University of Southern California Dissertations and Theses

When bullets are not enough: determinants of chemical weapons usage and abstention

(USC Thesis Other)

When bullets are not enough: determinants of chemical weapons usage and abstention

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content WHEN BULLETS ARE NOT ENOUGH:
DETERMINANTS OF CHEMICAL WEAPONS USAGE
AND ABSTENTION

By

Kymberly G. MacNeal

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

December 2016

Copyright 2016 Kymberly G. MacNeal
ii

Dedication

To Bruce and Michele MacNeal, my parents and ardent supporters. I do
not tell you enough how much I love you.

To Lena Kurtz, great-aunt and even greater cheerleader. You are deeply
missed.

iii

Acknowledgements

I would like to thank the three wonderful members of my committee –
Patrick James, Robert English, and Jane Figueiredo – for their support of this
dissertation. In particular, I would like to thank Professor James for his
encouragement and for doing everything in his power to ensure this project’s
timely completion. I would like to thank my fellow members of the Political
Science and International Relations Ph.D. Program at the University of Southern
California – may committee decisions be ever in your favor and may you embark
on the careers of your dreams. In particular, I would like to thank Nicolas de
Zamaróczy, for his willingness to destroy papers in progress and give sage advice,
and Matthew Mendez, for his friendship and encouragement through the rough
patches. My days will be that much dimmer without you both by my side. I would
like to thank the POIR staff, particularly Cathy Ballard and Veridiana Chavarin.
As unofficial ‘den-mothers,” your advice and paperwork reminders have kept us
on track and loving every minute of it. I would like to thank my family for
indulging my thirst for knowledge and putting up with random factoids and
decidedly odd conversations at the dinner table.

iv

Table of Contents

Dedication ............................................................................................................... ii
Acknowledgements ................................................................................................ iii
List of Illustrations ................................................................................................. vi
Chapter 1- Introduction ........................................................................................... 1
Theoretical Significance ...................................................................................... 6
Practical Significance .......................................................................................... 8
Structure .............................................................................................................. 9
Chapter 2- What’s the Fuss About a Label? ......................................................... 12
Ban on the Use of Chemical Weapons- Taboo? ............................................... 14
Why the Word Taboo is Important ................................................................... 23
What It Should Be Called Instead ..................................................................... 47
Conclusion ......................................................................................................... 53
Chapter 3- How Do We Know What We Know? ................................................. 55
What Is QCA and How Is It Different from Normal Statistics? ....................... 56
The Problem of Limited Diversity .................................................................... 64
Supersets: A Methodological Change for Unknown Causes ............................ 67
Chemical Weapons Usage: Who Are We Talking About? ............................... 71
How Can We Tell the Difference Between Users and Abstainers? .................. 78
Assessing the Driving Cases ............................................................................. 93
Conclusion ......................................................................................................... 94
Chapter 4- Expectations and Surprises ................................................................. 96
Is The Use of Chemical Weapons Consistent from 1918 To 2014? ................. 98
Why Do Countries Abstain from Using Chemical Weapons? ........................ 116
What Cases Drive the Use (And Non-Use) of Chemical Weapons? .............. 132
Are Supersets Useful for Identifying Potential Sufficient Conditions? .......... 141
What Have We Learned? ................................................................................ 146
Conclusion ....................................................................................................... 151
v

Chapter 5- What Have We Learned & Where Do We Go From Here? ............. 154
Future Research ............................................................................................... 164
Chemical Weapons: A Historical Evil and Continuing Problem? .................. 169
References ........................................................................................................... 173
Appendix A- List of Figures ............................................................................... 202
Appendix B- List of Tables................................................................................. 215

vi

List of Illustrations

Figures
Chapter 3
Figure 1- Definition of a superset 203
Figure 2- Superset S with subsets A, B, C 204
Figure 3- Effect of superset creation on explanatory condition
scores
205
Figure 4- Number of times a state was accused of using chemical
weapons
206
Figure 5- Chemical weapons use by region 207
Figure 6- Chemical weapons use by and against the state 208
Figure 7- Chemical weapons usage by location in or out of the state 209

Chapter 4
Figure 8- Sufficient conditions for the outcome ‘chemical weapons
use’
210
Figure 9- Sufficient conditions for the outcome ‘non-lethal
weapons first
211
Figure 10- Necessary conditions for the outcome ‘chemical
weapons use’ in 1918 Dataset
212
Figure 11- Sufficient conditions for the outcome ‘non-lethal
weapons first’ in 1918 dataset
213
Figure 12- Sufficient conditions for the outcome ‘non-use of non-
lethal weapons first’
214

vii

Tables
Chapter 3
Table 1- List of instances of chemical weapons use and their
acceptance status
216
Table 2- Average chemical weapons use across cases 219
Table 3- Statistical testing for use of chemical weapons by and
against the state
220
Table 4- Military vulnerability set membership scores 221
Table 5- Political vulnerability set membership scores 225
Table 6- Conflict location set membership scores 230
Table 7- Outcome condition set membership scores 234

Chapter 4
Table 8- Summary table of all fuzzy set scores for states that have
been accused of using chemical weapons
239
Table 9- Analysis of necessary conditions for the outcome
‘chemical weapons use’
242
Table 10- Analysis of necessary conditions for the outcome ‘non-
lethal weapons first’
243
Table 11- Analysis of sufficient conditions for the outcome
‘chemical weapons use’
244
Table 12- Analysis of sufficiency conditions for the outcome ‘non-
lethal weapons first’
245
Table 13- Analysis of necessary conditions for the outcome ‘use of
chemical weapons” in the 1918-1945 Dataset
246
viii

Table 14- Analysis of necessary conditions for the outcome ‘non-
lethal first” in the 1918-1945 Dataset
247
Table 15- Analysis of sufficient conditions for the outcome
‘chemical weapons use’ in the 1918-1945 dataset
248
Table 16- Analysis of sufficient conditions for the outcome ‘non-
lethal weapons first’ in the 1918-1945 dataset
249
Table 17- Analysis of necessary conditions for the outcome ‘use of
chemical weapons” in the 1945-2014 Dataset
250
Table 18- Analysis of necessary conditions for the outcome ‘non-
lethal weapons first” in the 1945-2014 Dataset
251
Table 19- Analysis of sufficient conditions for the outcome
‘chemical weapons use’ in the 1945-2014 dataset
252
Table 20- Analysis of sufficient conditions for the outcome ‘non-
lethal weapons first’ in the 1945-2014 dataset
253
Table 21- Analysis of necessary conditions for the outcome ‘non-
use of chemical weapons
254
Table 22- Analysis of necessary conditions for the outcome ‘non-
use of non-lethal weapons first’
255
Table 23- Analysis of sufficient conditions for the outcome ‘non-
use of chemical weapons’
256
Table 24- Analysis of sufficient conditions for the outcome ‘non-
use of non-lethal weapons first’
257
Table 25- Analysis of necessary conditions for the outcome ‘non-
use of chemical weapons’ in the 1918-1945 dataset
258
Table 26- Analysis of necessary conditions for the outcome ‘non-
use of non-lethal weapons first’ in the 1918-1945 dataset
259
Table 27- Analysis of sufficient conditions for the outcome ‘non-
use of chemical weapons’ in the 1918-1945 dataset
260
ix

