-05-2018/
Hayes, J., & James, P. (2017). Systemism, analytic eclecticism, and the democratic peace. In A.
Freyberg-Inan, E. Harrison, & P. James (Eds.), Evaluating progress in International
Relations: How do you know? (pp. 109-124). Taylor & Francis.
Hedges, C. (1999). Kosovo’s next masters? Foreign Affairs, 78(3), 24–42.
Hennerblicher, F. (2012). The origin of Kurds. Advances in Anthropology, 2(2), 64-79.
https://doi.org/10.4236/aa.2012.22008

346

Henriksen, D. (2013). NATO's gamble: Combining diplomacy and airpower in the Kosovo
crisis, 1998-1999. Naval Institute Press.
Higgins, A. (1996, January 22). Profile: Dzhokhar Dudayev: Lone wolf of Grozny. The
Independent.
https://www.independent.co.uk/voices/profile-dzhokhar-dudayev-lone-wolf-of-grozny-
1569145.html
Hislope, R. (2004). Crime and honor in a weak state: Paramilitary forces and violence in
Macedonia. Problems of Post-Communism, 51(3), 18–26.
Horowitz, D. L. (1981). Patterns of ethnic separatism. Comparative Studies in Society and History,
23(2), 165-195.
Horowitz, D. L. (1985). Ethnic groups in conflict. University of California Press.
Hughes, J. (2001). Chechnya: The causes of a protracted post‐Soviet conflict. Civil Wars, 4(4), 11-
48. https://doi.org/10.1080/13698240108402486
Hughes, J. (2005). The peace process in Chechnya. In R. Sakwa (Ed.), Chechnya: From past to
future (pp. 265-288). Anthem Press.
Hughes, J. (2007). Chechnya: From nationalism to jihad. University of Pennsylvania Press.
https://doi.org/10.9783/9780812202311
Human Rights Watch. (2001, October 26). Under orders: War crimes in Kosovo.
https://www.hrw.org/legacy/reports/2001/kosovo/
Human Rights Watch. (2004). Background on the crisis in Iraq: Ansar al-Islam in Iraqi
Kurdistan. https://www.hrw.org/legacy/backgrounder/mena/ansarbk020503.htm

347

Human Rights Watch. (2020). World report 2020: Syria.
https://www.hrw.org/world-report/2020/country-chapters/syria
Humanitarian Data Exchange. (2020). Iraq - Roads network (HDX v1.44.8) [Data set]. United
Nations Office for the Coordination of Humanitarian Affairs.
https://data.humdata.org/dataset/iraq-roads
Husanović, J. (2000). “Post‐conflict” Kosovo: An anatomy lesson in the ethics/politics of human
rights. International Journal of Human Rights, 4(3-4), 263–280.
https://doi.org/10.1080/13642980008406903
Ibrahim, M. A. (2020, February 3). The rivaling philosophies of Barzani and Ocalan weigh over
Syria's Kurds. Syria Direct.
https://syriadirect.org/news/the-rivaling-philosophies-of-barzani-and-ocalan-weigh-over-
syrias-kurds-timeline/
Ilyon. (2008). Samguk Yusa: Legends and history of the Three Kingdoms of ancient Korea (T-H.
Ha, Trans.). Yonsei University Press. (Original work published 1512)
International Crisis Group. (1998, October 1). Albania Crisis Briefing.
https://reliefweb.int/report/albania/albania-crisis-briefing
International Crisis Group. (2018, February 20). Winning the post-ISIS battle for Iraq in Sinjar.
https://www.crisisgroup.org/middle-east-north-africa/gulf-and-arabian-
peninsula/iraq/183-winning-post-isis-battle-iraq-sinjar
International Institute for Strategic Studies. (1999). The Kosovo Liberation Army. Strategic
Comments, 5(4), 1-2. https://doi.org/10.1080/1356788990541

348

Iosifides, T., Lavrentiadou, M., Petracou, E., & Kontis, A. (2007). Forms of social capital and the
incorporation of Albanian immigrants in Greece. Journal of Ethnic and Migration Studies,
33(8), 1343-1361. https://doi.org/10.1080/13691830701614247
Isozaki, A. (2017). Understanding the North Korean regime. Wilson Center.
https://www.wilsoncenter.org/sites/default/files/media/documents/publication/ap_underst
andingthenorthkoreanregime.pdf
Jaimoukha, A. M. (2005). The Chechens: A handbook. RoutledgeCurzon.
James, P. (2019). What do we know about crisis, escalation and war? A visual assessment of the
International Crisis Behavior Project. Conflict Management and Peace Science, 36(1), 3–
19. https://doi.org/10.1177/0738894218793135
James, P., & Mohammadian, C. (2017). Ethnic conflict: Questions and answers. Ethnopolitics,
16(1), 89–97. https://doi.org/10.1080/17449057.2016.1235827
Jang, S., & Suh, J. (2017). Development and security in international aid to North Korea:
Commonalities and differences among the European Union, the United States and South
Korea. Pacific Review, 30(5), 729–749. https://doi.org/10.1080/09512748.2017.1294614
Jenne, E. K. (2015). Ethnic bargaining: The paradox of minority empowerment. Cornell University
Press.
Jenne, E. K., Saideman, S. M., & Lowe, W. (2007). Separatism as a bargaining posture: The role
of leverage in minority radicalization. Journal of Peace Research, 44(5), 539–558.
https://doi.org/10.1177/0022343307080853

349

Jin, H. J., Tyler-Smith, C., & Kim, W. (2009). The peopling of Korea revealed by analyses of
mitochondrial DNA and Y-chromosomal markers. PLOS One, 4(1), e4210.
https://doi.org/10.1371/journal.pone.0004210
Jinadu, L. A. (2004). Explaining and managing ethnic conflict in Africa: Towards a cultural theory
of democracy. African Journal of Political Science, 9(1), 1-26.
https://doi.org/10.4314/ajps.v9i1.27357
Joe, W. J. (1982). Traditional Korea: A cultural history. Chungʼang University Press.
Jonsson, M. (2014). The Kosovo conflict: From humanitarian intervention to state capture. In S.
Cornell & M. Jonsson (Eds.), Conflict, crime, and the state in Postcommunist Eurasia (pp.
177-198). University of Pennsylvania Press.
https://doi-org.libproxy1.usc.edu/10.9783/9780812208986
Jorga, N. (1908). Geschichte des Osmanischen Reiches. Gotha.
Judah, T. (2001, May 3). Greater Albania - A fading dream. Institute for War and Peace Reporting.
https://reliefweb.int/report/albania/greater-albania-fading-dream
Judah, T. (2002). Kosovo: War and revenge. Yale University Press.
Jung, H. J., Park, G., Ji, M., & Rich, T. S. (2021). Why help North Korea? Foreign aid to North
Korea and its determinants, 2002–2018. Pacific Focus, 36(2), 202–228.
https://doi.org/10.1111/pafo.12186
Jwaideh, W. (2006). The Kurdish national movement: Its origins and development. Syracuse
University Press.
Kaldor, M. (2012) New and Old Wars: Organized violence in a global era (3
rd
ed.). Polity.

350

Kaldor, M. (2013). In defence of New Wars. Stability: International Journal of Security and
Development, 2(1), 1-16. https://doi.org/10.5334/sta.at
Kalyvas, S. N. (2003). The ontology of ‘political violence’: Action and identity in civil wars.
Perspectives on Politics, 1(3), 475-94. https://doi.org/10.1017/S1537592703000355
Kalyvas, S. N. (2006). The logic of violence in civil war. Cambridge University Press.
Kalyvas, S. N. (2008). Ethnic defection in civil war. Comparative Political Studies, 41(8), 1043-
68. https://doi.org/10.1177/0010414008317949
Kalyvas, S. N., & Kocher, M. A. (2007). Ethnic cleavages and irregular war: Iraq and Vietnam.
Politics and Society, 35(2), 183–223. https://doi.org/10.1177/0032329207302403
Karcha, R. (1959). Soviet propaganda concerning the rehabilitated peoples of the Northern
Caucasus. Caucasian Review, 8, 3–16.
Kang, D. C. (2009). Between balancing and bandwagoning: South Korea's response to
China. Journal of East Asian Studies, 9(1), 11–28.
https://doi.org/10.1017/S1598240800002794
Kang, D. C. (2010). East Asia before the West: Five centuries of trade and tribute. Columbia
University Press.
Kaseda, Y. (2016). Japan’s policy regarding Korean rapprochement and unification. In T.-H.
Kwak & S.-H. Joo (Eds.), One Korea: Visions of Korean unification (pp. 186-205). Taylor
& Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315598895
Kataoka, T. (1974). Resistance and revolution in China: The Communists and the Second United
Front. University of California Press.

351

Kaufman, S.J. (2001). Modern hatreds: The symbolic politics of ethnic war. Cornell University
Press.
Kavkaz Center. (2017, July 31). Journalists published evidence of mass execution of Chechens by
order of Kadyrov.
https://www.kavkazcenter.com/russ/content/2017/07/31/115205/zhurnalisty-
opublikovali-dokazatelstva-massovoj-kazni-chechentsev-po-prikazu-kadyrova.shtml
Keen, D. (2008). Complex emergencies. Polity.
Keen, D. (2012). Greed and grievance in civil war. International Affairs, 88(4), 757–777.
https://doi.org/10.1111/j.1468-2346.2012.01100.x
Kerle, N., & de Leeuw, J. (2009). Reviving legacy population maps with object-oriented image
processing techniques. IEEE Transactions on Geoscience and Remote Sensing, 47(7),
2392-2402. https://doi.org/10.1109/TGRS.2008.2010853
Kim, H. (1983). The influence of Christianity on modern Korean political thought. Korean
Journal, 23(12), 4-16.
Kim, J. (2012). A history of Korea from “land of the morning calm” to states in conflict. Indiana
University Press.
Kim, J. (2020, September 25). Letters revealed: North and South Korea exchange well wishes over
COVID-19. NK News.
https://www.nknews.org/2020/09/letters-revealed-north-and-south-korea-exchange-well-
wishes-over-covid-19/

352

Kim, T. (2016). China and Korean unification strategies: A Korean perspective. In T.-H. Kwak
& S.-H. Joo (Eds.), One Korea: Visions of Korean unification (pp. 129-147). Taylor &
Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315598895
King, G., Keohane, R. O., & Verba, S. (1994). Designing social inquiry: Scientific inference in
qualitative research. Princeton University Press.
Knapp, M., Flach, A., Ayboga, E., Graeber, D., Abdullah, A., & Biehl, J. (2016). Revolution in
Rojava: Democratic autonomy and women’s liberation in Syrian Kurdistan. Pluto Press.
Koktsidis, P.-I., & ten Dam, C. (2008). A success story? Analysing Albanian ethno-nationalist
extremism in the Balkans. East European Quarterly, 42(2), 161–190.
Komala. (2020). Socio-political background of birth of Komala.
https://www.komalainternational.org/read-about
Koontz, K. (2019). Borders beyond borders: The many (many) Kurdish political parties of Syria.
Middle East Institute.
https://www.mei.edu/publications/borders-beyond-borders-many-many-kurdish-political-
parties-syria
Korean Central News Agency. (2000, March 22). Moves to legislate on "suffrage" in Japan
condemned. Korean Central Broadcasting Committee, North Korea.
https://web.archive.org/web/20141012070653/http://www.kcna.co.jp/item/2000/200003/
news03/22.htm
Koryashkin, P., & Marychev, A. (1999, November 29). Maskhadov calls for killing mufti Kadyrov,
enemy no 1. ITAR - TASS News Wire.

353

Kramer, M. (2005). Guerrilla warfare, counterinsurgency and terrorism in the North Caucasus:
The military dimension of the Russian-Chechen conflict. Europe-Asia Studies, 57(2), 209-
290. https://doi.org/10.1080/09668130500051833
Krasniqi, P. (2000, July 18). Albanians spurn K-for. Institute for War and Peace Reporting.
https://reliefweb.int/report/serbia/albanians-spurn-k
Kretsi, G. (2002). The secret past of the Greek-Albanian borderlands. Cham Muslim Albanians:
Perspectives on a conflict over historical accountability and current rights. Ethnologia
Balkanica, 6, 171-195.
Ku, Y. (2017a). North Korean economy. In Y. Ku, I. Lee, & J. Woo (Eds), Politics in North and
South Korea: Political development, economy, and foreign relations (pp. 130-150). Taylor
& Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315627014
Ku, Y. (2017b). South Korean economy. In Y. Ku, I. Lee, & J. Woo (Eds), Politics in North and
South Korea: Political development, economy, and foreign relations (pp. 71-90). Taylor
& Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315627014
Kurdistan Communities Union. (2018, March 23). KCK: Security maintained in Shengal, we
withdraw the guerrillas [Press release]. Ajansa Nûçeyan a Firatê.
https://anfenglishmobile.com/kurdistan/kck-security-maintained-in-shengal-we-
withdraw-the-guerrillas-25684
Kurdistan Democratic Party. (1996, March 20). Kurdistan Democratic Party leader comments on
situation in Kurdistan [Press release]. BBC Summary of World Broadcasts.