Table 28- Analysis of sufficient conditions for the outcome ‘non-
use of non-lethal weapons first’ in the 1918-1945 dataset
261
Table 29- Analysis of necessary conditions for the outcome ‘non-
use of chemical weapons’ in the 1945-2014 dataset
262
Table 30- Analysis of necessary conditions for the outcome ‘non-
use of non-lethal weapons first' in the 1945-2014 dataset
263
Table 31- Analysis of sufficient conditions for the outcome ‘non-
use of chemical weapons’ in the 1945-2014 dataset
264
Table 32- Analysis of sufficient conditions for the outcome ‘non-
use of non-lethal weapons first’ in the 1945-2014 dataset
265
Table 33- Analysis of attack ratios for the outcome ‘chemical
weapons use’ for 1918-1945 dataset
266
Table 34- Analysis of attack ratios for the outcome ‘chemical
weapons use’ for 1945-2014 dataset
267
Table 35- Analysis of attack ratios for the outcome ‘non-lethal
chemical weapons first’ for 1918-1945 dataset
268
Table 36- Analysis of attack ratios for the outcome ‘non-lethal
chemical weapons first’ for 1945-2014 dataset
269
Table 37- Analysis of attack ratios for the outcome ‘non-use of
chemical weapons’ for 1918-1945 dataset
270
Table 38- Analysis of attack ratios for the outcome ‘non-use of
chemical weapons’ for 1945-2014 dataset
271
Table 39- Analysis of attack ratios for the outcome ‘non-use of
non-lethal weapons first’ for 1918-1945 dataset
272
Table 40- Analysis of attack ratios for the outcome ‘non-use of
non-lethal weapons first’ for 1945-2014 dataset
273
Table 41- Analysis of the predictive power of supersets in the
1918-1945 and 1945-2014 datasets
274
x

Table 42- Summary of fsQCA analysis for the use and nonuse of
chemical weapons and their escalation from 1918 to
2014
275

1

Chapter 1
Introduction

2

“Gas! GAS! Quick, boys! – An ecstasy of fumbling,
Fitting the clumsy helmets just in time,
But someone still was yelling out and stumbling
And floundering like a man in fire or lime. –
Dim through the misty panes and thick green light
As under a green sea, I saw him drowning.

In all my dreams before my helpless sight
He plunges at me, guttering, choking, drowning.

If in some smothering dreams, you too could pace
Behind the wagon that we flung him in,
And watch the white eyes writhing in his face
His hanging face, like a devil’s sick of sin;
If you could hear, at every jolt, the blood
Come gargling from the froth-corrupted lungs,
Obscene as cancer, bitter as the cud
Of vile, incurable sores on innocent tongues, -
My friend, you would not tell with such high zest
To children ardent for some desperate glory,
The old Lie: Dulce et decorum est
Pro patria mori.”
-Wilfred Owen, “Dulce et decorum est”

This project begins with a failure. It was not a necessary failure; many
people worked tirelessly to prevent these events from occurring. Hampered by a
general unwillingness to intervene and a mistaken belief in the general adherence
to rules decided a century ago, the world watched as innocents became the newest

3

victims of an old weapon. For the first time in a generation, chemical weapons
were unleashed upon the earth.
A civil war broke out in 2011 in Syria as a consequence of the wave of
changes unleashed by the Arab Spring in 2010. In the fall of 2012, Western
intelligence agencies warned that the Syrian regime was moving its chemical
assets, potentially indicating that they were preparing to use those weapons (Eric
Schmitt and David E. Sanger. 2013). Several countries warned Assad that
chemical weapons use was unacceptable, and could result in unidentified
consequences (Anonymous 2012; Peter Baker and Michael R. Gordon. 2012;
Peter Baker. 2012; Eric Schmitt and David E. Sanger. 2013). On March 19, 2013
the first attack using chemical weapons in Syria occurred, killing approximately
24 people and injuring another 124 (Baum 2015, 137). Attacks occurred
sporadically through the end of August, with the largest occurring on August 21,
killing 1,429 people and injuring approximately 2,170 others (United Nations
Mission to Investigate Allegations of the Use of Chemical Weapons in the Syrian
Arab Republic 2013a; United Nations Mission to Investigate Allegations of the
Use of Chemical Weapons in the Syrian Arab Republic 2013b; United States of
America 2013). The US blamed this attack on the Assad regime, citing
intelligence that placed personnel from Assad’s chemical weapons branch in the
area and chemical filled munitions originating from government-controlled areas
of Damascus (Day August 27, 2013; United States of America 2013, 2-3).

4

China and Russia blocked US attempts in the UN Security Council to
authorize the use of force in response to this violation of international law.
Tensions between Russia and the US grew so strained that a misstep could have
resulted in a physical confrontation over Syria. Due to a combination of
diplomatic skill and serendipity, a deal was reached to allow the Syrian
government to surrender their chemical arsenal in exchange for calling off a
potential US intervention in the civil war.
The removal of chemical weapons from Syria began in December 2013,
but did not stop the use of these weapons. Beginning in April 2014, crude chlorine
bombs were used against civilian populations within Syria. While the US declined
to place blame, circumstantial evidence strongly suggests that the Assad regime is
responsible (Organization for the Prohibition of Chemical Weapons. Office of the
Director-General December 18, 2014; Hannah Lucinda Smith. 2015). Chemical
weapons use continued until March 6, 2015, when the UN Security Council
threatened to authorize the use of force if these attacks did not cease (Anonymous
2015).
The events in Syria from 2013 to 2015 highlight a gap in the literature in
international relations. States have used chemical weapons in the modern era,
often blatantly contravening their obligations under different treaties. To date,
there has been no systematic study of why these states would choose to use

5

chemical weapons. Beyond case studies of chemical weapons acquisition, there
has not been an attempt to understand why states use their chemical arsenals.
This lack of scholarship can be partially attributed to a truism in
international relations. It is a well-known fact that states do not use chemical
weapons. Known as the “chemical weapons taboo,” the ban on chemical weapons
use has been institutionalized since the early 1900s, and has been augmented
since then by additional treaties despite the flagrant use of these weapons during
World War I.0F
1
In the light of the events in Syria, this truism deserves to be put
under scrutiny. If it is true that states do not use their chemical arsenals, then why
did Assad choose to do so? This is doubly true after 2013, when Syria officially
acceded to the Chemical Weapons Convention (CWC).
This dissertation seeks to answer those questions. Firstly, this project will
attempt to identify the conditions under which states will use their chemical
weapons arsenals. Are there consistent conditions under which states routinely
use chemical weapons, or are their decisions more subject the circumstantial
pressures? Secondly, this project will strive to ascertain the conditions under
which states will refrain from using their chemical weapons. There are several
countries with chemical weapons that have chosen not to use those arsenals for

1
These international agreements include the Strasbourg Agreement (1675), the Hague
Conventions (1899 and 1907), Geneva Protocol (1925), and the Chemical Weapons Convention
(1993).

6

decades. Are the conditions under which those decisions made generalizable?
Thirdly, there have been a significant number of states that chose to begin their
chemical weapons use with non-lethal agents and escalated to lethal agents. Under
what conditions did those states choose to escalate? Are they different from the
decision to use chemical weapons in general? What are the conditions under
which states choose not to escalate?