354

Kurdistan Democratic Party. (1996, September 12). Kurdistan will stay independent from Iraq,
Barzani says [Press release]. Agence France Presse.
Kurdistan Democratic Party. (1997, July 1). Kurdistan Democratic Party responds to rival PUK's
peace plan [Press release]. BBC Summary of World Broadcasts.
Kurdistan Democratic Party. (2017, March 4). Bela xwe nefroşin Kurdistanê Bextiyar Şaxî [Press
release]. Partiya Demokrat a Kurdistanê.
https://www.kdp.info/a/d.aspx?s=040000&l=16&a=98926
Kurdistan Democratic Party. (2017, March 20). Di bersiva PYDê de. Partiya Demokrat a
Kurdistanê [Press release]. https://www.kdp.info/a/d.aspx?s=040000&l=16&a=99392
Kurdistan Democratic Party. (2019, October 15). Barzanî pêşmergeyekî berjewendîxwaz e ji bo
netewe û welat [Press release]. Partiya Demokrat a Kurdistanê.
https://www.kdp.info/a/d.aspx?s=040000&l=16&a=117968
Kurdistan National Council. (2020). About the KNC. Kurdish National Council Geneva – Berlin.
https://knc-geneva.ezks.org/?page_id=49&lang=en
Kutschera, C. (1994). Mad dreams of independence: The Kurds of Turkey and the PKK. Middle
East Research and Information Project.
https://merip.org/1994/07/mad-dreams-of-independence/
Kwak, T-H. (2016). One Korea unification vision through neutralization. In T.-H. Kwak & S.-H.
Joo (Eds.), One Korea: Visions of Korean unification (pp. 10-30). Taylor & Francis.
https://doi-org.libproxy1.usc.edu/10.4324/9781315598895

355

Lake, D. A., & Rothchild, D. S. (1996). Containing fear: The origins and management of ethnic
conflict. International Security, 21(2), 41–75.
Lankov, A. (2002). From Stalin to Kim Il Sung: The formation of North Korea 1945–1960. Rutgers
University Press.
Lapidus, G. (1998). Contested sovereignty: The tragedy of Chechnya. International
Security, 23(1), 5-49. https://doi.org/10.2307/2539261
Lawless, J. (1987). Regression methods for Poisson process data. Journal of the American
Statistical Association, 82(399), 808–815.
Lee, E. D., Daniels, B. C., Myers, C. R., Krakauer, D. C., & Flack, J. C. (2020). Scaling theory of
armed-conflict avalanches. Physical Review. E, 102(4-1), 042312–042312.
https://doi.org/10.1103/PhysRevE.102.042312
Lee, I. (2017). Legacies of Japanese colonial rule and the Korean War. In Y. Ku, I. Lee, & J. Woo
(Eds), Politics in North and South Korea: Political development, economy, and foreign
relations (pp. 11-28). Taylor & Francis.
https://doi-org.libproxy1.usc.edu/10.4324/9781315627014
Lee, J., & Oba, M. (2019). Hitting an all-time low. Comparative Connections, 21(1), 105–138.
Lee, K.-B. (1984). A new history of Korea (E. W. Wagner & E. J. Schultz, Trans.). Harvard
University Press.
Lee, K.-M., & Ramsey, R. (2011). A history of the Korean language. Cambridge University Press.

356

Lew, Y. I. (2013). The making of the first Korean president: Syngman Rhee’s quest for
independence, 1875-1948. University of Hawaii Press.
https://doi.org/10.1515/9780824839147
Li, D., & Xia, Y. (2014). Jockeying for leadership: Mao and the Sino-Soviet split, October 1961 –
July 1964. Journal of Cold War Studies, 16(1), 24–60.
https://doi.org/10.1162/JCWS_a_00430
Licklider, R. (1995). The consequences of negotiated settlements in civil wars, 1945-1993.
American Political Science Review, 89(3), 681-690. https://doi.org/10.2307/2082982
Lieven, A. (1998). Chechnya: Tombstone of Russian power. Yale University Press.
Lijphart, A. (1969). Consociational democracy. World Politics, 21(2), 207–225.
https://doi.org/10.2307/2009820
Lijphart, A. (1977). Democracy in plural societies: A comparative exploration. Yale University
Press.
Lilja, J., & Hultman, L. (2011). Intraethnic dominance and control: Violence against coethnics in
the early Sri Lankan Civil War. Security Studies, 20(2), 171-197.
https://doi.org/10.1080/09636412.2011.572676
Lipa, D. [@DUALIPA] (2020, July 19). au•toch•tho•nous adjective (of an inhabitant of a place)
indigenous rather than descended from migrants or colonists . Twitter.
https://twitter.com/DUALIPA/status/1284928447912050688

357

Littell, J. (2006). The security organs of the Russian Federation. A brief history 1991-2005. Post-
Soviet Armies Newsletter.
Lloshi, X. (1999). Albanian. In U. Hinrichs & U. Büttner (Eds.), Handbuch der Südosteuropa-
Linguistik (pp.278-299). Otto Harrassowitz Verlag.
Lombardi, B. (2009). Balkan intrigue: German intelligence and Kosovo. International Journal of
Intelligence and Counterintelligence, 22(3), 470-490.
https://doi.org/10.1080/08850600902896936
Lyall, J. (2010). Are coethnics more effective counterinsurgents? Evidence from the Second
Chechen War. American Political Science Review, 104(1), 1-20.
https://doi.org/10.1017/S0003055409990323
Maitre, B. R. (2009). What sustains ‘Internal Wars’? The dynamics of violent conflict and state
weakness in Sudan. Third World Quarterly, 30(1), 53–68.
https://doi.org/10.1080/01436590802622318
Majko, P. (1999, March 25). Albania welcomes NATO strikes [Press release]. Radio Free
Europe/Radio Liberty. https://reliefweb.int/report/albania/albania-welcomes-nato-strikes
Majko, P. (1999, March 24). Albania concentrates troops along Kosova border and prepares for
new refugee influx [Press release]. Radio Free Europe/Radio Liberty.
https://reliefweb.int/report/albania/albania-concentrates-troops-along-kosova-border-and-
prepares-new-refugee-influx

358

Mamon, M. (2015, July 13). How a Chechen from Georgia became a feared leader of ISIS. The
Intercept.
https://theintercept.com/2015/07/13/chechen-georgia-became-feared-leader-isis/
Marshall, M. G. (2017). Major episodes of political violence 1946-2016. Center for Systemic
Peace.
Marshall, M. G. (2020). Current status of the world's major episodes of political violence. Monthly
reports to US Government's Political Instability Task Force.
Marshall, M. G., & Elzinga-Marshall, G. (2017). Global Report 2017: Conflict, governance, and
state fragility. Center for Systemic Peace. http://www.systemicpeace.org/globalreport.html
Martin, T. (1998). The origins of Soviet ethnic cleansing. Journal of Modern History, 70(4), 813-
861. https://doi.org/10.1086/235168
Marty, D. (2010). Committee on Legal Affairs and Human Rights: Inhuman treatment of people
and illicit trafficking in human organs in Kosovo. Council of Europe, Parliamentary
Assembly.
https://assembly.coe.int/committeedocs/2010/20101218_ajdoc462010provamended.pdf
Masters, B. (2001). Christians and Jews in the Ottoman Arab World: The Roots of Sectarianism.
Cambridge University Press.
Matloff, J. (2017). No friends but the mountains: Dispatches from the world's violent highlands.
Basic Books.

359

Maza, C. (2014, July 22). Georgia’s Pankisi Gorge, home to many of the Chechens fighting in
Syria. Balkanist.
https://balkanist.net/georgia-is-home-to-many-of-the-chechen-fighters-in-syria/
McDowall, D. (2004). A modern history of the Kurds (3
rd
ed.). I.B. Tauris.
McGarry, J., & O’Leary, B. (2006). Consociational theory, Northern Ireland’s conflict, and its
agreement. What consociationalists can learn from Northern Ireland. Government and
Opposition, 41(1), 43–63. https://doi.org/10.1111/j.1478-9302.2011.00235.x
McLauchlin, T., & Pearlman, W. (2012). Out-group conflict, in-group unity? Exploring the effect
of repression on intramovement cooperation. Journal of Conflict Resolution, 56(1), 41–
66. https://doi.org/10.1177/0022002711429707
Mece, M. H. (2017). Living in uncertainty: Resurgence of blood feud in Albanian post-socialist
society and its consequences on children and young adults. Contemporary Readings in Law
and Social Justice, 9(2), 28–38. https://doi.org/10.22381/CRLSJ9220172
Mello, P. A. (2010). In search of new wars: The debate about a transformation of war. European
Journal of International Relations, 16(2), 297 – 309.
https://doi.org/10.1177/1354066109350053
Mercer, J. (1995). Anarchy and identity. International Organization, 49(2), 229-252.
https://doi.org/10.1017/S0020818300028381
Merton, R. K. (1968). Social theory and social structure. Free Press.

360

Messner, J.J., Haken, N., Taft, P., Onyekwere, I., Blyth, H., Fiertz, C., Murphy, C., Quinn, A., &
Horowitz, M. (2018). Fragile states index. The Fund for Peace.
https://fragilestatesindex.org/
Milliyet. (2014, March 13). Kürt cephesinde sürpriz isim.
https://www.milliyet.com.tr/gundem/kurt-cephesinde-surpriz-isim-1925250
Ministry of Foreign Affairs. (2011). 2. Diplomacy for peace and stability on the Korean peninsula.
Ministry of Foreign Affairs, Republic of Korea [White paper].
https://www.mofa.go.kr/eng/brd/m_5684/down.do?brd_id=761&seq=310627&data_tp=
A&file_seq=2
Ministry of Foreign Affairs. (2019). II. Establishing peace on the Korean peninsula and
strengthening cooperation in the region. Ministry of Foreign Affairs, Republic of Korea
[White paper].
https://www.mofa.go.kr/viewer/skin/doc.html?fn=20210226121047492.pdf&rs=/viewer/r
esult/202110
Monshi, E. B. (1979). History of Shah Abbas the Great. Mazda.
Moran, P. A. P. (1950). Notes on continuous stochastic phenomena. Biometrika, 37(1-2), 17–23.
https://doi.org/10.1093/biomet/37.1-2.17
Morina, D. (2017, July 3). Kosovo ex-commander Lushtaku cleared of war crimes. Balkan
Transitional Justice.
https://balkaninsight.com/2017/07/03/senior-kosovo-ex-guerrillas-war-crimes-verdict-
07-03-2017/

361

Mueller, J. E. (1970). Presidential popularity from Truman to Johnson. American Political Science
Review, 64(1), 18–34. https://doi.org/10.2307/1955610
Mueller, J. E. (2004). The remnants of war. Cornell University Press.
Mukherjee, B. (2006). Why political power‐sharing agreements lead to enduring peaceful
resolution of some civil wars, but not others? International Studies Quarterly, 50(2), 479–
504. https://doi.org/10.1111/j.1468-2478.2006.00410.x
Mulaj, K. (2008). Resisting an oppressive regime: The case of Kosovo Liberation Army. Studies
in Conflict and Terrorism, 31(12), 1103-1119.
https://doi.org/10.1080/10576100802510660
Murad, N., & Krajeski, J. (2017). The last girl: My story of captivity, and my fight against the
Islamic State. Tim Duggan Books.
Naegele, J. (2001, March 21). Macedonia: Ethnic Albanian rebels flex muscles. Radio Free
Europe/Radio Liberty.
https://reliefweb.int/report/serbia/macedonia-ethnic-albanian-rebels-flex-muscles
Naegele, J. (2002, April 16). Macedonia: Intra-party clashes rattle ahead of fall elections. Radio
Free Europe/Radio Liberty.
https://reliefweb.int/report/former-yugoslav-republic-macedonia/macedonia-intra-party-
clashes-rattle-ahead-fall-elections

362

Naegele, J. (2002, April 17). Macedonia: Members of disbanded UCK look back at last year's
fighting. Radio Free Europe/Radio Liberty.
https://reliefweb.int/report/former-yugoslav-republic-macedonia/macedonia-members-
disbanded-uck-look-back-last-years
Naím, M. (2005). Illicit: How smugglers, traffickers, and copycats are hijacking the global
economy (1
st
ed.). Doubleday.
Nasidze, I, Ling, E. Y., Quinque D., Dupanloup, I., Cordaux, R., Rychkov, S., Naumova, O.,
Zhukova, O., Sarraf-Zadegan, N., Naderi, G. A., Asgary, S., Sardas, S., Farhud, D. D.,
Sarkisian, T., Asadov, C., Kerimov, A., Stoneking, M. (2004). Mitochondrial DNA
and Y-chromosome variation in the Caucasus. Annals of Human Genetics, 68(3), 205-
21. https://doi.org/10.1046/j.1529-8817.2004.00092.x
National Bureau of Statistics of China (2019). 25-17 Administrative division and population of
ethnic minority autonomous areas (2018). China Statistical Yearbook.
http://www.stats.gov.cn/tjsj/ndsj/2019/indexeh.htm
Nelson, S. M. (1993). The archaeology of Korea. Cambridge University Press.
NextGIS. (2021). Primary routes of the Chechen Republic [Data set]. https://nextgis.com/
North Atlantic Treaty Organization. (1999, June 9). Military technical agreement between the
international security force ("KFOR") and the governments of the Federal Republic
of Yugoslavia and the Republic of Serbia [Press release].
https://www.nato.int/kosovo/docu/a990609a.htm

363

North Atlantic Treaty Organization. (2001, March 22). KFOR's got a tight grip on the border
[Press release]. https://reliefweb.int/report/serbia/kfors-got-tight-grip-border
North Korea. (2012). Juche idea: Answers to hundred questions. Foreign Languages Publishing
House on behalf of DPR Korea.
O, T. (2016). The collapse of North Korea: Challenges, planning and geopolitics of unification.
Palgrave Macmillan.
Obradović, J. (2013). Ethnic conflict and war crimes in the Balkans: The narratives of denial in
post-conflict Serbia. I.B. Tauris.
Öcalan, A. (2017). Democratic confederalism: Kurdistan, woman’s revolution and democratic
confederalism. Pluto Press.
O’Leary, B. (2016). Power-sharing and partition amid Israel–Palestine. Ethnopolitics, 15(4), 345-
365. https://doi.org/10.1080/17449057.2016.1210353
Olson, M. (2002). The logic of collective action; public goods and the theory of groups (20
th
ed.).
Harvard University Press.
Olson, R. W. (1989). The emergence of Kurdish nationalism and the Sheikh Said Rebellion, 1880-
1925. University of Texas Press.
Openshaw, S. (1984). The modifiable areal unit problem. Geo Books.
OpenStreetMap (2020). https://www.openstreetmap.org/#map=5/38.007/-95.844
O’Sullivan, D., & Perry, G. (2013). Spatial simulation: Exploring pattern and process. John Wiley
& Sons Inc.