Theoretical Significance
Answering the above questions has significant theoretical importance. As
mentioned above, the literature believes that the ban on the use of chemical
weapons has reached the level of taboo behavior in the international system (Price
and Tannenwald 1996; Price 1997). This implies that instances of this behavior
should be a rare event- a significant aberration in the conduct of statecraft that is
both reviled and institutionally disallowed. Because of the abhorrence of chemical
weapons, the literature believes that there is a general reflexive aversion to the use
of these weapons by decision makers, resulting in their infrequent usage
(Jefferson 2009; Price and Tannenwald 1996; Price 1997).
Empirically, this perception of chemical weapons use can be viewed as
inaccurate. States have been accused of using chemical weapons in over 40
separate instances since 1918. While it is true that the use of chemical weapons

7

can be labelled a “rare” event, it all depends on the comparison group. When
compared to the use of guns or tanks in warfare, chemical weapons use is clearly
a rare event. When compared to other weapons of mass destruction (WMD),
however, their use is far more common. Nuclear weapons were only ever used in
one conflict and biological weapons were only ever covertly used in a few. This
dissertation will reexamine the literature on the chemical weapons taboo and
explain why there is such gap between theoretical expectations and empirical
data.
This dissertation is theoretically significant for a second reason. The
method this project will use to identify why states use and refrain from using
chemical weapons is fuzzy set Qualitative Comparative Analysis (fs/QCA), a
variant of Qualitative Comparative Analysis (QCA). QCA was developed as a
“middle ground” between traditional quantitative methods and small-n case
studies in the social sciences. This author proposes that the use of artificially
constructed supersets can serve as a screening tool in QCA to identify which
conditions (out of a large set) will later prove significant. A proof of concept test
for the use of supersets will be conducted on the international use of chemical
weapons dataset to evaluate whether or not this innovative method could prove
useful for international relations.

8

Practical Significance
Understanding why states use chemical weapons is not simply a matter of
academic interest; it has clear implications for the conduct of international affairs.
Firstly, identifying the conditions under which states will use their chemical
weapons creates a window of opportunity for decision makers to prevent the
future use of chemical weapons. If decision makers are interested in forestalling
the use of these weapons, as several indicated they were in the Syrian incident,
they need to know ahead of time what could potentially signal their intended use.
By understanding the conditions under which chemical weapons were used in the
past, recognition of these conditions recurring can serve as that early warning.
Once the warning signs are recognized, the international community has an
opportunity to alter those conditions to prevent regimes from using their weapons.
This could potentially save thousands of lives, most of them civilian.
The second practical reason why understanding chemical weapons use
would prove useful deals with the efficient use of resources. The primary
monitoring system for states about the movements of others in the international
system is their intelligence community. Being able to identify states at high risk
for using chemical weapons (experiencing conditions which have led to past use)
could allow those agencies to scale up their activities regarding that state.
Conversely, identifying countries at low risk (experiencing conditions which did
not lead to chemical weapons use in the past) would allow those agencies to divert

9

their resources to areas of greater need without compromising their ability to warn
of chemical weapons use. By understanding why states do and do not use
chemical weapons, intelligence communities can more efficiently use their limited
resources at their disposal.

Structure
In order to answer the questions of why chemical weapons are and are not
employed, this dissertation will be structured as follows. Chapter 2 will discuss
the gap between the literature and empirical evidence. The definition of what a
chemical weapon is will be given, followed by the current understanding of the
chemical weapons taboo. An alternate label for the disinclination of states to use
chemical weapons will be proposed, taking into account international reactions to
use and their frequency.
Chapter 3 will discuss how the international use of chemical weapons will
be systematically studied. First, a description of fs/QCA methodology and
critiques will be put forward. Next, a method from epidemiology to identify the
driving cases for chemical weapons use and abstention will be discussed. Lastly,
the logic behind the creation of supersets will be laid out, explaining how these
artificial sets could be used to screen for potential sufficient conditions.

10

The results of the fs/QCA study will be presented in Chapter 4. The results
of the fs/QCA analyses for the use and non-use of chemical weapons will be
presented, as well as analyses for chemical weapons escalation and non-
escalation. Fighting in a conflict in the Global South against a stronger opponent
and facing a threat to regime survival were determined to be necessary conditions
for chemical weapons use in the interwar period. Conditions associated with
chemical weapons use in the post WWII period were not identified. Non-use in
both periods required being more powerful than one’s opponent while possessing
a security commitment. Attack ratio analyses will then be presented to explain
why fs/QCA had difficulty distinguishing between those states that escalated
chemical weapons use and those that did not. Lastly, the proof of concept for
supersets will be tendered, highlighting the strengths and weaknesses of this new
method.
Chapter 5 will conclude this project. The main arguments and findings
from each chapter will first be presented. Then, topics of future research
stemming from this dissertation will then be discussed. The topic of chemical
weapons use is a complex one that is not well understood, necessitating additional
study in order to assist policy makers in making their decisions. The use of
supersets as a screening tool needs to be further studied, as it could prove to be a
method highly suited to pilot studies. Lastly, this chapter will present the author’s
reflection on the current international attitude towards chemical weapons use and

11

the likelihood that these weapons will be used again in international and domestic
conflict.

12

Chapter 2
What’s the Fuss About a Label?

13

A systematic evaluation of why states choose to use their chemical
weapons is absent from the current international relations literature. This lack of
scholarship is at least partially due to the concept of the chemical weapons taboo,
i.e. the belief that states refrain from using their chemical weapons. This chapter
aims to assess whether the disinclination by states to use chemical weapons can
rightly be labelled as such. To accomplish this, this chapter is laid out in the
following manner. First, the state of the current literature on the chemical
weapons taboo will be outlined, paying particular attention to the different
proposed sources of the taboo. Next, an exploration of the understanding of the
taboo phenomena in the literature will be undertaken. Insights from anthropology
and sociology will be incorporated to more clearly identify what a taboo is at its
core. Challenges to taboo as an appropriate label for the disinclination to use
chemical weapons will be presented, and an assessment will be conducted as to
how well this label describes the chemical weapons non-use phenomenon.
Finally, this chapter will propose that a “prohibitive norm” would be a better
description of states’ non-use of chemical weapons, relaxing some of the
assumptions inherent in the taboo label. The implications of this relabeling for
perceptions of individual and societal risk will also be discussed.

14

Ban on the Use of Chemical Weapons- Taboo?
Chemical warfare is best described as the use of chemicals to reduce am
adversary’s military effectiveness or to terrorize a civilian population. Instead of
relying on bullets and explosives (kinetic weapons) to harm military personnel
and matériel, chemical weapons utilize the chemical properties of the agents to
poison living things and render objects and locations unsafe to interact with.
Chemical warfare agents can affect soldiers directly through contact, inhalation,
or ingestion, producing confusion, incapacitation, or death. Agents of chemical
warfare can also be used to destroy a country’s food supplies, economic and
manufacturing infrastructure, or livestock (Thomas and Thomas 1970, 3). The
chemical weapons taboo is simply the normative prohibition against the use of
this class of weapons.1F
2

The decision over whether to allow the use of novel weapons in the
practice of warfare has always been hotly debated. Issues of morality, technical
skill, and overall fairness have guided the arguments over tools now taken for
granted, such as the crossbow, gun, and airplane (Price 1997). The majority of the
time, once-controversial new weapons come to be viewed as just another tool in a
country’s arsenal after repeated use and the passage of time. Yet, legal bans on the

2
Norms are understood to be standards of appropriate behavior for actors with a given identity.
For more discussion, see (Finnemore and Sikkink 1998, 887-917; Jepperson, Wendt, and
Katzenstein 1996, 54; North 1990).

15

use of lethal chemical weapons have existed in international law for over 100
years.2F
3
Why did this particular type of weapon acquire a taboo and legal ban on
its usage when other weapons did not? Scholars debate why the perceived
chemical weapons taboo emerged. Three sets of hypotheses have been proposed:
the essentialist argument, the military calculus argument, and the constructivist
argument.

The Essentialist Argument In this context, essentialism claims that there is an
evolutionarily derived biological mechanism responsible for the human aversion
to the use of poison and poison weapons. Humans do not use poison because it
has been encoded into our very DNA. This aversion takes the form of either a
physiological adaptation (instinctively loathing poisonous objects), or a mental
link connecting poison with the supernatural world. Cole is a proponent of the
second adaptation. He finds that the use of chemicals for defense by animals is
replicated by some pre-farming societies (Cole 1998, 122-124). In humans, Cole
claims that this disinclination to use poison as a regular tool is due to the mental

3
The Hague Convention of 1899 banned the use of ‘projectiles the object of which is the diffusion
of asphyxiating or deleterious gases’ between contracting powers. It shall be noted that in this
paper when I refer to a taboo, I am referencing an unwritten but real normative disinclination to
perform a certain type of behavior. When I refer to a ban, I am referring to a formal treaty
outlawing certain types of activities. Taboos and bans are usually mutually reinforcing, but not
always.