364

O'Sullivan, D., & Unwin, D. (2010). Geographic information analysis (2
nd
ed.). John Wiley &
Sons.
Özoğlu, H. (2004). Kurdish notables and the Ottoman State: Evolving identities, competing
loyalties, and shifting boundaries. SUNY Press.
Paine, S. C. M. (2002). The Sino-Japanese War of 1894–1895: Perceptions, power, and primacy.
Cambridge University Press.
Park, H. Y. (2011). Shared national memory as intangible heritage: Re-imagining two Koreas as
one nation. Annals of Tourism Research, 38(2), 520–539.
https://doi.org/10.1016/j.annals.2010.11.013
Patriotic Union of Kurdistan. (1996, May 26). Patriotic Union of Kurdistan offers to discuss peace
plan [Press release]. BBC Summary of World Broadcasts.
Patriotic Union of Kurdistan. (1996, October 28). PUK spokesman says no SAIRI, Iranian forces
present in northern Iraq [Press release]. BBC Summary of World Broadcasts.
Pearlman, W., & Cunningham, K. G. (2011). Nonstate actors, fragmentation, and conflict
processes. Journal of Conflict Resolution, 56(1), 3–15.
https://doi.org/10.1177/0022002711429669
Perritt, H. H. (2008). Kosovo Liberation Army: The inside story of an insurgency. University of
Illinois Press.
Pettersson, T., & Magnus, Ö. (2020). Organized violence, 1989-2019. Journal of Peace Research,
57(4), 727–742. https://doi.org/10.1177/0022343320934986

365

Pfonner, M. R., & James, P. (2020). The Visual International Relations Project. International
Studies Review, 22(2), 192–213. https://doi.org/10.1093/isr/viaa014
Pike, J. (2013). Albanian civil war. Global Security.
https://www.globalsecurity.org/military/world/war/albania.htm
Polian, P. M. (2004). Against their will: The history and geography of forced migrations in the
USSR. Central European University Press.
Polybius. (2010). Polybius’ Histories (B. C. McGing, Trans.). Oxford University Press.
Posen, B. R. (1993). The security dilemma and ethnic conflict. Survival, 35(1), 27-47.
https://doi.org/10.1080/00396339308442672
Ptolemy. (2000). Ptolemy’s Geography: An annotated translation of the theoretical chapters (J.
L. Berggren & A. Jones, Trans.). Princeton University Press.
Quackenbush, S. L. (2016). Centers of gravity and war outcomes. Conflict Management and Peace
Science, 33(4), 361-380. https://doi.org/10.1177/0738894215570430
Radio Free Europe/Radio Liberty. (2002, April 5). A shootout disrupts Macedonia's calm. RFE/RL
Balkan Report Volume 6, Number 14.
https://reliefweb.int/report/former-yugoslav-republic-macedonia/shootout-disrupts-
macedonias-calm
Radio Free Europe/Radio Liberty. (2007, October 22). Is Moscow seeking to subvert Chechen
resistance? Newsline – October 22, 2007. https://www.rferl.org/a/1143977.html

366

Radio Free Europe/Radio Liberty. (2016, July 15). Pentagon admits 'Omar the Chechen' died this
week, not earlier.
https://www.rferl.org/a/pentagon-admits-omar-the-chechnyan-died-this-week-not-in-
march-air-strike-mosul-iraq/27859533.html
Raleigh, C., Linke, A., Hegre, H., & Karlsen, J. (2010). Introducing ACLED: An Armed Conflict
Location and Event Dataset. Journal of Peace Research, 47(5), 651-660.
https://doi.org/10.1177/0022343310378914
Renner, I., & Warton, D. (2013). Equivalence of MAXENT and Poisson point process models for
species distribution modeling in ecology. Biometrics, 69(1), 274–281.
https://doi.org/10.1111/j.1541-0420.2012.01824.x
Reuveny, R., & Thompson, W. R. (2010). Coping with terrorism: Origins, escalation,
counterstrategies, and responses. SUNY Press.
Rhee, S. (2015). The diary of Syngman Rhee 1904-34 & 1944 = Yi Sŭng-Man Yŏngmun Ilgi
Yŏnginbon 1904-34 & 1944. Taehan Min’guk Yŏksa Pangmulgwan.
Riphenburg, C. J. (2005). Ethnicity and civil society in contemporary Afghanistan. Middle East
Journal, 59(3), 31-51. https://doi.org/10.3751/59.1.12
Robertson, C., & Feick, R. (2018). Inference and analysis across spatial supports in the big data
era: Uncertain point observations and geographic contexts. Transactions in GIS, 22(2),
455–476. https://doi.org/10.1111/tgis.12321

367

Rogg, I., & Rimscha, H. (2007). The Kurds as parties to and victims of conflicts in Iraq.
International Review of the Red Cross, 89(868), 823–842.
https://doi.org/10.1017/S1816383108000143
Romano, D. (2006). The Kurdish nationalist movement: Opportunity, mobilization and identity.
Cambridge University Press.
Romano, D. (2020, August 24). Intensifying Turkish war against Kurds marks Treaty of Sevres
centenary. Syrian Observatory for Human Rights. https://www.syriahr.com/en/181005/
Ron, J. (2003). Frontiers and ghettos: State violence in Serbia and Israel. University of California
Press. https://doi.org/10.1525/9780520936904
Rossi, M. (2014). Ending the impasse in Kosovo: Partition, decentralization, or consociationalism?
Nationalities Papers, 42(5), 867-889. https://doi.org/10.1080/00905992.2014.937683
Rudaw. (2015, August 3). Barzanî: Kurdên êzîdî nasnameya netewî ya kurdî parastine.
https://www.rudaw.net/kurmanci/kurdistan/0308201516
Rudaw. (2017, March 3). Di navbera Pêşmergeyên Rojava û YBŞê de şer derket!
https://www.rudaw.net/kurmanci/kurdistan/03032017
Rudaw. (2017, March 5). Pasdarên Îranê li Şingalê ligel YBŞê civiyane.
https://www.rudaw.net/kurmanci/kurdistan/050320177
Russell, J. (2011). Chechen elites: Control, cooption or substitution? Europe-Asia Studies, 63(6),
1073–1087. https://doi.org/10.1080/09668136.2011.585758

368

Saideman, S. M. (2002). Overlooking the obvious: Bringing international politics back into ethnic
conflict management. International Studies Review, 4(3), 63-86.
https://doi.org/10.1111/1521-9488.00265
Saideman, S. M. (2003). Explaining the international relations of secessionist conflicts:
Vulnerability versus ethnic ties. International Organization, 51(4), 721–753.
https://doi.org/10.1162/002081897550500
Saideman, S. M., & Ayres, R. W. (2008). For kin or country: Xenophobia, nationalism, and war.
Columbia University Press.
Sakwa, R. (2005). Introduction: Why Chechnya? In R. Sakwa (Ed.), Chechnya: From past to
future (pp. 1-20). Anthem Press.
Salehyan, I., & Gleditsch, K. S. (2006). Refugees and the spread of civil war. International
Organization, 60(2), 335-366. https://doi.org/10.1017/S0020818306060103
Sambanis, N. (2001). Do ethnic and nonethnic civil wars have the same causes? A theoretical and
empirical inquiry (Part 1). Journal of Conflict Resolution, 45(3), 259–282.
Sambanis, N., & Schulhofer-Wohl, J. (2009). What’s in a line? Is partition a solution to civil war?
International Security, 34(2), 82–118. https://doi.org/10.1162/isec.2009.34.2.82
Sambanis, N., Skaperdas, S., & Wohlforth, W. (2015). Nation-building through war. American
Political Science Review, 109(2), 279-296. https://doi.org/10.1017/S0003055415000088
Sarkees, M. R., & Wayman, F. W. (2010). Resort to war: A data guide to inter-state, extra-state,
intra-state, and non-state wars, 1816-2007. CQ Press.

369

Sayari, S., & Hoffman, B. (1991). Urbanization & insurgency: The Turkish case, 1976-1980. Rand
Corporation.
Schultz, K. A. (2017). Mapping interstate territorial conflict: A new data set and applications.
Journal of Conflict Resolution, 61(7), 1565–1590.
https://doi.org/10.1177/0022002715620470
Seth, M. J. (2016). A concise history of Korea: From antiquity to the present. Rowman &
Littlefield.
Seymour, L., Bakke, K., & Cunningham, K. (2016). E pluribus unum, ex uno plures: Competition,
violence and fragmentation in ethnopolitical movements. Journal of Peace
Research, 53(1), 3–18. https://doi.org/10.1177/0022343315605571
Sil, R., & Katzenstein, P., J. (2010). Analytic eclecticism in the study of world politics:
Reconfiguring problems and mechanisms across research traditions. Perspectives on
Politics, 8(2), 411–431. https://doi.org/10.1017/S1537592710001179
Sima, Q. (1993). Records of the Grand Historian: Han Dynasty II (B. Watson, Trans.). Columbia
University Press. (Original work published c. 86 BC)
Simmel, G. (1955). Conflict and the web of group affiliations (K. W. Wolff & R. Bendix, Trans.).
Free Press.
Simmons, B. A. (2005). Rules over real estate: Trade, territorial conflict, and international borders
as institution. Journal of Conflict Resolution, 46(6), 823-848.
https://doi.org/10.1177/0022002705281349

370

Singer, J. D. (1960). International conflict: Three levels of analysis. World Politics, 12(3), 453–
460. https://doi.org/10.2307/2009401
Singer, J. D. (1961). The level-of-analysis problem in international relations. World Politics, 14(1),
77–92. https://doi.org/10.2307/2009557
Sinjar Resistance Unit (2017, March 2). YBŞ/YJŞ Command: KDP should withdraw its forces
immediately [Press release]. Hawar News Agency.
https://web.archive.org/web/20170307105736/http://en.hawarnews.com/ybsyjs-
command-kdp-should-withdraw-its-forces-immediately
Sinjar Resistance Units. (2017, March 4). YBŞ/YJŞ vows to respond to every attack on Shengal
soil [Press release]. Hawar News Agency.
https://web.archive.org/web/20170305113542/http://en.hawarnews.com/ybsyjs-vows-to-
respond-to-every-attack-on-shengal-soil/
Siroky, D. S., & Cuffe, J. (2015). Lost autonomy, nationalism and separatism. Comparative
Political Studies, 48(1), 3-34. https://doi.org/10.1177/0010414013516927
Skendi, S. (2015). The Albanian national awakening, 1878-1912. Princeton University Press.
https://doi.org/10.1515/9781400847761
Skjelderup, M., Ainashe, M., & “Qare,” A. M. A. (2020). Militant Islamism and local clan
dynamics in Somalia: The expansion of the Islamic Courts Union in Lower Jubba
province. Journal of Eastern African Studies, 14(3), 553-571.
https://doi.org/10.1080/17531055.2020.1789929

371

Smith, H. (2014). Russia as a great power: Status inconsistency and the two Chechen
wars. Communist and Post-Communist Studies, 47(3), 355-363.
https://doi.org/10.1016/j.postcomstud.2014.09.005
Soh, S. C. (2009). The comfort women: Sexual violence and postcolonial memory in Korea and
Japan. University of Chicago Press.
Song, N. (2018). Making borders in modern East Asia: The Tumen River demarcation, 1881–
1919. Cambridge University Press.
Souleimanov, E. A., & Aliyev, H. (2015). Blood revenge and violent mobilization: Evidence from
the Chechen wars. International Security, 40(2), 158-180.
https://doi.org/10.1162/ISEC_a_00219
Spät, E. (2018). Yezidi identity politics and political ambitions in the wake of the ISIS attack.
Journal of Balkan and Near Eastern Studies, 20(5), 420-438.
https://doi.org/10.1080/19448953.2018.1406689
Staniland, P. (2012). Between a rock and a hard place: Insurgent fratricide, ethnic defection, and
the rise of pro-state paramilitaries. Journal of Conflict Resolution, 56(1), 16-40.
https://doi.org/10.1177/0022002711429681
Starr, H. (2005). Territory, proximity, and spatiality: The geography of international
conflict. International Studies Review, 7(3), 387-406.
https://doi.org/10.1111/j.1551-2916.2005.00506.x
Starr, H. (2013). On geopolitics: Space, place, and International Relations. Paradigm.

372

State News Agency. (2009, December 19). Traitor of the nation award. Chechen-press.
https://thechechenpress.com/developments/7215-premiya-predatel-natsii.html
State News Agency. (2015, November 23). The appeal of the chairman of the cabinet of ministers
of the Chechen Republic of Ichkeria to the G20 leaders. Chechen-press.
https://thechechenpress.com/developments/11130-the-appeal-of-the-chairman-of-the-
cabinet-of-ministers-of-the-chechen-republic-of-ichkeria-to-the-g20-leaders.html
Steele, J. (2006, July 10). Shamil Basayev: Chechen politician seeking independence through
terrorism. The Guardian.
https://www.theguardian.com/news/2006/jul/11/guardianobituaries.chechnya
Stein, G. H. (1984). The Waffen SS: Hitler's elite guard at war, 1939–45. Cornell University Press.
Stewart, F. (2008). Horizontal inequalities and conflict: Understanding group violence in
multiethnic societies. Palgrave Macmillan.
Stojarova, V. (2010). Nationalist parties and the party systems of the Western Balkans. In
V. Stojarova & P. Emerson (Eds.). Party politics in the Western Balkans (pp. 42-58).
Taylor & Francis. https://doi-org.libproxy1.usc.edu/10.4324/9780203866221
Stueck, W. (1995). The Korean War: An international history. Princeton University Press.
https://doi.org/10.1515/9781400821785
Stueck, W. (2002). Rethinking the Korean War: A new diplomatic and strategic history. Princeton
University Press.

373

Stueck, W., & Yi, B. (2010). An alliance forged in blood: The American occupation of Korea, the
Korean War, and the US–South Korean alliance. Journal of Strategic Studies, 33(2), 177-
209. https://doi.org/10.1080/01402391003590200
Sugar, P. F. (1977). Southeastern Europe under Ottoman rule, 1354–1804. University of
Washington Press.
Suhrke, A. (1975). Irredentism contained: The Thai-Muslim case. Comparative Politics, 7(2),
187-203. https://doi.org/10.2307/421548
Sundberg, R., & Melander, E. (2013). Introducing the UCDP Georeferenced Event Dataset.
Journal of Peace Research, 50(4), 91-98. https://doi.org/10.1177/0022343313484347
Syrian Observatory for Human Rights. (2019, May 5). The pro-Turkey factions of “Olive Branch
and Euphrates Shield” withdraw from the areas they have advanced into in the
northern countryside at the expense of the Kurdish forces after heavy shelling and
landmine explosions. https://www.syriahr.com/en/126695/
Syrian Observatory for Human Rights. (2020, October 9). “Peace Spring” a year on: A new map
of influence…growing humanitarian disaster and continuous Turkish violations.
https://www.syriahr.com/en/188093/
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin &
S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33-37). Brooks/Cole.
Tamm, H. (2016). Rebel leaders, internal rivals, and external resources: How state sponsors affect
insurgent cohesion. International Studies Quarterly, 60(4), 599–610.
https://doi.org/10.1093/isq/sqw033

374

Tarrow, S. (1994). Power in movement: Social movements, collective action and politics.
Cambridge University Press.
Taydas, Z., Enia, J., & James, P. (2011). Why do civil wars occur? Another look at the theoretical
dichotomy of opportunity versus grievance. Review of International Studies, 37(5), 2627-
2650. https://doi.org/10.1017/S026021051100012X
Temby, O. (2013). What are levels of analysis and what do they contribute to international
relations theory? Cambridge Review of International Affairs, 28(4), 721–742.
https://doi.org/10.1080/09557571.2013.831032
ten Dam, C. (2010). How to feud and rebel: Violence-values among the Chechens and
Albanians. Iran & the Caucasus, 14(2), 331–366.
https://doi.org/10.1163/157338410X12743419190340
Thies, C. G. (2004). State building, interstate and intrastate rivalry: A study of post-colonial
developing country extractive efforts, 1975-2000. International Studies Quarterly, 48(1),
53-72. https://doi.org/10.1111/j.0020-8833.2004.00291.x
Tian, J. Q. (2006). Government, business, and the politics of interdependence and conflict across
the Taiwan Strait (1
st
ed.). Palgrave MacMillan.
Tilly, C. (1985). War making and state making as organized crime. In P. Evans, D. Rueschemeyer,
& T. Skocpol (Eds.), Bringing the state back in (pp. 169-191). Cambridge University Press.
Tobler, W. R. (1979). Smooth pycnophylactic interpolation for geographical regions. Journal of
the American Statistical Association, 74(365), 519-530.
https://doi.org/10.1080/01621459.1979.10481647