16

link between three factors: poison, magic, and medicine. Medicine is an essential
tool for survival, and can often be used to harm at high doses. Misuse of the
materials that heal in order to kill may offend deep-rooted sensibilities,
particularly because healing is often done with the aid of the occult. The use of
healing substances to harm could risk offending the spirits (Cole 1998, 125;
Mandelbaum 1981, 38).
A proponent of the first adaptation, Jefferson suggests that the chemical
weapons taboo should be conceptually constructed in terms of a visceral reaction
through the disgust response which becomes socially formalized as a social
prohibition (Jefferson 2009, 209). This visceral reaction to the threat of poison has
roots in an innate psychological response to the possibility of contamination. The
potential of pollution over time has been conceived of as less acceptable than the
‘immediate violent effects upon the body by conventional violent trauma’
(Jefferson 2009, 209). She adds that even though taboos are based in part on
visceral reactions, they are also ‘culturally embedded phenomena dependent on
contextual factors for their existence and enforcement’ (Jefferson 2009, 209). An
example of the marriage of biological response and state policy can be found in
US history. In the early part of the 20
th
century, conditions in the meatpacking
industry were astonishingly horrid. Not only were workers exposed to hazardous
working conditions and drastically overworked, no modern sanitation practices
were employed. It was common for rodents, their excrement, and other

17

contaminants to be ground into the meat and sold to customers for human
consumption. This practice went largely unreported until Upton Sinclair published
The Jungle in 1906, in which he described the industry in lurid detail. The public
was horrified that it could be eating vermin in their breakfasts and demanded
institutional change. Later that year, Congress passed both the Federal Meat
Inspection Act and the Pure Food and Drug Act of 1906 and established the
Bureau of Chemistry (later to be known as the Food and Drug Administration).
These laws prohibited contaminated and unsanitary meat products from being
sold and forbid the interstate and international trafficking of adulterated food and
drug products (Anonymous 1906, 768-772; Anonymous 1906). These laws and
others prevent the contamination of the US food supply, and were initiated by the
visceral response of a public who viewed the consumption of vermin-tainted food
utterly unacceptable. The continued cultural practice of avoiding contaminated
food ensures the relevance, and strict adherence to, of these laws.
The link between poison and the taboo against chemical weapons occurred
on the battlefields of the First World War. The norm against poison rested on an
entrenched custom and visceral response, while the norm against the use of
chemical weapons was prompted by a desire to limit new war technologies
(Jefferson 2009, 210). The longstanding unease with the military use of poison in
the European tradition was in part derived from customary law stemming from the
just war tradition as well as a more basic biological drive to avoid contamination

18

or being poisoned. Prior to the First World War, European states signed the
Hague Convention of 1899, specifically disallowing the use of asphyxiating gases
in warfare. At the time, many of the modern chemical agents were mostly
experimental and had not been proven to be militarily useful or were not yet
synthesized.
During WWI, the lines between the prohibition against the use of poison
in warfare and the use of chemical weapons blurred. In the emotional and
psychological reactions to the use of chemical weapons on the Western Front, the
mental distinction between the taboos against poison and chemical weapons
melded, eliciting the same disgust response based on their shared contaminating
properties (Jefferson 2009, 210). This mental linking between poison and the new
chemical weaponry is evident in the works of British poet Wilfred Owen, who
described the effects of the use of chemical agents as “If you could hear, at every
jolt, the blood come gargling from the froth-corrupted lungs, obscene as cancer,
bitter as the cud of vile, incurable sores on innocent tongues…”(Owen 1963).
Here, the results of the new weapons were described by a clearly horrified author
who equated it with disease and corruption. This new conceptual link between
poison and chemical weapons was further entrenched by the legal ban in the
Geneva Protocol (1925), which helped to reaffirm the taboo against chemical
weapons (Jefferson 2009, 210).

19

Moon argues that the taboo against the use of poison, and by extension
chemical weapons, can be found deeply entrenched in Western literature. In
various canonical works, a norm against the use of poison as a practice is widely
prohibited in military ethics (Van Courtland Moon 2008, 55). Additionally, the
revulsion against the use of poison is part of the ‘deep cultural taboos which
define what we are as a civilization, what we consider most threatening to our
values and to the ways in which we define ourselves’ (Van Courtland Moon 2008,
55). Through his investigation of the imagery patterns in Western literature, he
finds that the pain caused by poison is equated with treachery, destructive human
emotions, pollution, and environmental degradation (Van Courtland Moon 2008,
55-56).

The Military Calculus Argument The second line of argument regarding chemical
weapons non-use involves a variant of deterrence theory. Moon concludes that it
was strategic deterrence at a high level that prevented chemical weapons usage in
Europe throughout WWII.3F
4
Firstly, both top military officers and their civilian
bosses, with memories of WWI fresh in their minds, were aware that initiation of
chemical warfare was incredibly dangerous (Van Courtland Moon 1984, 31).

4
Mueller agrees with this assessment, although he mentions that there would likely be some short-
term gain from using chemical weapons in combat (Mueller 1989).

20

Major cities in Europe were held as mutual hostages of the opposing side’s air
force. Secondly, at the start of the war in 1939, the Allies announced that they
would abide by the Geneva Protocol, while reserving the right to retaliate, clearly
signaling to the Axis powers that there was no advantage to be gained by
preemption and that any use of these weapons on the battlefield would spark all-
out chemical warfare.
While strategic deterrence may have served as a systemic restraint to
chemical weapons usage, the most salient hindrance was tactical (Van Courtland
Moon 1984, 3-35). During the first three years of the war, the Axis powers were
bent on a quick victory, and chemical weapons would have hampered military
operations.4F
5
The Allies chose not to use chemical weapons given the vulnerability
of their civilian populations, even though it could have slowed the Axis advance
into Western Europe and the southwest Pacific (Van Courtland Moon 1984, 31-
32). To deter instances of chemical warfare in Europe and the Pacific, the Allies
adopted the doctrine of massive retaliation in 1942, linking all theaters of the war
and raising the stakes of chemical warfare considerably.5F
6
Exaggerated estimates
of Axis chemical weapons holdings reinforced deterrence, as each side assumed

5
In early Blitzkrieg campaigns, the use of chemical weapons would have complicated overall
strategy. In the desert campaigns, logistics and atmospheric conditions would have negated their
usefulness (Van Courtland Moon 1984, 31).
6
Formal Allied chemical weapons policy was formulated in 1942 in response to reports of
chemical weapons use in China and Russia. Despite the fact that these incidents were clearly
contrary to international law, there was little response from either Western powers or the League
of Nations (Van Courtland Moon 1984, 32).

21

that the other would retaliate, resulting in massive military and civilian casualties.
Without the belief of policymakers that chemical warfare would deliver
significant advantages, the risks to all sides seemed too high to justify initiating
chemical warfare (Van Courtland Moon 1984, 33).

The Constructivist Argument The third set of researchers are those who advocate
a constructivist account of the chemical weapons taboo and are skeptical of the
ability of either the military calculus or the essentialist accounts to fully explain
why people attached a stigma to chemical weapons use. The constructivist
argument posits that the chemical weapons taboo is an instance of a very special
and contingent social construction about what it means for a state to be civilized.
Since legal, technical, and deterrent explanations are insufficient to
explain chemical weapons non-use, constructivist arguments focus on how
discursive practices shaped the conception of chemical weapons as ‘forbidden’
weapons. Modern conceptions of acceptable practices in war are a continuation of
the debate from the Middle Ages in which certain features of poison were given a
special meaning that stressed the lack of defense and its easy availability as a
weapon of the weak (Price 1997, 26). Warfare has served as a contest of strength
between two opposing sovereigns, with honor demanding straight-forward
confrontation. Poison serves as a force-equalizer capable of derailing the

22

masculine business of war, and was therefore delegitimized as a military tool and
restricted to use in the political realm.6F
7

The discursive practices of diplomats also served to stigmatize chemical
weapons (Price and Tannenwald 1996, 119-120). Price and Tannenwald assert
that the chemical weapons taboo is a socio-political construction resulting from its
institutionalization in international law beginning with the Hague Conference of
1899. As a case of a genuine moral rejection of a means of warfare, it arose at a
particular juncture in history when people questioned the unrestricted use of
technology in warfare amongst ‘civilized’ states (Price 1997, 164). The Hague
Declaration (1899) was the result of an era that wanted to limit warfare through
international law, meaning that this agreement was an instance of mutual
anticipatory hand-tying and self-denial among the major powers (Price 1997,
167). The effect of the treaty was amplified by the chemical and gas warfare
lobbies during the interwar period as they created mass hysteria disproportionate
to the actual chemical weapons threat (Price and Tannenwald 1996, 126-129;
Price 1997, 167). These lobbyists greatly exaggerated new chemical weapons
developments in order to secure tariffs on chemicals in order to ensure the
survival of chemical warfare departments and constructed a picture of chemical
weapons use that was far out of proportion to the actual danger they posed at the

7
It was also considered a weak and womanly way to achieve a goal.

23

time (Brown 1968; Price 1995, 73-103). In the interwar period, chemical weapons
were the weapons most susceptible to counter measures and posed less of a threat
than other weapons systems (Price and Tannenwald 1996). The force of the
“impending chemical apocalypse” hysteria legitimized the further
institutionalization of the chemical weapons “taboo” in the Geneva Protocol
(1925).7F
8

Why the Word Taboo is Important
Words in any language can be extremely powerful. They have the power
to make or break reputations, start wars, or end intractable conflicts. Words are
used to communicate ideas and concepts from person to person, and can condition
individual expectations about other’s future behavior. Because words can play
such an important role in individual action, it behooves the individual to pick their
words carefully when confronted with situations that have significant
consequences. Slight changes in inflection and collocation can drastically change

8
Critics of the chemical lobby argued that chemical weapons were in fact at least as humane as
conventional weaponry. One US Senator argued that a ban on the use of chemical weapons would
prevent the US “from using gas against the next savage race with which we find ourselves in war,
and would compel us to blow them up, or stab them with bayonets, or riddle them and sprinkle
them with shrapnel, or puncture them with machine-gun bullets, instead of blinding them for an
hour or so until we could disarm them” (Anonymous 1927, 150). This idea that gas was a humane
alternative to kinetic weapons was eventually eclipsed by those who saw the horrors they inflicted,
becoming the dominant theme in chemical weapons discourse. Those unconvinced that chemical
weapons are a less humane option and who understood its tactical advantages they can bestow are
still be found today, and have historically found a home within the military.

24

meaning (i.e. I’m fine. vs. I’m fine). Words used to describe a particular gender
could be extremely insulting if misused. Therefore, an evaluation of the term
“taboo” with regards to the non-use of chemical weapons needs to be undertaken.
Does this term adequately represent what is meant by its users? Is the definition of
taboo congruent with observed behavior? Are there other terms that better
describe the empirical record?

What is a Taboo? In her work on the nuclear taboo, Tannenwald defines and
outlines the requirements for a “nuclear taboo,” drawing on the anthropological
and sociological literature focusing on the individual level. For her, a nuclear
taboo is the “de facto prohibition against the first use of nuclear weapons. The
taboo is not the behavior (of non-use) itself but rather the normative belief about
the behavior” [author’s emphasis].8F
9
Taboos are unconsciously taken-for-granted
notions about what qualifies as acceptable social behavior and are rarely, if ever,
thought about and assessed for practical utility (Tannenwald 2007, 13). Taboos
are a specific kind of normative prohibition that refers to danger, eliciting awful
or uncertain consequences or sanctions if violated (Radcliffe-Brown 1939, 26).

9
For norm, Tannenwald means a shared expectation about appropriate behavior for a given
identity found in (Jepperson, Wendt, and Katzenstein 1996, 54; Tannenwald 2007, 10)

25

In response to criticisms, Tannenwald contends that taboos contain two
intersubjective and phenomenological components that qualify them as true
taboos and not merely traditions (Tannenwald 2007, 13). Firstly, taboos function
as “bright-line” norms, or a social red-lines which, if crossed, result in a new
social reality (Steiner 1956, 116; Tannenwald 2007, 11). The main difference
between a taboo and a conventional norm is the reaction to violation: non-taboo
norms are judged on the relative degree of the severity of the violation (one bomb
versus 100 bombs). Taboos, are judged on the absoluteness of their violation; the
use of even one bomb changes social reality. Pandora’s Box has been opened and
a return to innocence requires extensive measures be taken (Tannenwald 2007,
11). As the second intersubjective element, Tannenwald asserts that a taboo need
not possess all the objective characteristics of a taboo if it possesses a crucial
component: a taboo is a taboo if people believe it to be so (Tannenwald 2007, 12).
Tannenwald argues that if a decision maker regards a taboo as a taboo (regardless
of whether it actually is or not), this perception will affect their decision-making
and subsequent behavior. A leader’s perception of appropriate behavior in
international society becomes reality, inducing further changes in behavior.9F
10
For
instance, if a leader of a state became convinced that the use of landmines was
forbidden/immoral/illicit, that decision maker would choose to not use landmines,

10
As has been often observed, “if men define situations as real, they are real in their
consequences.” (Thomas and Thomas 1928, 572)

26

even if their use was not illegal and could benefit that state in a conflict. If the
belief in the forbiddenness of landmines were to persist over time, landmines
would become taboo (unthinkable) for that leader. If other leaders were to
develop similar policy inclinations, the fervent belief in the non-use of landmines
could cross the line from belief into a military reality in which no one uses
landmines.10F
11

In addition to Tannenwald’s criteria, taboos also exhibit two important
features that differentiate them from conventional norms. Taboos are both self-
enforcing and self-perpetuating. While the consequences of taboo breakage are
always somewhat indistinct as noted by Radcliffe-Brown, they are always
assumed to result in terrible consequences. “[The consequences of taboo
breakage] may be left to the excited imagination of the taboo-breaker, who
believes as firmly in the sequence of cause and effect (violation followed by
punishment) as does the modern man in the inevitable action of natural laws”
(Webster 1942, 18). Violators will always be punished via natural laws of cause
and effect, rendering human or superhuman mediation unnecessary (Steiner 1956,
22).

11
There is an obvious potential real-world consequence to this argument. If a single decision
maker, or small group of decision makers, were to believe in the reality of a taboo that did not
exist for the rest of the international system, they would set themselves up to be potential victims
of that non-taboo. It is therefore necessary for those who believe in international taboos to
accurately assess the state of taboo support in the system.

27

“Death- certain, sudden, and in terrible form- is not seldom the fate which is
announced to the taboo-breaker. In the midst of Eden grows the tree of the
knowledge of good and evil, and God has forbidden man to eat of its fruit,
saying ‘In the day that thou eatest thereof, thou shalt surely die’” As a matter of
fact, the taboo-breaker does often die, so acute is the fear caused by even an
involuntary transgression” (Webster 1942, 24).