375

Totten, G., & Jo, Y. (1991). South Korea and the politics of Korean unification. In Q. Zhao &
R. Sutter (Eds.), Politics of divided nations: China, Korea, Germany and Vietnam -
Unification, conflict resolution and political development (pp. 61-88). University of
Maryland Press.
Toumanoff, C. (1963). Studies in Christian Caucasian history. Georgetown University Press.
Tsai, P.-J. (2012). Application of Moran’s test with an empirical Bayesian rate to leading health
care problems in Taiwan in a 7-year period (2002–2008). Global Journal of Health
Science, 4(5), 63-77. https://doi.org/10.5539/gjhs.v4n5p63
Tufte, E. (2006). Beautiful evidence. Graphics Press LLC.
Valeriano, B. (2012). Becoming rivals: The process of rivalry development. In J. A. Vasquez (Ed.)
What do we know about war? (2
nd
ed., pp. 63-82). Rowman & Littlefield.
van Bruinessen, M. (1992). Agha, shaikh and state. Zed Books.
van de Mieroop, M. (2015). A history of the ancient Near East, ca. 3000-323 BC. Wiley-
Blackwell.
Vasquez, J. A. (1995). Why do neighbors fight? Proximity, interaction, or territoriality. Journal of
Peace Research, 32(3), 277-293. https://doi.org/10.1177/0022343395032003003
Vatchagaev, M. (2014). The politicization of Sufism in Chechnya. Caucasus Survey, 1(2), 25–35.
Vickers, M. (1998). Between Serb and Albanian: A history of Kosovo. Columbia University Press.
Vinci, A. (2006). Greed-grievance reconsidered: The role of power and survival in the motivation
of armed groups. Civil Wars, 8(1), 25–45. https://doi.org/10.1080/13698240600886032

376

Vogt, M., Bormann, N.-C., Rüegger, S., Cederman, L.-E., Hunziker, P., & Girardin, L. (2015).
Integrating data on ethnicity, geography, and conflict: The Ethnic Power Relations data set
family. Journal of Conflict Resolution, 59(7), 1327–42.
https://doi.org/10.1177/0022002715591215
von Clausewitz, K. (1917). Vom Kriege. Insel-Verlag.
Waltz, K. N. (1959). Man, the state, and war: a theoretical analysis. Columbia University Press.
Waltz, K. N. (1979). Theory of international politics. McGraw-Hill.
Walzer, M. (2000). Just and unjust wars (3
rd
ed.). Basic Books.
Warren, T. C., & Troy, K. K. (2015). Explaining violent intraethnic conflict: Group fragmentation
in the shadow of state power. Journal of Conflict Resolution, 59(3), 484-509.
https://doi.org/10.1177/0022002713515400
Warton, D., & Shepherd, L. (2010). Poisson point process models solve the “pseudo-absence
problem” for presence-only data in ecology. Annals of Applied Statistics, 4(3), 1383–
1402. https://doi.org/10.1214/10-AOAS331
Weber, M. (2015). Politics as a vocation. Weber’s Rationalism and Modern Society (T. Waters &
D. Waters, Eds.). Macmillan. (Original work published 1919)
Weller, N., & Barnes, J. (2014). Finding pathways: Mixed-method research for studying causal
mechanisms. Cambridge University Press.
Wheeler, N. J. (2002). Saving strangers: Humanitarian intervention in international society.
Oxford University Press.

377

White, Jr., M. S., & Griffin, P. (1985). Piecewise linear rubber-sheet map transformation. The
American Cartographer, 12(2), 123-131. https://doi.org/10.1559/152304085783915135
Whitmeyer, K. (2015). Religious outbidding in Chechnya. Caucasus Survey, 3(1), 25-40.
Williams, B. G. (2011). Afghanistan declassified: A guide to America’s longest war. University
of Pennsylvania Press. https://doi.org/10.9783/9780812206159
Williams, D., & Guttschuss, E. (2012). Spiraling violence: Boko Haram attacks and security force
abuses in Nigeria. Human Rights Watch.
https://www.hrw.org/report/2012/10/11/spiraling-violence/boko-haram-attacks-and-
security-force-abuses-nigeria
Wilson, B. (2007). The evolution of Iranian warfighting during the Iran–Iraq War: When
dismounted light infantry made the difference. Infantry. U.S. Army: Foreign Military
Studies Office.
Wixman, R. (1982). Ethnic nationalism in the Caucasus. Nationalities Papers, 10(2), 137-156.
https://doi.org/10.1080/00905998208407937
Wohlstetter, A. (1968). Illusions of distance. Foreign Affairs, 46(2), 242–255.
https://doi.org/10.2307/20039298
Woo, J. (2017). South Korean democratization. In Y. Ku, I. Lee, & J. Woo (Eds), Politics in North
and South Korea: Political development, economy, and foreign relations (pp. 29-48).
Taylor & Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315627014

378

Wucherpfennig, J., Weidmann, N. B., Girardin, L., Cederman, L.-E., & Wimmer, A. (2011).
Politically relevant ethnic groups across space and time: Introducing the GeoEPR dataset.
Conflict Management and Peace Science, 28(5), 423–37.
https://doi.org/10.1177/0738894210393217
Yao, X. (2000). An introduction to Confucianism. Cambridge University Press.
https://doi.org/10.1017/CBO9780511800887
Yavuz, M. & Gunter, M. (2001). The Kurdish nation. Current History, 100(642), 33–39.
https://doi.org/10.1525/curh.2001.100.642.33
Yazdi, S. A. (1989). Zafarnama, In A Century of Princes: Sources on Timurid History and Art (W.
M. Thackston, Ed.). Aga Khan Program for Islamic Architecture. (Original work published
1425)
Yi, K. (1984). A new history of Korea. Published for the Harvard-Yenching Institute by Harvard
University Press.
Yi, S., & Lee, H. E. (2016). The making of international law in Korea: From colony to Asian
power. Brill.
Yildiz, K. (2015). The Future of Kurdistan: The Iraqi dilemma. Pluto Press.
https://doi.org/10.2307/j.ctt183p57j
Young, B. R. (2021). Guns, guerillas, and the Great Leader: North Korea and the Third World.
Stanford University Press. https://doi.org/10.1515/9781503627642

379

Yu, W., Ai, T., & Shao, S. (2015). The analysis and delimitation of Central Business District using
network kernel density estimation. Journal of Transport Geography, 45, 32-47.
https://doi.org/10.1016/j.jtrangeo.2015.04.008
Yun, P. (2016). Guest editor’s introduction: Manchuria and Korea in East Asian history.
International Journal of Korean History, 21(1), 1-9.
https://doi.org/10.22372/ijkh.2016.21.1.1
Zedalis, R. J. (2012). Oil and gas in the disputed Kurdish territories: Jurisprudence, regional
minorities, and natural resources in a federal system. Routledge.
https://doi.org/10.4324/9780203116746
Zhao, Q., & Sutter, R. (1991). Introduction: Unification, conflict resolution, and political
development. In Q. Zhao & R. Sutter (Eds.), Politics of divided nations: China, Korea,
Germany and Vietnam - Unification, conflict resolution and political development
(pp. 1-6). University of Maryland Press.
Zhebin, A. (2016). The future vision of a unified Korean peninsula: A Russian perspective. In T.
Kwak & S. Joo (Eds.), One Korea: Visions of Korean unification (pp. 148-165). Taylor
& Francis. https://doi-org.libproxy1.usc.edu/10.4324/9781315598895
Zucchino, D. (1995, February 9). Chechen opposition says Dudayev a worse threat than
Russians. Houston Chronicle (Pre-1997 Fulltext).

380

Appendix A: Reference list for statistical software packages
The following packages were used in the computation of the statistical analyses using R
statistical software, version 3.6.3.
Allaire, J. J., Xie, Y., McPherson, J., Luraschi, J., Ushey, K., Atkins, A., Wickham, H., Cheng, J.,
Chang, W., & Iannone, R. (2020). rmarkdown: Dynamic documents for R. R package
version 2.1. https://CRAN.R-project.org/package=rmarkdown
Baddely, A., Turner, R., Rubak, E. (2020). spatstat: Spatial point pattern analysis, model-fitting,
simulation, tests. R package version 1.64-1. https://CRAN.R-project.org/package=spatstat
Bivand, R. (2019). classInt: Choose univariate class intervals. R package version 0.4-2.
https://CRAN.R-project.org/package=classInt
Bivand, R. (2020). maptools: Tools for handling spatial objects. R package 1.0-1.
https://CRAN.R-project.org/package=maptools
Bivand, R. (2020). spdep: Spatial dependence: Weighting schemes, statistics. R package version
1.1-5. https://CRAN.R-project.org/package=spdep
Bivand, R., Keitt, T., & Rowlingson, B. (2020). rgdal: Bindings for the 'Geospatial' Data
Abstraction Library. R package version 1.5-16.
https://CRAN.R-project.org/package=rgdal
Bivand, R., & Piras, G. (2019). spatialreg: Spatial regression analysis. R package version 1.1-5.
https://CRAN.R-project.org/package=spatialreg
Bivand, R., & Rundel, C. (2019). rgeos: Interface to geometry engine - open source ('GEOS'). R
package version 0.5.2. https://CRAN.R-project.org/package=rgeos

381

Brundson, C., & Hongyan, C. (2014). GISTools: Some further GIS capabilities for R. R package
version 0.7-4. https://CRAN.R-project.org/package=GISTools
Cheng, J., Karambelkar, B., & Xie, Y. (2019). leaflet: Create interactive web maps with the
JavaScript 'Leaflet' library. R package version 2.0.3.
https://CRAN.R-project.org/package=leaflet
Hijmans, R. J. (2018). rspatial: rspatial.org. R package version 1.0-0.
https://CRAN.R-project.org/package=rspatial
Hijmans, R. J. (2020). raster: Geographic data analysis and modeling. R package version 3.3-13.
https://CRAN.R-project.org/package=raster
Hlavac, M. (2018). stargazer: Well-formatted regression and summary statistics tables. R package
version 5.2.2. https://CRAN.R-project.org/package=stargazer
McIlroy, D. (2020). mapproj: Map projections. R package version 1.2.7.
https://CRAN.R-project.org/package=mapproj
Neuwirth, E. (2014). RColorBrewer: ColorBrewer Palettes. R package version 1.1-2.
https://CRAN.R-project.org/package=RColorBrewer
Pebesma, E. (2020). sf: Simple features for R. R package version 0.9.5.
https://CRAN.R-project.org/package=sf
Pebesma, E., & Bivand, R. (2020). sp: Classes and methods for spatial data. R package version
1.4-2. https://CRAN.R-project.org/package=sp

382

Peterson, B., & Carl, P. (2020). PerformanceAnalytics: Econometric tools for performance and
risk analysis. R package version 2.0.4.
https://CRAN.R-project.org/package=PerformanceAnalytics
R Core Team and contributors worldwide (2020). The R base package version 3.6.3.
https://CRAN.R-project.org/package=base
Revelle, W. (2019). psych: Procedures for psychological, psychometric, and personality research.
https://CRAN.R-project.org/package=psych
Wickham, H. (2018). reshape: Flexibly Reshape Data. R package 0.8.8.
https://CRAN.R-project.org/package=reshape
Wickham, H., Chang, W., Henry, L., Pedersen, T. L., Takahashi, K., Wilke, C., Woo, K., Yutani,
H., & Dunnington, D. (2020). ggplot2: Create elegant data visualisations using the
grammar of graphics. R package version 3.3.2.
https://CRAN.R-project.org/package=ggplot2
Wickham, H., Jackson, S., & Korpella, M. (2019). ggmap: spatial visualization with ggplot2. R
package version 3.0.0. https://CRAN.R-project.org/package=ggmap

383

Appendix B: Historical demographic maps selected for map processing
This work required three historical ethnographic maps showing the distribution of Kurds,
Chechens, and Albanians prior to the final establishment of internationally recognized borders
(IRBs). This was necessary because the dataset I employed for ethnic distribution, the GeoEPR,
only dates to 1946. However, a historical map for Korean distribution was not required, as the
current border was finalized in 1948.
The goal of this stage of the analysis was to use pre-border demographic maps to process,
georeference, and relate to the GeoEPR dataset as a means of creating pre-1946 GIS layers for the
three ethnic groups in vector format. This, along with qualitative research into pre-border political
and economic dynamics, identified changes in population and territory for each group. Visualizing
changes is not only useful in its own right; it also provided clues as to where historical review may
best be directed. While there were a great many maps available, criteria for selection eliminated
several. Dozens of maps were found in online archives, but most were rejected out of hand for not
possessing ideal attributes for map processing.
Possible candidates required querying linked data to find other versions of the maps online
that were scanned at higher resolutions, making them more fit for use (Chiang et al., 2014;
Parundekar et al., 2010). In the end, the Library of Congress and David Rumley’s Collections had
the highest quality images available. These maps were originally created by government sources,
such as military cartography units, royal cartographical societies, and well-regarded demographic
cartographers of the day.
The better candidate maps were considered for having clear legends and color-coding. All
are in the Latin script, in either English or German. The priority was to find maps with high quality

384

legends and clear colors representing population, which makes maps suitable to relate to the
GeoEPR.
Figure B.1 displays the first historical map used in this work, which was utilized in the case
of Kurdish factions:
Figure B.1
Pre-Lausanne map of Kurdish distribution

Note. Map of Eastern Turkey in Asia, Syria and Western Persia (Ethnographical). By the Royal
Geographical Society (UK), 1910/updated in 1918. Scale of 1:2,000,000. Format: TIFF
(300 DPI). Downloaded from the Library of Congress.