Webster mentions several cases wherein taboo-breakers essentially willed
themselves to death as a result of their taboo violation, so convinced they were in
the reality of cause and effect. While taboos with this level of psychological
strength are seldom seen in Western cultures, taboo-breakage is usually
accompanied by profound psychological effects, including fear, guilt, pain, and
other emotions.11F
12

Taboos fade and die when they are systematically violated, particularly if
there are few observable consequences for the violators. Over time, the
expectation of punishment diminishes as the taboo is routinely infringed upon.

12
These emotions can have enormous psychological power. The deleterious effects of these
emotions, and the consequences of the individual’s attempts to deal with these emotions can result
in a taboo’s punishment. A perfect example is the Lady Macbeth’s descent into madness (“Out,
damn’d spot!”) after committing regicide in Macbeth. Another example of the psychological and
physiological power of taboos is the “finger in fast food” urban myth. In these mostly false viral
stories, a piece of a worker (usually a finger) is accidentally served to a customer, who then
promptly purges themselves of the offending food, suffers great psychological stress, and
immediately sues the company. These stories tap into the Western taboo against the consumption
of human flesh, and the fear that the modern consumer doesn’t actually know what is in their
mass-produced food. Those who accidentally violate this taboo (the customer) would be expected
to exhibit feelings of nausea, revulsion, anger, and a desire for retribution when their violation is
revealed. In at least one incident in which this story actually happened, the one who accidentally
ate human flesh was a Buddhist vegetarian who consequently stopped eating, lost 15 pounds in
less than a week, and required ongoing therapy (Anonymous 2005).

28

Even in the face of substantial social disproval and sanction, taboos can still be
broken down if infringed upon long enough by a significant number of people.
In addition to being self-enforcing, taboos are self-perpetuating. Self-
perpetuation occurs mainly due to the fact that they reside in the unconscious
parts of the human mind and are rarely examined or questioned. Violations of
taboos are not assessed for their relative utility except in situations of extreme
adversity. While the option exists for people to break social taboos (ex: eating
other people for lunch), very few people even entertain the idea that human flesh
could be served as a meal, let alone seriously consider or carry out the thought.
The unthinkingness of taboo perpetuation is augmented when they become
linked to other taboos. The collective force of those prohibitions keeps violation
literally unthinkable.
“A particular taboo, once well established, tends to multiply endlessly…an
object becomes tabooed which for any reason reminds one of something else
tabooed. Thus prohibition is piled upon prohibition, as Ossa on Pelion and
Pelion on Olympus, to anticipate every single possibility of danger in the
perilous maze of a world where all things are potentially dangerous” (Webster
1942, 13)

Taboos are so strong because of a complex web of issue linkage. The collective
weight of social prohibition prevents one from breaking even one of them. When
taboos are obeyed in society, they become even more firmly entrenched,
perpetuating themselves from generation to generation and gaining power as older

29

generations police younger ones. Once the taboo has been socially established,
the conscience or sense of duty compels the individual to accept the prohibition
and refrain from the prohibited action. This applies even when the taboo is placed
upon things that have little overt normative significance (Webster 1942, 371).
The two criteria in addition to Tannenwald’s describe a closed logical
system. After the initial input (ex: no you cannot eat other people and you will be
punished if you do), the two components of self-perpetuation and self-
enforcement work together to reinforce adherence. Continued adherence creates a
sort of social momentum that perpetuates the social rule. While the logic
describing this phenomenon is circular and can be thought of as a type of
“perpetual motion machine,” this only applies as long as no outside forces act
upon the system. Instances of taboo breakage, particularly in cases where there is
little to no punishment disrupt this social momentum (ex: introducing gravity and
friction into the perpetual motion machine). Given enough disruptions, the
circular logic of the taboo could come flying apart. Once this occurs, and a taboo
loses its self-perpetuating and self-enforcement mechanisms, it is then demoted to
the status of a norm or suggestion.

Implications of the Taboo Label From the definition, taboos are self-perpetuating
and self-enforcing normative prohibitions against certain actions or behaviors.

30

However, the use of the word taboo has implications beyond the simple
definition. When the word “taboo” is used to describe aberrant behavior, it
implies very specific things about the occurrence of that behavior that forms a
subtext to the conversation.12F
13
The use of the word taboo provides information
about how often one can expect to encounter the behavior and what to expect if
one does.
The use of the word taboo indicates that one will come across violations of
the social norm against the behavior very infrequently. This is tied into the notion
of self-perpetuation. Since adherence to the taboo is an unconscious behavior, it is
a form of social programming. People adhere to the taboo because we were
instructed by others in society (either consciously or unconsciously) to view
adherence as correct behavior. Violation of taboo, therefore, can only occur under
two circumstances: an accident or intentional use. Accidents occur infrequently
by definition. Intentional defiance of a taboo is rare because one must first

13
This subtext should be familiar to persons fluent in any language. It is the meaning behind the
words of any conversation. What one says is just as important as how one says it; choices of
diction can radically change the meaning of a conversation using grammatically equivalent words.
For example, it has been noted that in the English language much of the language used to describe
women or their actions is loaded with the notion that they are a form of property and are therefore
devalued as people (Bolinger 1980, 92; Lakoff 1973, 48). Correct, “ladylike” language was
required of women for centuries, while masculine form of address was typically frowned upon,
such as swearing, assertive language, or inattention to grammar or pronunciation (Lakoff 1973,
45-80; Romaine 2000). The same sort of linguistic behavior was and still is used against people of
color in the United States; the difference being that instead of property (nowadays) things typically
associated with “blackness” are lesser and the concerns of the black community are overstated and
should be ignored or devalued. English is not the only language in which this occurs. As noted by
many French students, many words denoting bad things are considered feminine (ex: catastrophe,
night).

31

recognize that it exists. As with many unconscious beliefs and behaviors, it is
unlikely for individuals to consciously identify them without intending to do so.
Secondly, the use of the word taboo signals what can be expected of the
consequences for engaging in that behavior. As mentioned above, the
consequences are understood to be dire, if undefined. This is because taboo
violators have just disobeyed deeply and subconsciously held beliefs by the
community about what is socially appropriate. Violation therefore necessitates a
response proportionate to the perceived depth of defilement from society. A
severe and sustained response is often the result. Because of this, taboos can still
exist in the face of occasional violation, as long as the social response is
tremendously negative or punitive.

Cases of Use- Challenge to Tabooness Instances in which individuals violate
taboos serve as challenges to their continued existence in two ways. First, an
intentional violation of a taboo is an indicator that the self-perpetuation
mechanism has failed. Taboo acts are no longer unthinkable; someone thought of
it, potentially performed a cost-benefit assessment over violating the social
stricture, and then crossed the red line on purpose. The moral restraint felt by
those who adhere to the social rules has broken down and what was once
inconceivable has broken into reality. Accidental violation of taboos is not quite

32

so bad; instead of intentionally committing a taboo act, circumstances conspired
to result in the breaking of the taboo (ex: accidentally consuming human flesh or
engaging in sexual intercourse with a stranger that turns out to be a long-lost
twin). Even then, however, the shadow of stigma will remain on the violator.
Secondly, the social response to taboo violation can serve as a challenge to
tabooness. If society doesn’t punish the violator, the response by society violates
the bright-line character of a taboo. In order to retain its force, a taboo must place
an enduring social stigma on the violator, ensuring that that individual will never
be reconciled to the community (ex: those who intentionally kill and eat their
neighbors are forced out of every neighborhood they try to settle in). In this way,
the relationship between the violator and society has been irrevocably changed.
To ensure that this bright-line is drawn and understood, extreme social outrage
needs to occur (to preserve individual perception that the taboo should be
obeyed), resulting in punishment and/or ostracization (to make clear that any
future violators will be socially sanctioned).