385

The borders of Iran, Iraq, Syria, and Turkey were formalized with the Treaty of Lausanne
in 1923. The updated version of the map was created only a few years prior to the finalization of
these borders. Most of the ethnic cleansing in the former Ottoman province of Anatolia ended in
1917, which places this map between those events and the Treaty of Lausanne. As such, it covered
an ideal time frame for use in this project. It contained the aftermath of the latest period of
migration but occurred before those which came about after the finalization borders but before the
beginning of the GeoEPR.
Its suitability is passable. It is in the Latin script, in English, and appears to use a common
font. Its text is less than ideal due to varying degrees of curvature and spacing, as well as the
inclusion of topography. However, the color coding and legend are clear, and the boundaries of
ethnic predominance are what I am interested in. The format is an ideal type and the image quality
is acceptable at 300DPI.
The major limitation with this map was its propensity to display areas of mixed populations
as bars with squared off edges. This issue was mitigated by including areas in which Kurds were
in the majority in the statistical computations of territorial predominance and excluding areas in
which they were not in the majority. In practice, yellow areas that were squared off were counted,
while non-yellow areas were not.
Figure B.2 displays the historical demographic map used for processing the distribution of
Chechens prior to the establishment of the borders of the Soviet Union and the Russian Federation,
which are covered by GADM and GeoEPR sources.

386

Figure B.2
Pre-Soviet map of Chechen distribution

Note. Ethnographische Karte von Russland (Sudliches Blatt). Created by August Heinrich
Peterman and Aleksandr Rittikh (Justes Perthes, Gotha, Germany), 1878. Scale of
1:3,700,000. Format: JPEG2000. Downloaded from David Rumsey Map Collection.

Russia completed its conquest of what is now Chechnya in the mid-1860s, just prior
to the update to this map, although sporadic resistance continued. The entirety of the Chechen and
Ingush peoples were deported after World War II, and they returned in 1956. That time period is

387

covered by the GeoEPR, so a map made after the Russian conquest and prior to the Soviet
deportation was ideal. This is because the geographic extent of the Chechen population, along with
qualitative research into historical populations prior to deportation, may help to explain political
development over time. There were several potential maps to choose from that were made during
this period. Most of them exhibited low suitability for processing due to scans of less than 300DPI
or being scanned and saved as inappropriate formats.
This map is from the correct time frame. It has a legend, and it utilizes a unique purple
color for Chechen distribution that is easily distinguished from neighboring populations. It is in
JPEG2000 format, which may not be the very best quality format for the image, but it is acceptable.
Additionally, the image can be zoomed in to a very high resolution with Irfan v4.56, which makes
up for possible shortcomings.
Figure B.3 displays the historical demographic map used for the distribution of Albanians
prior to the solidification of IRBs following the end of the Socialist Federal Republic of Yugoslavia
in which Kosovo became part of the Federal Republic of Yugoslavia:

388

Figure B.3
Pre-Socialist Federal Republic of Yugoslavia map of Albanian distribution

Note. Die Völker des Donauraumes und der Balkanhalbinsel. Created by the Second
Cartography Battalion, German Heer (Berlin, Germany), 1940. Scale of
1:3,000,000. Format: TIFF (300DPI). Downloaded from the Library of Congress.

A hypothetical Greater Albania as it was understood by earlier generations of Albanian
irredentists would include all of the contiguous territory inhabited by Albanians before the end of

389

the Second World War. An ideal ethnographic map would come from the period of time before
the Axis-supported Albanian Kingdom (1939-1943), prior to the splitting of ethnic Albanians by
IRBs following the war by communist regimes.
This map is dated from shortly before the fixing of modern borders, squarely in the desired
time frame. Furthermore, it contains a clear legend and color-coding scheme. Its image quality is
acceptable. Though text processing was not utilized, important text, is clear and unobstructed,
which is helpful for assigning control points.
All map processing was conducted using ArcGIS Pro. Processed maps will also be
maintained in a format usable with QGIS, PostgreSQL, and PostGIS in the event of institutional
limitations.
Appendix B References
Chiang, Y.-Y., Leyk, S., & Knoblock, C. A. (2014). A survey of digital map processing techniques.
ACM Computing Surveys, 47(1), 1-44. https://doi.org/10.1145/2557423
Esri (2020). ArcGIS Pro. https://www.esri.com/en-us/home
Parundekar, R., Knoblock, C. A., & Ambite, J. L. (2010). Aligning ontologies of geospatial linked
data. CEUR Workshop Proceedings, 691.
Peterman, A. H., & Rittikh, A. (1878). Ethnographische Karte von Russland (Sudliches Blatt).
Justes Perthes.
https://www.davidrumsey.com/luna/servlet/detail/RUMSEY~8~1~306216~90079057:Et
hnographische-Karte-von-Russland-
?sort=pub_list_no_initialsort%2Cpub_date%2Cpub_list_no%2Cseries_no&qvq=q:sudlic

390

hes%20blatt;sort:pub_list_no_initialsort%2Cpub_date%2Cpub_list_no%2Cseries_no;lc:
RUMSEY~8~1&mi=2&trs=15
PostGIS (2020). Spatial and Geographic objects for PostgreSQL, version 3.0.4. https://postgis.net/
PostgreSQL Global Development Group (2020). PostgreSQL 13.1. https://www.postgresql.org/
QGIS.org (2020). QGIS Geographic Information System. QGIS Association. http://www.qgis.org
Royal Geographical Society. (1910/1918). Map of Eastern Turkey in Asia, Syria and Western
Persia (ethnographical). https://www.loc.gov/item/2007633929/.
Second Cartography Battalion (1940). Die Völker des Donauraumes und der Balkanhalbinsel.
German Heer. https://www.loc.gov/item/2007627467/
Skiljian, I. (2020). IrfanView v.4.56. https://www.irfanview.com/

391

Appendix C: An elaboration on the statistical analyses
The formulae for the statistical analyses and their explanations are further elaborated upon
in this appendix. They will be presented in the order in which they were discussed in Chapter
Three.
Map processing
Let us begin with map processing. Rubbersheeting triangulates between two maps: a stable
one with known features, and the map the user wishes to transform. The process is known as
piecewise linear homeomorphism (PLH) and the variety used in this analysis was broadly
influenced by White and Griffin (1985).
In PLH, control points are represented by 𝑃 𝑖 ( 1 ≤ 𝑖 ≤ 𝑛 ) ; 𝑢 𝑖 is a coordinate vector for the
map to be subjected to rubbersheeting; 𝑣 𝑖 is the coordinate vector in the comparison map. Both 𝑢 𝑖
and 𝑣 𝑖 represented the solved values for 𝑃 𝑖 in their respective maps. x represents all potential
coordinates for the map that is to be transformed. Finally, the transformation matrix is constructed
by solving for 𝑇 𝑗 ( 1 ≤ 𝑗 ≤ 2𝑛 + 2) . The matrix is constructed as a three-by-three matrix. 𝑇 𝑗 is
solved through:
𝑣 𝑘 = 𝑇 𝑗 𝑢 𝑘 ; (1)
𝑣 𝑙 = 𝑇 𝑙 𝑢 𝑚 ; (2)
𝑣 𝑚 = 𝑇 𝑚 𝑢 𝑚 (3)
and 𝑝 𝑘 , 𝑝 𝑙 , and 𝑝 𝑚 are the vertices of j. Vector-notation equations are then determined with:

[𝑣 𝑥 𝑣 𝑦 1] = [
𝑡 11
𝑡 12
𝑡 13
𝑡 21
𝑡 22
𝑡 23
𝑡 31
𝑡 32
𝑡 33
][𝑢 𝑥 𝑢 𝑦 1].

(4)

392

Vertices for j are then solved:
𝑣 𝑥 = 𝑢 𝑥 𝑡 11
+ 𝑢 𝑦 𝑡 21
+ 𝑡 31
, (5)
𝑣 𝑥 = 𝑢 𝑥 𝑡 12
+ 𝑢 𝑦 𝑡 22
+ 𝑡 32
, (6)
𝑣 𝑥 = 𝑢 𝑥 𝑡 13
+ 𝑢 𝑦 𝑡 23
+ 𝑡 33
(7)
Coordinates x are then located on the map to be transformed as x’ = 𝑇 𝑗 𝑥 . The corresponding
coordinates on the known map are then identified using the displacement of the closest control
point with:
𝑣 𝑛 +𝑖 = 𝑢 𝑛 +𝑖 + ( 𝑣 𝑗 − 𝑢 𝑗 ) , (8)
where j is selected in a manner that allows |𝑣 𝑛 +𝑖 − 𝑣 𝑗 | to be over 1 ≤ 𝑗 ≤ 𝑛 .
Objects of contestation
We turn now to the tests involving the Effects of Distance (P2), beginning with Pearson’s
Correlation. This simply tests the linear relationship between covariates x and y. Its formula is
given as:

𝑟 =
∑ ( 𝑥 𝑖 − 𝑥 ̅ ) ( 𝑦 𝑖 − 𝑦 ̅)
√∑ ( 𝑥 𝑖 − 𝑥 ̅ )
2
∑ ( 𝑦 𝑖 − 𝑦 ̿)
2
,

(1)
where r is the correlation coefficient. The range of values and means of values in the two samples
are computed and values are returned ranging from -1 to 1, which correspond to perfect disparity
to perfect correlation, respectively (Wright, 1921).
The next step in the analysis was ordinary least squares (OLS) regression modeling. The
following is primarily influenced by Dougherty (2002). OLS models a s linear relationship
between the sum of the squares of differences between the DV and EVs. Its formula is given as:

393

𝑌 = 𝛽 0
+ ∑ 𝛽 𝑗 𝑋 𝑖𝑗
+ 𝘀 𝑗 =1…𝑝 , (1)
where Y is the DV; 𝛽 0
is the model’s intercept; X represents a particular EV, while j is the range
of EVs; and 𝘀 is the random error.
OLS models are a starting point for spatial analytics but are quite limited due to their
assumption of a normal distribution for the data. As this is not always the case, we may first try
and transform skewed data to fit a normal distribution through logarithmic transformation
(Manning & Mullahy, 2001).
We begin with the linear model:
𝑌 = 𝛽 0
+ ∑ 𝛽 𝑗 𝑋 𝑖𝑗
+ 𝘀 𝑗 =1…𝑝 , (1)
however, 𝘀 does not have a normal distribution, given the nature of spatial data. And so, a log-
transformation is used with the formula:
𝐸 ( 𝑦 |𝑥 ) = 𝑓 ( 𝑥 ) exp ( 𝑥 , 𝛿 ) , (2)
where 𝐸 ( 𝑦 |𝑥 ) is the transformed data based on a function ((𝑓 ( 𝑥 ) ) of the log-scale prediction of the
error term of the linear model (exp ( 𝑥 , 𝛿 ) ).
Next, the data was tested for spatial autocorrelation using Moran’s I (Moran, 1950). The
formula is given as:

𝐼 =
𝑁 𝑊 ∑ ∑𝑤 𝑖𝑗
( 𝑥 𝑖 − 𝑥 ̅ ) ( 𝑥 𝑗 − 𝑥 ̅ )
𝑗 𝑖 ∑ ( 𝑥 𝑖 − 𝑥 ̅ )
2
𝑖 ,

(1)
where I is the resultant statistic ranging from -1 (perfect dispersal) to 1 (perfect autocorrelation). i
and j are the indexing functions for the variable of interest x, which give us the number of spatial
units N. 𝑤 𝑖𝑗
is the spatial weights matrix, while W is the sum of all matrices.

394

Getis-Ord General G and Gi* statistics test for clustering in the data. The General G statistic
is given as:

𝐺 =
∑ ∑ 𝑤 𝑖 𝑛 𝑗 =1
,𝑥 𝑖 𝑥 𝑗 𝑛 𝑖 =1
∑ ∑ 𝑥 𝑖 𝑥 𝑗 𝑛 𝑗 =1
𝑛 𝑖 =1
, ∀
𝑗 ≠ 𝑖 ,

(1)
where i and j are features being compared for attribute value x. Again, w is a spatial weights matrix.
The total number of features from the dataset is represented by n. Finally, ∀
𝑗 ≠ 𝑖 means that j and
i cannot refer to the same feature as the other.
The z-score (standard deviation indicating clustering probabilities with an associated p-
score) for the General G is then computed as:

𝑧 𝐺 =
𝐺 −E[𝐺 ]
√V[𝐺 ]
,

(2)
where

E[𝐺 ] =
∑ ∑ 𝑤 𝑖 𝑛 𝑗 =1
,𝑥 𝑖 𝑥 𝑗 𝑛 𝑖 =1
∑ ∑ 𝑥 𝑖 𝑥 𝑗 𝑛 𝑗 =1
𝑛 𝑖 =1
, ∀
𝑗 ≠ 𝑖 ,

(3)
and
V[𝐺 ] = E[𝐺 2
] − E[𝐺 2
]. (4)
Once autocorrelation and clustering were detected, the next step to create a spatially lagged
autoregressive model to accurately assess the EVs in P2 with an improved maximum likelihood
estimator (MLE) (Anselin, 2002; Brueckner, 2003; Seya et al., 2020; Whallon, 1974). The formula
for this model is:

𝛾 𝑖 = 𝜌 ∑ 𝑤 𝑖𝑗
𝛾 𝑗 𝑁 𝑗 =1
+ ∑ 𝑥 𝑘 ,𝑖 𝐾 𝑘 =1
𝛽 𝑘 + 𝘀 𝑖 ,
(1)

395

where ∑ 𝑥 𝑘 ,𝑖 𝐾 𝑘 =1
𝛽 𝑘 + 𝘀 𝑖 is the adjusted linear model. 𝜌 represents the model’s specified spatial
parameters. The spatial weights matrix is once again 𝑤 𝑖𝑗
𝛾 𝑗 .
The robustness check selected for the spatial lag model was the studentized Breusch-Pagan
(1979; Koenker, 1981) test. This is a chi-squared test in which the auxiliary regression equation’s
test statistic is distributed 𝑛𝑋
2
on k degrees of freedom. This assesses the null hypothesis of
homoscedasticity at the threshold of p < 0.05.
Rival cooperation
Let us now move on to the analyses of cooperation resulting from Out-group Incursion
(P3), beginning with kernel density estimation (KDE). The information for KDE is drawn mainly
from Gramacki (2018). The basic formula for a univariate KDE is:

𝑓 ̂
( 𝑥 , ℎ) =
1
𝑛 ℎ
∑ 𝐾 𝑛 𝐼 =1
(
𝑥 − 𝑋 𝑖 ℎ
) =
1
𝑛 ∑ 𝐾 ℎ
( 𝑥 − 𝑋 𝑖 ) ,
𝑛 𝐼 =1

(1)
where𝑓 ̂
is the kernel—or density—to be determined. Xi is the sample of instances that are to be
rendered into kernels. The scale Kh is determined by:
𝐾 ℎ
( 𝑢 ) = ℎ
−1
𝐾 ( ℎ
−1
𝑢 ) . (2)
The bandwidth h is a standard deviation of a normal distribution. It is a positive number selected
by the user, and it is based on the size of the data. The square root of the number of observations
is frequently used. The distribution is given as:

𝑁 ( 𝑥 ; 𝜇 , 𝜎 ) =
1
√( 2𝜋 ) 𝜎 exp (−
( 𝑥 − 𝜇 )
2
2𝜇 2
),

(3)
and h is given as:

396

𝑓 ̂
( 𝑥 , ℎ) =
1
𝑛 ∑ 𝑁 ( 𝑥 ; 𝑋 𝑖 , ℎ
2
) ,
𝑛 𝐼 =1

(4)
where the kernels are given as:

𝐾 (
𝑥 − 𝑋 𝑖 ℎ
) =
1
√( 2𝜋 ) 𝜎 exp (−
( 𝑥 − 𝑋 𝑖 )
2
2ℎ
2
).

(5)
So, taken together, the formula for the KDE becomes:

𝑓 ̂
( 𝑥 , ℎ) =
1
𝑛 ℎ
∑
𝑛 𝐼 =1
=
1
𝑛 ∑
𝑛 𝐼 =1
1
√( 2𝜋 ) 𝜎 exp (−
( 𝑥 − 𝑋 𝑖 )
2
2ℎ
2
) .

(6)

K- and L- function analyses follow KDE. A K-function analysis was performed to examine
when cooperative action events may become unrelated to each other. This function classifies the
data by its likelihood of spatial randomness (Wang et al., 2020). Its equation is given as:

𝐾 ̂
( 𝑠 , 𝑡 ) =
𝐴 ∙ 𝐷 𝑛 2
∑
𝑛 𝑖 =1
∑
𝑛 𝑗 ≠1
𝑙 𝑠 ,𝑡 ( 𝑑 𝑖𝑗
, 𝑢 𝑖𝑗
)
𝜔 𝑖𝑗
𝜐 𝑖𝑗
,

(1)
where A and D are the study area, with A indicating the geographic space and D refers to the
temporal limit. 𝑙 𝑠 ,𝑡 ( 𝑑 𝑖𝑗
, 𝑢 𝑖𝑗
) is an indicator function that works in conjunction with the spatial and
temporal weights matrices of w and v.
The L-function transforms the K-function, which is particularly useful for kernels that may
result from lower numbers of observations than others. Its formula is:

𝐿 ̂
( 𝑠 , 𝑡 ) = √𝐾 ̂
( 𝑠 , 𝑡 ) /2𝜋𝑡 − 𝑠 ,
(1)
The average nearest neighbor (ANN) analysis was next. The ANN compares an observed
distribution to that of a constructed distribution under CSR (Dixon, 2002). Its equation is:

397

𝐺 ( 𝑤 ) = 1 − 𝑒 −𝜌𝜋 𝑤 2
,
(1)
And the Monte Carlo simulation creates N random locations are simulated and repeated R times
under the null hypothesis, which is CSR. The p-value is derived through
1+𝑁 1+𝑅 .
The Poisson point pattern (PPP) analysis was the final portion of the analysis of rival
cooperation. This test predicts the probability of an outcome based on underlying spatial effects
on a phenomenon of interest. It consists of an observed spatial phenomenon and multiple user-
defined parameters that might account for its occurrence (Diggle & Rowlingson, 1994). The
phenomenon is indicated by xi ,…, n. A PPP equation that accounting for multiple values is given
as:

𝑓 ( 𝑥 ; 𝜃 1
, … , 𝜃 𝑘 , ∅) = ∏ 𝑔 ( 𝑥 − 𝑥 0𝑘 ; 𝜃 𝑘 ) exp
𝑞 𝑘 =1
{∑ ∅
𝑗 𝑧 𝑗 ( 𝑥 )
𝑟 𝑗 =1
},

(1)
where f(x; 𝜃 1
,…,θk,∅) is the unknown effect of some independent process on the outcome. q
represents the number of event-locations. g is the increase in probability of a particular explanatory
variable. zj (x) indicates the potential covariates.
x0 requires that f(x – x0; θ) → 1. Thus, (x – x0) ́(x – x0) → ∞, which implies that ρ(x) → ρ/(1
+ ρ). The unknown value can only yield a value between 0 and 1. A log-likelihood function for ρ
and θ would take the Bernoulli form:

𝐿 ( 𝜌 , 𝜃 ) = ∑ log 𝜌 ( 𝑥 𝑖 )
𝑛 𝑖 = 1
+ ∑ log{ 1 − 𝜌 ( 𝑥 𝑖 ) }
𝑛 +𝑚 𝑖 =𝑛 + 1
,

(2)
where 𝑛 + 𝑚 becomes the number of events and control locations. Substitution and maximizing
𝐿 ( 𝜌 , 𝜃 ) gives us the next portion of the equation:

398

𝐿 ( 𝜌 , 𝜃 ) = 𝑛 log 𝜌 + ∑ log 𝑓 ( 𝑥 𝑖 − 𝑥 0
; 𝜃 )
𝑛 𝑖 = 1
− ∑ log 𝜌 { 1 + 𝜌 𝑓 ( 𝑥 𝑖 − 𝑥 0
; 𝜃 ) }
𝑛 =𝑚 𝑖 = 1
,

(3)
Finally, arranging for parametrization allows the equation to be reduced to:
𝐿 0
( 𝜌 ) = 𝑛 log 𝜌 − ( 𝑛 + 𝑚 )log( 1 + 𝑝 ) , (4)
where maximization takes place with 𝜌 ̂ = 𝑛 /𝑚 . Hypothesis testing with regards to 𝜃 is conducted
using the weights matrix of each parameter’s observed values. The test performs numerical
maximization of the likelihood ratio method. This statistically compares 𝐷 = 2{ 𝐿 ( 𝜌 , ̂ 𝜃 ̂
)−
𝐿 0
( 𝜌 ̂
0
)} as it occurs with each parameter.

Appendix C References
Anselin, L. (2002). Under the hood: Issues in the specification and interpretation of spatial
regression models. Agricultural Economics, 27(3), 247-267.
https://doi.org/10.1111/j.1574-0862.2002.tb00120.x
Breusch, T., & Pagan, A. (1979). A simple test for heteroscedasticity and random coefficient
variation. Econometrica, 47(5), 1287-1294. https://doi.org/10.2307/1911963
Brueckner, J. K. (2003). Strategic interaction among governments: An overview of empirical
studies. International Regional Science Review, 26(2), 175-188.
https://doi.org/10.1177/0160017602250974
Diggle, P. J., & Rowlingson, B. S. (1994). A conditional approach to point process modelling of
elevated risk. Journal of the Royal Statistical Society Series A (Statistics in Society),
157(3), 433–440. https://doi.org/10.2307/2983529

399

Dixon, P.M. (2002.) Nearest-neighbor contingency table analysis of spatial segregation for
several species. Écoscience, 9(2), 142-151.
https://doi.org/10.1080/11956860.2002.11682700
Dougherty, C. (2002). Introduction to econometrics (2
nd
ed.). Oxford University Press.
Getis, A., & Ord, J. (1992). The analysis of spatial association by use of distance statistics.
Geographical Analysis, 24(3), 189–206.
https://doi.org/10.1111/j.1538-4632.1992.tb00261.x
Gramacki, A. (2018). Nonparametric kernel density estimation and its computational aspects.
Springer International Publishing. https://doi.org/10.1007/978-3-319-71688-6
Koenker, R. (1981). A note on studentizing a test for heteroscedasticity. Journal of
Econometrics, 17(1), 107–112. https://doi.org/10.1016/0304-4076(81)90062-2
Manning, W. G., & Mullahy, J. (2001). Estimating log models: To transform or not to
transform? Journal of Health Economics, 20(4), 461–494.
https://doi.org/10.1016/S0167-6296(01)00086-8
Moran, P. A. P. (1950). Notes on continuous stochastic phenomena. Biometrika, 37(1-2), 17–23.
Seya, H., Yoshida, T., & Yamagata, Y. (2020). Spatial econometric models. In Y. Yamagata &
H. Seya (Eds.), Spatial analysis using big data: Methods and urban applications (pp.
113-158). Academic Press. https://doi.org/10.1093/biomet/37.1-2.17
Wang, Y., Gui, Z., Wu, H., Peng, D., Wu, J., & Cui, Z. (2020). Optimizing and accelerating
space–time Ripley ’s K function based on Apache Spark for distributed spatiotemporal

400

point pattern analysis. Future Generation Computer Systems, 105, 96-118.
https://doi.org/10.1016/j.future.2019.11.036
Whallon, R. (1974). Spatial analysis of occupation floors II: The application of nearest
neighbor analysis. American Antiquity, 39(1), 16–34. https://doi.org/10.2307/279216
White, Jr., M. S., & Griffin, P. (1985). Piecewise linear rubber-sheet map transformation.
American Cartographer, 12(2), 123-131. https://doi.org/10.1559/152304085783915135
Wright, S. (1921). Correlation and causation. Journal of Agricultural Research, 20(7), 557–585.

401

Appendix D: Case selection table
This appendix displays the table used in case selection. It is based on information from the
Center for Systemic Peace. The table is intended to serve as an eventual defined universe of
potential cases for further study as explained in Chapter Nine.

Appendix D References
Marshall, M. G. (2017). Major episodes of political violence 1946-2016. Center for Systemic
Peace.
Marshall, M. G. (2020). Current status of the world's major episodes of political violence. Monthly
reports to US Government's Political Instability Task Force.
Marshall, M. G., & Elzinga-Marshall, G. (2017). Global Report 2017: Conflict, governance, and
state fragility. Center for Systemic Peace. http://www.systemicpeace.org/globalreport.html

402

Table D.1
Case selection process
Conflict Periods
1

Year(s) Type(s) Group(s) Statelessness Division Claim of Case
State(s) of Violence by IRB self- Types
determination (Y/N)

1965-OG EW Palestinians A Y Y DSN
Israel/ (N)
Palestine
1988-OG EW Armenians, N Y Y DIG
Azerbaijan Azeris (N)
1991 EV Nepalese N Y NA —
Bhutan
1991-1993 EW Abkhaz, Y/Y Y/Y Y/Y DSN
Georgia Ossetians (N)
1991-1993 CEV Kalenjia, N N N —
Kenya Masai,
Kikuyi, Luo
1991-1995 EW Croats, Serbs N Y Y DIG
Croatia (N)
1991-1992 EV Russians N Y Y DIG
Moldova (N)
1991-2002 CEV(S) Hindus, N Y NA —
India Muslims

1992-1995 EW Serbs, N Y Y DIG
Yugoslavia Croats, (N)
Bosniaks
1992-1999 EV Cassamance Y N N —
Senegal
1993-2005 CEV Hutus, Tutsis A Y NA —
Burundi

1
Legend appears under Note below.

403

Year(s) Type(s) Group(s) Statelessness Division Claim of Case
State(s) of Violence by IRB self- Types
determination (Y/N)

1994 CEV Hutus, Tutsis A Y NA`
Rwanda
1994 EV Zapatistas Y N Y SSS
Mexico (N)
1994 EW(S) Sunnis, N N N
Yemen Shiites
1994-1996 EW Chechens Y N Y SSS
Russia (Y)
1994-1998 EW Hutus, Tutsis A Y NA
Rwanda
1994-1998 EW Kurds Y Y & N Y DSN
Iraq (Y)
1997-2005 EV Acehnese Y N Y SSS
Indonesia (N)
1997-2010 CEV Delta, Ijaw, Y N N
Nigeria Itsekeri
1997-OG EW/EV(S) Pashtuns Y & N Y & N Y DSN
Afghanistan, Hazara, (N)
Pakistan Turkics
1998-1999 EW Kosovar N Y Y DIG
Yugoslavia (Y)
1998 EW Abkhazians A Y Y DSN
Georgia (N)
1998-2003 CEV Malaita, Y N N
Solomon Isatabu
Islands
1999-2002 CEV Moluccas Y N N
Indonesia
1999-2000 EW Oromo Y N Y SSS
Ethiopia (N)
1999-2006 EW Chechens Y N Y SSS
Russia (Y)

404

Year(s) Type(s) Group(s) Statelessness Division Claim of Case
State(s) of Violence by IRB self- Types
determination (Y/N)

2001 CEV Dayaks, Y N N
Indonesia Madurese
2001 EW Hutus, A Y NA
Rwanda regime
2001-2004 EV Kano A Y NA
Nigeria
2003-OG CEV(S) Darfur A/N/Y Y/N NA/Y DSN
Sudan (several (N)
groups)
2004-OG CEV Malaysians N N N
Thailand
2004-OG EW Kurds Y Y Y DSN
Turkey (Y)
2004-OG EV Balochis Y Y Y DSN
Iran, Pakistan (N)
2004-2014 EV/CW Houthis N N N
Yemen
2005-OG CEV Seleka Y N N
Central
African
Republic
2006-2010 CEV Toroboros, A Y NA
Chad Janjawid
2007-2008 EV Palestinians A Y Y DSN
Lebanon (N)
2007-OG EV Oromo, Y/N N/Y Y/Y SSS,
Ethiopia Somalis DIG
(N,N)
2008-OG EV Chechens, Y/Y/Y N/N/N Y/Y/Y SSS
Russia Dagestanis, (Y)
Ingushetians

405

Year(s) Type(s) Group(s) Statelessness Division Claim of Case
State(s) of Violence by IRB self- Types
determination (Y/N)

2009-OG EV Uighurs Y Y Y DSN
China (N)
2009-OG CEV Lou Nuer, Y N NA
South Sudan Murle
2010
Kyrgyzstan CEV Kyrgiz, N Y NA
Uzbeks
2010-OG EW/ EW(S) Boko Haram, Y Y Y DSN
Nigeria Kanuri,
Others (N)
2011-OG EW/ EW(S)/ Sunnis, N/N/ Y/Y/ NA/NA/ DSN
Iraq CEV Shiites, Y/Y/ Y/Y/ Y/Y/ (Y)
Kurds, Y/Y Y/Y Y/Y
Yazidis,
Assyrians,
Turkmen
2011-OG EW/ EW(S)/ Sunnis, N/Y/ Y/Y/ NA/N/ DSN
Syria CEV Druze, A/N/ Y/Y/ N/Y/ (Y)
Alawites, Y/Y Y/Y Y/Y
Kurds,
Assyrians,
Turkmen
2012-OG EV Sindhi Y Y Y DSN
Pakistan (Y)
2013-OG CEW Nuer, Dinka Y/A Y/Y NA
South Sudan
2013-2015 EV Mai Mai Y N Y SSS
Democratic (many others) (N)
Republic
of Congo
2014-OG EW Russians N Y Y DIG
Ukraine (N)
2015-OG EW/ EW(S) Boko Haram/ Y Y Y DSN
Cameroon Kanuri, Others (N)

406

Year(s) Type(s) Group(s) Statelessness Division Claim of Case
State(s) of Violence by IRB self- Types
determination (Y/N)

2015-OG EW(S) Houthis N N N
Yemen
2016-OG EV Kamwinda Y N N
Dem. Rep. Nsapu
of Congo
2017-OG EV Rohingya Y Y Y DSN
Myanmar (N)

Note. See below for an explanation of terms.