Does the “Chemical Weapons Taboo” Qualify as Such? Now that the concept of
taboo has been defined, do states’ reluctance to employ chemical weapons
constitute an international taboo? There are two ways to assess this. The first way
to assess the validity of the chemical weapons taboo is by identifying the

33

frequency of violation. If the chemical weapons taboo exists, usage of chemical
weapons is expected to be fairly low, similar to the instances of nuclear weapons
use. The second evaluation method looks at the international response to chemical
weapons use. If there are overwhelmingly negative or punitive responses to
chemical weapons use by members of the international community, the argument
for a chemical weapons taboo has merit.
The number of instances in which states have used chemical weapons is
debated. As can be imagined, there is an incentive to cover up state use of
chemical weapons to prevent intervention by other states and the UN or to avoid
stigma. At the same time, there is an incentive for civilians to claim that chemical
weapons have been used in order to attract international attention (and hopefully
support) for their plight, whether or not that is the truth. Short of dispatching
evidence collection teams armed with test tubes and sophisticated equipment, the
determination of whether or not chemical weapons have been used typically relies
on the expert opinions of physicians operating in the area, eye-witness
testimonies, and the descriptions of the effects of the supposed chemical weapon
coupled with the user’s known chemical holdings. While usually non-systematic,
most of the determinations of chemical weapons use in the last century rest on
analysts’ gut instincts, ability to back those up with evidence provided by
international aid organizations, and interpretations of deliberately vague
international treaties.

34

Over 40 accusations of chemical weapons use have been levelled at
various governments since the end of WWI, per the definition presented in the
Chemical Weapons Convention.13F
14
Only about ten instances of chemical weapons
use are widely discussed in the literature and are agreed upon as verified by
historians.14F
15
These include Britain in the USSR (1919), Italy in Libya (1921),
Spain and France in the Rif War (1921), Italy in Ethiopia (1935), Japan in the
Second Sino-Japanese War (1937), Egypt in the Yemeni Civil War (1963), Iraq in
the Iran-Iraq War (1983-1988), Libya in Chad (1987), and Iraq in Halabja (1988).
Syria (2013-2015) is now the tenth confirmed instance of chemical weapons use.
A number of disputed accusations are fated to remain unresolved due to lack of
evidence collection.15F
16
Additionally, terrorist organizations have resorted to
chemical weapons in their attacks, including usage by Aum Shinrikyo in Japan
(1994), rebels in Chechnya (1997), and insurgents in Iraq (2007).
If all the accusations of chemical weapons use are true, it does not appear
that a taboo against the use of chemical weapons exists. With 40 instances in

14
These accusations include usage of 1) those chemicals defined in the convention as chemical
weapons and 2) chemicals allowed under the convention for policing matters that were used as
weapons in warfare. It does not include instances in which herbicides were used against
vegetation, as these do not contravene the convention.
15
It should be noted that most chemical weapons histories cover additional cases, with little
agreement on which additional cases should be included. The use of chemical weapons constitutes
a segment of “forgotten history” in this author’s opinion, and the true accepted instances of
chemical weapons should be much higher.
16
These include the US in North Korea and China (1951), the US in Vietnam (1964), Vietnam and
the USSR in Laos (1975), Vietnam in Kampuchea (1978), the USSR in Afghanistan (1979), and
Serbia in Bosnia (1995).

35

approximately 100 years, an accusation of chemical weapons use appears every 2-
3 years on average. If only the ten accepted instances of chemical weapons use
are used to gauge chemical weapons tabooness, the taboo is still in trouble when
compared to the nuclear taboo as described by Tannenwald. When compared to
the number conflicts these states fought in during the post WWI era, the taboo
falls in further doubt. On average, a state was accused of using chemical weapons
in 21.5% of conflicts. Of those states that were accused of using chemical
weapons, they were accused of using them in 37.8% of the conflicts they were
involved in.16F
17
If the idea that the use of chemical weapons constitutes a taboo in
the international system was valid, one would expect that states would only use
them in a small percentage of the conflicts they engage in, if at all. This does not
appear to be the case, as around 1/5
th
of the conflicts engaged in by chemical
weapons possessor states experienced accusations of usage.
A second way to assess whether the disinclination to use chemical
weapons qualifies as a taboo is investigating the international response to obvious,
well-publicized violations. Five well-known state uses of chemical weapons have
been chosen: Italy vs. Ethiopia (1935-1936), Japan vs. China (1937-1945), Egypt
vs. Yemen (1963-1967), Iraq vs. Iran (1983-1988), and the recent use of chemical
weapons by the Syrian regime against insurgent forces (2013-2015). These

17
For more information on how these statistics were calculated, please see Chapter 3.

36

incidents form hoop tests regarding the hypothesis that the non-use of chemical
weapons is a taboo. While passing a hoop tests confirms the relevance of a
hypothesis, failing one eliminates the hypothesis as valid (Bennett 2010, 210; Van
Evera 1997, 31-32). If evidence of a taboo cannot be found in these cases in
which the violation of the chemical weapons “taboo” was internationally
recognized, the chemical weapons taboo does not qualify as such.
During the Italo-Ethiopian war fought between 1935 and 1936, the Italians
used chemical weapons during their colonial expansion into Ethiopia, despite the
prohibition enshrined in the Geneva Protocol which they signed in June 1925.
Preparations for the acquisition of Ethiopia began as early as 1924-1925, when
Italian officers began assessing Ethiopian defenses and developing contingency
plans (Badoglio 1937, 4-5; Del Boca 1969, 15; Del Boca April 5, 1965).
Mussolini made the official decision to invade Ethiopia upon realizing that Libya
and Italian Somaliland were poor colonies and could never solve Italy’s
demographic problems (Baskerville 1937, 8). Despite signing two treaties with
Ethiopia and ratifying the Geneva Protocol in 1928, Italy shipped over 1,000
chemical munitions to Mogadishu between 1930 and 1932, reportedly to be used
in the event of an Ethiopian invasion of Somalia (Del Boca 1969, 8; Sbacchi
2005, 47). By 1935, the Associazone Commercio Asfissianti in Italy had
produced 500 tons of mustard gas. In June of that year, the Commander-in-Chief
of the Italian Forces in Africa, Emilio De Bono, ordered the shipment of 75,000

37

arsine artillery shells to Eritrea and Somalia (Sbacchi 2005, 48). The Italians were
clearly anticipating the massive use of chemical weapons in the future conquest of
Ethiopia. On September 27, 1935 Mussolini ordered General De Bono to invade
Ethiopia, which he did on October 3 from Eritrea (Selassie 1949, 22, 28).
The first instance of chemical weapons use occurred on October 10, 1935
on the town of K’orahe and was ordered by General Graziani (Sbacchi 2005, 49).
Chlorine was reported to have been used at Ogaden six days later, the same day
that De Bono granted Graziani official permission to use chemical weapons
against K’orahe (Anonymous 1935; Sbacchi 2005, 49). On October 17, Mussolini
countermanded De Bono’s decision, reserving the authority to order chemical
weapons use for himself (Sbacchi 2005, 49). Italian forces experienced delays,
and on October 29, an impatient Mussolini authorized Graziani to use chemical
weapons on the southern front to overcome enemy resistance (Ministero
dell'Africa Italiana 1935; Ministero delle Colonie 1935; Sbacchi 2005, 49). He
later gave his generals carte blanche regarding chemical weapons use on
December 16, 1935 (Del Boca 1969, 78). Many hundreds of tons of chemical
weapons were used extensively through the end of the war on May 5, 1936.
Soviet commentators estimated that 15,000 of the 50,000 Ethiopian casualties
were due to chemical weapons use (Stepanov and Popov 1962).