Year(s)
OG = Ongoing. The conflict has yet to terminate.
Types of Violence
EW = Ethnic War: Two or more factions with political-military forces organized along ethnic lines
engaged in warfare. Warfare is armed conflict involving battlefield deaths greater than zero, in
which a political aim is apparent.
EW(S) = Ethnic War (with Sectarian focus): A variety of ethnic war, but when religious cleavages
are the dominant source of the conflict. This depends on the claims of the fighters involved,
whether or not political grievances or resources may have greater social scientific explanatory
power.
EV = Ethnic Violence: Armed conflict that does not require a battlefield or two similarly organized
forces facing each other. One faction lacks a unified political-military force, and so may have a
militia or a political movement, but not both. This is often state-directed.
CEV = Communal Ethnic Violence. A variety of ethnic violence. Both involved parties lack a
unified political-military force. Often this is perpetrated by militias or arise from riots.
Statehood Status
A = Ambiguous. The status of statelessness is not clear. Palestinians, Ossetians, and Abkhazians
have recognition from several, but not all, member states of the UN. Others, if forming the core
population of a state at varying times, or of uncertain ethnic differentiation are also considered
ambiguous for purposes of assigning statelessness.
Other
Y = Yes.
N = No.
NA = Not Available/Not Applicable. If there is no information on a self-determination movement,
or if all parties involved have representative states, this value is assigned.

407

Appendix E: Preliminary tests of intra-Kurdish conflict and cooperation
This appendix contains the results of tests leading up to those presented in the Kurdish
case. They will be displayed in the order in which they were performed.
Roots of Rivalry (P1)
Let us begin with figure E.1:
Figure E.1
Georeferenced historical distribution of Kurdish populations

Note. Map of Eastern Turkey in Asia, Syria and Western Persia (Ethnographical), created in 1910
and updated in 1918. The distribution of Kurds has been processed in yellow and overlain
on the original map. Darker hues imply a denser population.

408

This was processed in ArcGIS Pro and served as a useful starting point for process tracing.
Effects of Distance (P2)
The first step in assessing the hypotheses for intraethnic conflict objects was the test for
Pearson correlation coefficients. The visual results of this test are displayed below in figures E.2a
and E.2b:

409

Figure E.2a
Summary statistics for episodes of intra-Kurdish conflict: Pearson correlation (distance from
regional centers)
1994-1997

2011-2020

Note. * Denotes Pearson correlation coefficient with a 95% confidence interval.

410

Figure E.2b
Summary statistics for episodes of intra-Kurdish conflict: Pearson correlation (distance from
borders)
1994-1997

2011-2020

Note. * Denotes Pearson correlation coefficient with a 95% confidence interval.

411

These tests were performed in R. Let us begin with Figure E.2a., which was the regional
centers model. Both periods of conflict appeared most similar to each other with regards to the
relationship between the number of conflict instances and distance from regional centers.
Interestingly, the distance from the national capital appeared to be positive and significant for the
first period, and negative, though insignificant, for the second. The distance from the primary route
was significant for both, yet negative for the first period and positive for the second.
Let us move to the second series of models in figure E.2b, which concern distance from
the border and the number of conflict events. Again, the distance from the border appeared to be
the most similar in each time period. Both were negative and significant. However, the distance
from the primary route differed for the first period in comparison to the regional centers model as
it was positive, rather than negative. Additional tests helped further examine these trends.
The next analysis involved OLS. The results of the OLS models, the logarithmically
transformed models, and the residual plots are displayed below in tables E.1a and E.1b, and figures
E.3a and E.3b:

412

Table E.1a
Summary statistics: Ordinary least squares (OLS) regression models of episodes of intra-Kurdish
conflict

OLS Models
1994-1997 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.129**
(0.050)

Distance from Primary Route (km)

0.302
(0.266)
0.246***

(0.027)
Distance from Regional Center (km)

-0.260***
(0.064)

Distance from Border (km)

-0.116***
(0.040)
Constant

-11.258
(15.964)
3.005
(1.791)
Observations 62 45
R2 0.552 0.786
Adjusted R2 0.529 0.776
Residual Std. Error 10.080
(df = 58)
3.808
(df = 42)
F Statistic

23.796
***

(df = 3; 58)
77.297
***

(df = 2; 42)
Note: *p<0.1; **p<0.05; ***p<0.01

413

OLS Models
2011-2020 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.001
(0.012)

Distance from Primary Route (km)

0.351***
(0.059)
0.103
(0.151)
Distance from Regional Center (km)

-0.081**
(0.032)

Distance from Border (km)

-2.566
(1.572)
Constant

7.493
(5.467)
14.035***
(4.448)
Observations 26 15
R
2
0.758 0.368
Adjusted R
2
0.725 0.262
Residual Std. Error
2.659
(df = 22)
4.473
(df = 12)
F Statistic
23.011
***

(df = 3; 22)
3.491
*

(df = 2; 12)
Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

414

Table E.1b
Summary statistics: Logarithmically transformed regression models of episodes of intra-Kurdish
conflict with stepwise selection based on Bayesian information criterion (BIC)

Log-Transformed OLS Models with BIC
1994-1997 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.101**
(0.044)

Distance from Regional Center (km)

-0.252***
(0.063)

Distance from Border (km)

-0.116***
(0.040)

Distance from Primary Route (km)

-0.116***

(0.027)

Observations 62 45
R2 0.542 0.786
Adjusted R2 0.526 0.776
Residual Std. Error 10.104
(df = 59)
3.808
(df = 42)
F Statistic

34.886***
(df = 2; 59)
77.297
***

(df = 2; 42)
Note: *p<0.1; **p<0.05; ***p<0.01

415

OLS Models
2011-2020 Intra-Kurdish Conflict

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.349
***

(0.054)

Distance from Regional Center (km)

-0.082
**

(0.031)

Distance from Border (km)

-3.210**
(1.231)
Distance from Primary Route (km)

7.952
***

(0.895)
16.483
***

(2.570)
Observations 26 15
R
2
0.758 0.343
Adjusted R
2
0.737 0.293
Residual Std. Error
2.601
(df = 23)
4.380
(df = 13)
F Statistic
36.070
***

(df = 2; 23)
6.796
**

(df = 1; 13)

Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

416

Figure E.3a
Plots of log-transformed regression models for episodes of intra-Kurdish conflict: Regional
Centers Model (1)
1994-1997

2011-2020

417

Figure E.3b
Plots of log-transformed regression models for episodes of intra-Kurdish conflict: Borders
Model (2)
1994-1997

2011-2020

418

These tests were performed using R. The OLS results were similar to the Pearson results, but the
log-transformed model appeared to have somewhat different outcomes with regards to distance
from the national capital. More conflict events appeared to be located farther from the national
capital. This is expected under the National Capital hypothesis (H2a).
The results of tests regarding the Regional Center (H2b) and Unpacific Borders (H2c)
hypotheses also demonstrate support. However, these tests are not always ideal for spatial data,
and so further investigation was required.
The next step was to test for autocorrelation using Moran’s I. This is displayed below in
table E.2, with Moran scatterplots featured in figure E.4:
Table E.2
Measures of spatial autocorrelation of intra-Kurdish conflict events: Moran’s I statistic

I statistic Expectation Variance z-score Pseudo p-
value

1994-1997
Global 0.7995 -0.0093 0.0013 22.067 < 2.2e-16
EBS 0.27795 NA NA NA 0.001

2011-2020
Global 0.6956 -0.0400 0.0097 7.4569 8.856e-14
EBS 0.2848 NA NA NA 0.003

Note. EBS is Empirical Bayesian Smoothing. All statistics are rounded to the nearest ten
thousandth. z-score is equal to one standard deviation. EBS tests were performed with Monte
Carlo simulations under randomization in which n = 1000, and so Expectation, Variance, and z-
score are not yielded.

419

Figure E.4
Moran scatterplots of spatial autocorrelation of intra-Kurdish conflict events: Empirical
Bayesian smoothing (EBS)

420

These tests were performed in R. It was apparent that the global Moran’s I test for autocorrelation
resulted in strong and significant autocorrelation for both periods. EBS results were not quite as
strong, but they were nevertheless statistically significant. This suggested that the data were more
autocorrelated than we would suspect under CSR.
Once autocorrelation was established, a test for clustering was performed. The results for
the Getis-Ord General G are displayed below in table E.3:
Table E.3
Measures of spatial clustering of intra-Kurdish conflict events: Getis-Ord General G

1994-1997
High-Low Clustering Report
Observed General G: 0.009390
z-score: 2.870830
p-value: 0.004094
Given the z-score of 2.8708304865078333, there is a less than 1% likelihood that this high-clustered
pattern could be the result of random chance.

General G Summary

Observed General G: 0.009390
Expected General G: 0.009346
Variance: 0.000000
z-score: 2.870830
p-value: 0.004094

421

2011-2020
High-Low Clustering Report
Observed General G: 0. 068451
z-score: 2. 474254
p-value: 0. 013351
Given the z-score of 2.474254451361571, there is a less than 5% likelihood that this high-clustered
pattern could be the result of random chance.

General G Summary

Observed General G: 0. 068451
Expected General G: 0. 040000
Variance: 0. 000132
z-score: 2. 474254
p-value: 0. 013351

This test was performed using ArcGIS Pro and exported directly from that program. It was
apparent from these results that instances of intraethnic conflict were clustered in both periods.
There was very little chance that the clusters could have resulted from CSR. As the clustering
results were so strong, a Getis-Ord Gi* test was not performed.
The next stage in the analysis of intraethnic conflict was spatially lagged autoregression
modeling. This is detailed in Chapter Four. The robustness tests for this stage of the analysis are
displayed here in table E.4:

422

Table E.4

Spatial analysis and modeling: Robustness checks of intra-Kurdish conflict

Global Moran’s I for regression residuals, n = 1000
1994-1997

Regional Centers Model Borders Model

Standard deviate
p-value
Observed
Expectation
Variance

5.1341
2.835e-07
0.2825
-0.0491
0.0042

Standard deviate
p-value
Observed
Expectation
Variance

14.533
< 2.2e-16
0.667
-0.0468
0.0024

2011-2020

Regional Centers Model Borders Model

Standard deviate
p-value
Observed
Expectation
Variance

5.1341
2.835e-07
0.2825
-0.0491
0.0042

Standard deviate
p-value
Observed
Expectation
Variance

14.533
< 2.2e-16
0.667
-0.0468
0.0024

423

Studentized Breusch-Pagan Test

1994-1997

Regional Centers Model Borders Model

BP
p-value

4.7875
0.108

BP
p-value

10.147
0.0063

2011-2020

Regional Centers Model Borders Model

BP
p-value

19.736
0.0002

BP
p-value

11.928
0.0026

These tests were performed in R. They assess how strong the spatial effect on the spatial lag models
were in comparison to the other regression models described above. Both the global Moran’s I and
the studentized Breusch-Pagan results for the spatial lag models suggest that the spatial aspect of
the data was too influential for the OLS and log-transformed models to have yielded reliable
results.
Even when rendered as homoscedastic data, the results remain statistically significant—
with the sole exception of the regional centers model for the first period—though it is only barely
out of a 90% confidence interval. These results suggest that, while the data cannot be completely
rendered as homoscedastic and independent, the spatial lag model is the better choice for such an
analysis.

424

Out-group Incursion (P3)
Let us now examine the results of tests for rival cooperation. The first was interpolation
of the clustering of conflict results using KDE, which is the subject of figure E.5:
Figure E.5
Kernel density estimation plot of cooperation between rival Kurdish factions

Note. Brighter colors denote more intense cooperation.

This test was performed in R. The subnational administrative units that witnessed rival cooperative
action are displayed but were not analyzed as a point-count in order to avoid the modifiable areal
unit problem (MAUP). Individual-level conflict events were analyzed. They were spatially joined
to areal units in order to remove unnecessary areas from the region of study. Within those units
are the kernels, or clusters, of cooperative events.

425

Blue regions denote very sparse areas of cooperation; actions here are likely to be unrelated
to any underlying process. Magenta regions feature cooperative events that are somewhat close to
each other, within a distance band of eight kilometers. Yellow regions result from cooperative
events that are closer together than five kilometers. There were several clusters, though the most
intense were located near Şingal, Iraq and Kobanî, Syria. Tentatively, this supports the Who Dies
and Where They Die hypothesis (H3a).
To determine the optimal ranges for considering cooperative events to be related to each
other, the KDE results were subjected to K- and L-function tests, as shown in figure E.6:

426

Figure E.6
Assessing spatial dependence and randomness

Note. Envelopes represent the confidence interval for spatial randomness.

427

These tests were performed in R. The Kobs(r) represented the trend in observed values that comprise
the density plot. Ktheo(r) is a hypothetical trend under randomness. As Kobs(r) crossed the threshold
of the Ktheo(r) envelope only rarely, the data are likely not random. Cooperative events appeared
to trend toward being unrelated to each other at around thirteen kilometers. They became
considered most likely to be unrelated at 37 kilometers apart.
The L-function analysis was somewhat more precise. The trend toward randomness began
closer to eight kilometers and were considered randomly occurring by 36 kilometers apart. As a
result of this analysis, clusters spaced between eight kilometers and 36 kilometers were considered
possibly related and those 36 kilometers apart or more were considered unrelated.
The last method for which CSR was accounted for was the ANN analysis, which is
displayed in figure E.7:

428

Figure E.7
Comparing the observed model to random models of rival KPF cooperation

Note. The blue line represents the observed average distance between neighboring events.

This analysis was performed in R. The ANN comprises all observations in the entire dataset, not
only the clusters. The average distance between neighbors within the observed events was close
to two kilometers. The expected distance was 21 to 37 kilometers for the randomly generated
distribution. The results were statistically significant, and so the null hypothesis of CSR driving
rival cooperative events was rejected.

429

Appendix F: Preliminary tests of intra-Chechen conflict
This appendix records the results of the initial tests that informed the Chechen case. The tests are
presented in the order in which they are referenced in the case chapter.
Roots of Rivalry (P1)
Let us begin with figure F.1:
Figure F.1
Georeferenced historical distribution of Chechen populations

Note. Ethnographische Karte von Russland (Sudliches Blatt) from 1878. Chechen distribution is
indicated by the vivid purple color.

430

This ethnographic map was processed and georeferenced using ArcGIS Pro. The level of detail in
this map made the assignment of control points quite simple. Unchanging topography, such as
mountain peaks, as well as the exactitude of historically prominent citadels provided features that
resulted in an especially close fit considering the geographic extent of the historical demographic
map.
Effects of Distance (P2)
Moving on to the study of the objects of contestation during periods of intraethnic conflict, the
tests for Pearson correlation coefficients came next. These results are visualized in figures F.2a
and F.2b:

431

Figure F.2a
Summary statistics for episodes of intra-Chechen conflict: Pearson correlation (distance from
regional centers)
1994-2009

2009-2017

Note. * Denotes Pearson correlation coefficient with a 95% confidence interval.

432

Figure F.2b
Summary statistics for episodes of intra-Chechen conflict: Pearson correlation (distance from
borders)
1994-2009

2009-2017

Note. * Denotes Pearson correlation coefficient with a 95% confidence interval.

433

These tests were performed using R. Both models in both periods were remarkably similar
to each other. As expected, conflict events seemed to cluster near the regional center and primary
routes. Interestingly, they also seemed to be correlated with proximity to the national center,
against expectations. These are early tests, so this relationship would certainly be worth exploring
further.
Moving on to the border models, the primary route’s importance seemingly disappeared.
Also, the border appeared completely unimportant. This remained the case even when the entirety
of the national border and internal borders of each neighboring krai were included. This was
against expectations again, so there was another relationship to further explore to see whether the
expectations for each hypothesis were unfounded.
The next stage of the analysis was OLS regression modeling, followed by logarithmic
transformation. The results of both are displayed in tables F.1a and F.1b. I omitted plots of
residuals in the interest of brevity.

434

Table F.1a
Summary statistics: Ordinary least squares (OLS) regression models of episodes of intra-
Chechen conflict

OLS Models
1994-2009 Intra-Chechen Conflict Continuum

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.372*
(0.201)

Distance from Primary Route (km)

1.549***
(0.314)
-0.135

(0.198)
Distance from Regional Center (km)

-2.288***
(0.173)

Distance from Border (km)

0.535***
(0.115)
Constant

-432.565
(297.302)

-5.754
(3.341)
Observations 263 11
R2 0.632 0.753
Adjusted R2 0.628 0.692
Residual Std. Error
36.524
(df = 259)
3.286
(df = 8)
F Statistic
148.223***
(df = 3; 259)
12.208***
(df = 2; 8)
Note: *p<0.1; **p<0.05; ***p<0.01

435

OLS Models
2009-2017 Conflict with Chechen Emirate

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

0.152*
(0.078)

Distance from Primary Route (km)

-0.295
(0.218)
0.130
(0.117)
Distance from Regional Center (km)

-0.664***
(0.076)

Distance from Border (km)

0.830***
(0.126)
Constant

-184.350
(115.659)
-20.658***
(3.938)
Observations 102 13
R
2
0.708 0.844
Adjusted R
2
0.699 0.812
Residual Std. Error
10.145
(df = 98)
1.454
(df = 10)
F Statistic
79.122***
(df = 3; 98)
26.963***
(df = 2; 10)
Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

436

Table F.1b
Summary statistics: Logarithmically transformed regression models of episodes of intra-
Chechen conflict with stepwise selection based on Bayesian information criterion (BIC)
Log-Transformed OLS Models with BIC
1994-2009 Intra-Chechen Conflict Continuum

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

1.588***
(0.315)

Distance from Regional Center (km)

-2.039***
(0.108)

Distance from Border (km)

0.498***
(0.099)
Distance from Primary Route (km)

117.830***
(2.990)
-5.541

(3.225)
Observations 263 11
R2 0.627 0.739
Adjusted R2 0.624 0.710
Residual Std. Error

36.694
(df = 260)
3.186
(df = 9)
F Statistic

218.579***
(df = 2; 260)
25.478***
(df = 1; 9)
Note:
*p<0.1; **p<0.05; ***p<0.01

437

Log-Transformed OLS Models with BIC
2009-2017 Conflict with Caucasus Emirate

Dependent variable:
Conflict Events

(1)
Regional
Centers
Model
(2)
Borders
Model

Distance from National Capital (km)

NA
NA

Distance from Regional Center (km)

-0.601***
(0.040)

Distance from Border (km)

0.872***
(0.121)
Distance from Primary Route (km)

40.239***
(1.461)
-20.921***
(3.973)
Observations 102 13
R
2
0.690 0.824
Adjusted R
2
0.687 0.808
Residual Std. Error
10.336
(df = 100)
1.470
(df = 11)
F Statistic
223.043***
(df = 1; 100)
51.568***
(df = 1; 11)
Note:
*
p<0.1;
**
p<0.05;
***
p<0.01

438

These tests were also performed with R. The OLS and log-transformed models yielded results for
both conflict periods that were more in line with expectations for the national capital; conflict
events seemed to occur farther from it, but the statistical significance was weak. The regional
center was a strong focal point in all models and for both periods.
Distance away from the border appeared to be a strong indicator of conflict, which was
very different from the findings from the case of Kurdish factions. However, the states of Georgia
and Russia are powerful actors and likely contained much of the movement across borders. That
said, there was at least some degree of cross-border traffic in each period of conflict, so it is still
puzzling that Chechen rivals did not try to intercept their foes near the border, even in the second
period when the forces of the Chechen Republic were assisted by the Russian Federation. Terrain
may have something to do with this finding, as discussed in Chapters Eight and Nine.
The primary route appeared to be more important in the second period, as well as more
significant. This supports the argument made in the main body of this work regarding the utility
of data and the recency of conflict. More recent conflicts will likely demonstrate the utility of that
particular hypothesis, while older ones will likely suggest that the primary route is an insignificant
object of contestation.
The next step was to assess autocorrelation with Moran’s I. The results are displayed in
table F.2:

439

Table F.2
Measures of spatial autocorrelation of intra-Chechen conflict events: Moran’s I statistic

I statistic Expectation Variance z-score Pseudo p-
value

1994-2009
Global 0.8897 -0.0037 0.0002 58.195 < 2.2e-16
EBS 0.1791 NA NA NA 0.001

2009-2017
Global 0.8105 -0.0088 0.0097 0.0006 8.856e-14
EBS 0.1733 NA NA NA 0.001

Note. EBS is Empirical Bayesian Smoothing. All statistics are rounded to the nearest ten
thousandth. z-score is equal to one standard deviation. EBS tests were performed with
Monte Carlo simulations under randomization in which n = 1000, and so Expectation,
Variance, and z-score are not yielded.

Moran’s I was calculated using R. The global Moran’s I results demonstrated extreme correlation
with highly significant values. EBS moderated the I statistic value considerably, but the values
remained significant, suggesting that CSR was unlikely to be at work.
Following Moran’s I, a Getis-Ord General G test for clustering was performed. The results
of this test are displayed below in table F.3:

440

Table F.3
Measures of spatial clustering of intra-Chechen conflict events: Getis-Ord General G
1994-2009
High-Low Clustering Report
Observed General G: 0.006135
z-score: 17.656092
p-value: 0.00000 0…
Given the z-score of 17.656091593855162, there is a less than 1% likelihood that this high-clustered
pattern could be the result of random chance.

General G Summary

Observed General G: 0.006135
Expected General G: 0.002851
Variance: 0.000000
z-score: 17.656092
p-value: 0 .0 0 0 0 0 0 …

441

2009-2017
High-Low Clustering Report
Observed General G: 0.008749
z-score: 4.621913
p-value: 0.000004
Given the z-score of 4.6219132988539116, there is a less than 5% likelihood that this high-clustered
pattern could be the result of random chance.

General G Summary

Observed General G: 0.008749
Expected General G: 0.006255
Variance: 0. 000132
z-score: 4.621913
p-value: 0.000004

The Getis-Ord General G test was performed with and exported from ArcGIS Pro. There was less
than a one percent chance of CSR in the first period and less than five percent in the second. And
so, clustering was established. As the clustering was above the 95 percent confidence interval, a
Getis-Ord Gi* test was not performed.
The analytical process was now ready for spatially lagged autoregression modeling. The
results and a discussion of them are found in Chapter Five. The robustness tests for the models are
displayed below in table F.4:

442

Table F.4
Spatial analysis and modeling: Robustness checks of intra-Chechen conflict

Global Moran’s I for regression residuals, n = 1000
1994-2009

Regional Centers Model Borders Model

Standard deviate
p-value
Observed
Expectation
Variance

47.281
< 2.2e-16
0.6186
-0.0099
0.0002

Standard deviate
p-value
Observed
Expectation
Variance

0.019597
0.9844
-0.1825
-0.1855
0.022

2011-2020

Regional Centers Model Borders Model

Standard deviate
p-value
Observed
Expectation
Variance

23.627
< 2.2e-16
0.5816
-0.0252
0.0007

Standard deviate
p-value
Observed
Expectation
Variance

0.6119
0.5406
-0.1085
-0.1622
0.0077

443

Studentized Breusch-Pagan Test

1994-1997

Regional Centers Model Borders Model

BP
p-value

9.4786
0.0236

BP
p-value

6.6787
0.0355

2011-2020

Regional Centers Model Borders Model

BP
p-value

14.936
0.0019

BP
p-value

7.4285
0.0243

The robustness checks were performed with R. The global Moran’s I test for regression residuals
lost its statistical significance. Thus, we must be cautious about the findings regarding the impact
of the primary routes in the borders model. We do not need to discard them entirely, however, as
the studentized Breusch-Pagan results remained significant for those models, albeit somewhat
reduced in magnitude. All tests remained significant for the regional centers models of each period.
Thus, we may be confident about the reliability of those findings. The spatial lag models retained
their utility over non-spatial methods.

444

Appendix G: Preliminary tests of intra-Albanian cooperation
Preliminary tests for cooperation among Albanian factions are contained in this appendix.
Like the other appendices, they are displayed in the order in which they were performed.
Roots of Rivalry (P1)
Let us begin with figure G.1, the processed georeferenced map of Albanian distribution
prior to the imposition of modern IRBs:
Figure G.1
Georeferenced historical distribution of Albanian populations

Note. Die Völker des Donauraumes und der Balkanhalbinsel in 1940.

445

Like the other historical demographic maps, this was processed using ArcGIS Pro.
Out-group Incursion (P3)
As there was no statistical analysis of intra-Albanian conflict, let us examine cooperative
action. This begins with a geographic interpolation of cooperative events using KDE, as shown in
figure G.2:
Figure G.2
Kernel density estimation plot of cooperation between Albanian factions

Note. Brighter colors denote denser cooperation.

This was generated using R. As with the Kurdish case, blue regions denote very sparse cooperative
clusters with increasing density of events as the colors move from darker to brighter. Clearly, the
density is much greater than in the Kurdish case. This is indicative of two things. First, irredentists

446

with a single representative state cooperate much more than divided stateless nations. Second, the
geographic area is smaller than in the Kurdish case, which also supports greater density due to the
shorter distances needed to travel to engage the enemy.
Moving on to the K- and L- function tests, the next goal was to determine optimal distance
bands for further analyses of cooperation. The results of these tests are displayed graphically in
figure G.3:

447

Figure G.3
Assessing spatial dependence and randomness

Note. The envelopes are confidence intervals that indicate spatial randomness.

448

The K- and L- functions were performed with R. Unlike the Kurdish case, cooperative events do
not approach CSR at all using the K-function. They do begin approaching CSR after about 25 km
when using the L-function. They do not cross it, however. The optimal range appears to lie between
eight and 25 km.
The final method to exclude CSR was the ANN analysis. The results of the ANN analysis
are displayed in figure G.4:
Figure G.4
Comparing the observed model to a random model

Note. The blue line is the average distance between neighboring events.

449

This analysis was also performed with R. The results indicate that the observed values were
vastly different from the simulated values, thus suggesting strongly that CSR is not at work with
these data. The associated pseudo p-value was statistically significant as well. The null
hypothesis of CSR was rejected.

Abstract (if available)

Abstract Why do ethnonationalist factions that ostensibly share goals and foes fight each other despite what we know about the pacific effects of coethnicity? When do they cooperate against out-group threats and when will they not? I argue that a four-fold categorical approach which distinguishes between varieties of organized intraethnic conflict will contribute to a better understanding of this underexplored phenomenon. Categorization is based on whether factions claiming to represent ethnic groups are divided or constrained by borders, and whether the group forms the core community of a nation-state or not.
This work theorizes that factions claiming to represent a stateless nation that is divided by borders will experience a form of organized intraethnic conflict that is distinct from conflicts that affect other recognized forms. These include separatists in a single state, state-backed irredentists divided by borders, and two states of co-ethnics divided by borders—divided societies—such as East and West Germany.
Organized intraethnic conflict is rare, but it may be predicted and modeled depending upon its categorization. The origins of conflict, the objects of contestation, and the conditions for cooperation vary by type. This work’s primary contribution is the recognition of conflict involving divided stateless nations (DSNs).
This work tests the proposal that DSN factional conflict is distinct from other forms using an analytically eclectic empirical strategy with mixed methods. It employs linear modeling and novel spatiotemporal analyses combined with process tracing and qualitative research.
Data is drawn from the Uppsala Conflict Data Program, NGO, IGO, and state sources. Qualitative sources include factional press releases and media. Interviews with fighters and members of the ethnic communities being studied are also incorporated.
Tests are applied to archetypical, extreme cases as a basis for baseline theorizing. The DSNs are represented by Kurdish factions. Chechen factions stand in for constrained separatists. Albanian factions are the irredentists. North and South Korea provide the divided society.
I find that the drivers of conflict for factions representing DSNs are more like divided societies and unlike infighting among separatists and irredentists. DSN factions are unique with regards to what they fight over. DSN factions will cooperate against outsiders if large numbers of unaffiliated coethnics are killed in institutionally competitive territory. Surprisingly, non-combatant status is only salient when accounting for factional affiliation. Rival separatists and divided societies will not cooperate even under this condition, while irredentists will cooperate regardless of who is killed and where.
Earlier generations of the literature have suggested that coethnicity is either a pacific force or a source for division and exploitation by state actors. A key implication for the literature is that precise effects of coethnicity can be predicted by understanding a group’s degree of territorial division and statehood status.
This work has an important implication for policy as well. The prevailing view is to consider organized intraethnic conflict as a singular phenomenon. Many pressing humanitarian crises and interethnic conflicts develop from this phenomenon, yet without making necessary distinctions these crises escalate. Proper identification could make for more positive conflict resolution outcomes.

Linked assets

University of Southern California Dissertations and Theses