38

Appeals to the League of Nations for help by the Ethiopian Emperor
began on December 30, 1935, claiming that the Italians had used chemical
weapons. The Italians denied the charges, but later asserted that they had only
used non-lethal chemicals and that it was only in response to the Ethiopian use of
dum-dum bullets (Coleman 2005, 47; Del Boca 1969, 81; Selassie 1949; Thomas
and Thomas 1970, 141).17F
18
Reports by aid workers in the area reveal that the first
claim was conclusively false (Maryam 1936, 605-643). The Italians argued that as
the Geneva Protocol did not prohibit either the use of chemical weapons against
non-signatories nor the use of chemical weapons as retaliation for use of other
illegal means of warfare they were actually in compliance with the Protocol
(Thomas and Thomas 1970, 142). There were debates in the League of Nations
whether the Italians had used chemical weapons illegally, but nothing was done to
intervene in the war or punish the use of chemical weapons.
During the fighting of the Second Sino-Japanese War (1931-1945), China
accused Japan of extensive use of both chemical and biological weapons. In
February 1932, Japan denied using toxic chemicals against the Chinese, claiming
that they were simply using smoke screens (Anonymous 1932, 363). Accusations
resurfaced in 1937 when the Chinese notified the League of Nations that Japan

18
The Italians alleged that the Ethiopians resorted to the use of dum-dum bullets beginning in
November 1935. These bullets expand on impact and were banned in warfare by the Hague
Convention of 1899. It should be noted that Ethiopian use of these munitions occurred nearly a
month after Italian chemical weapons use began, generally negating the Italian argument that they
used their weapons in response to Ethiopian violations of the Hague Convention.

39

had used poison gas in Nankow after fighting for two weeks (Koo 1937, 655-
657). This was swiftly followed by more accusations, where Japan was accused of
using both lethal and non-lethal chemical weapons, many times against unarmed
combatants (Chi-tsai 1938, 666; Chi-tsai 1938, 667-668; Chi-tsai 1938, 665; Koo
1938, 1133-1135; Koo 1938, 381). In all, it is estimated that Japan conducted
approximately 2,000 chemical weapons attacks during the conflict with China,
causing over 80,000 casualties, between 1937 and 1945 (McCarthy 2005, 169-
170).18F
19
Soviet authorities estimated that approximately 25% of artillery projectiles
and 30% of avian munitions used by the Japanese arsenal in China contained
chemical agents (Stepanov and Popov 1962).
Beginning in 1938, the Chinese government brought charges and evidence
against Japan to the League of Nations, requesting that the League intervene
(Anonymous 1938, 378-380; Chi-tsai 1938, 665; Koo 1938, 381). At the time,
additional Japanese chemical forces were en-route to China, and China requested
that the League forestall the future usage of chemical weapons against the
Chinese people. It is not evident that anything was done beyond the League
calling upon its member nations for more information despite the recognition that
the atrocities in China had occurred.

19
It should be noted that Japan signed the Geneva Protocol in 1925, although it did not accede to
the treaty until 1970.

40

A US House of Representatives report issued after WWII stated that the
Japanese were guilty of a number of chemical weapons attacks on Chinese forces
between 1937 and 1943 (Coleman 2005, 51; Thomas and Thomas 1970, 143).
The Japanese were reported to have used a wide number of chemicals, including
mustard gas and lewisite. In the indictment brought against Japan at the Tokyo
War Crimes Tribunal, the Japanese were charged with violating the customary
law against chemical weapons usage. However, this charge was tenuous at best
and the Japanese denied it (Coleman 2005, 51). In exchange for immunity from
war crimes charges, many Japanese scientists voluntarily joined the US chemical
weapons program.
As early as 1963, Egypt was accused of using tear gases against Royalist
forces in the Yemeni Civil War (I. Lundberg. 1969; McLean 1965; D. M. van
Rosmalen. 1967). The Egyptians denied the charge, and US investigators failed to
find any evidence supporting the accusations (Coleman 2005, 101; Thomas and
Thomas 1970, 151). Chemical weapons attacks on Yemeni villages were again
reported in January 1967. Over 40 different instances of chemical weapons use
were described in the press, although there is some question as to the veracity of
some accusations. Casualty figures were available for less than half of the alleged
incidents, but those instances alone totaled at least 1,400 people killed and 900
injured as a direct result of chemical exposure (Stockholm International Peace
Research Institute 1971, 159-160). Vesicants and lung irritants were rumored to

41

have been used beginning in 1965, but this was not substantiated until 1967. That
year the International Committee of the Red Cross verified that an unknown party
had used both types of chemical agents (Stockholm International Peace Research
Institute 1971, 160; Thomas and Thomas 1970, 151).
News reports exclusively blamed the Egyptians, even though their
government denied any responsibility. Some even speculated that the USSR had
supplied some of the agents used. Chemical weapons attacks continued despite
the accusations, prompting the British House of Commons to issue a statement
deploring their use in Yemen. The US also stated that it deplored the conduct of
the war, and offered to support international action to deal with the problem
(Thomas and Thomas 1970, 152). However, the US preferred that Saudi Arabia
present the case to the UN. Even though Egypt clearly violated the Geneva
Protocol (which it signed in June 1925) by using chemical weapons in Yemen, the
international system largely ignored the conflict due to political reasons and the
advent of the Six-Day War (Coleman 2005, 102; Thomas and Thomas 1970, 153).
In 1980, Tehran radio broadcasted allegations of Iraqi chemical weapons
usage against the town of Susangerd (Coleman 2005, 109). In mid-July 1982 Iran
claimed that Iraq had used riot-control agents to break up advancing infantry in
the assault on Basra. As they had enjoyed some initial success, the Iraqis began to
compensate for the scattering of their forces along a 650-mile front by using aerial

42

chemical munitions in the summer of 1983 (Spiers 2010, 104). That year, the
Iranian ambassador to the UN alleged that a total of 49 chemical weapons attacks
had taken place between 1980 and 1983. Casualty lists included 109 dead and
hundreds more wounded (Coleman 2005, 109). Iran later claimed that between
February and March 1984 Iraq used chemical weapons on 20 additional occasions
in violation of the Geneva Protocol (which it had ratified in 1931), wounding an
additional 2,200 people. The attack on Hoor-ul-Huzwaizeh in March 1984 was
conclusively verified by the UN Secretary General (Coleman 2005, 109). Both the
International Red Cross and the US supported the Iranian claims, even though
Iraq denounced the US claim as “political hypocrisy” and “full of lies” (Coleman
2005, 109). On March 30, 1984, the UN Security Council issued a statement
condemning the use of chemical weapons in the Iran-Iraq War. The US and other
countries restricted chemical exports to both belligerent countries in order to
prevent the production of any further chemical weapons.
As Iraqi forces built their stockpile, they developed their tactics and
employed chemical weapons more extensively in the defense of Faw (1986) and
Basra (1987) (Spiers 2010, 106). Even though the number of reports of chemical
weapons usage diminished overall, Iraq continued to use chemical weapons until
the conclusion of the war in August 1988 when Iraq finally accepted the UN
ceasefire plan. The Iraq Survey Group estimated that over 1,800 tons of mustard
gas, 140 tons of tabun, and over 600 tons of sarin were used in over 100,000

43

chemical munitions by Iraq during the conflict (Spiers 2010, 109). In total, Iran
claimed that of the 500,000-600,000 casualties it suffered during the Iran-Iraq
War, 60,000 of them were due to chemical weapons exposure (Anonymous 1995,
20; Ali 2001, 43-58).19F
20
There are some allegations that Iran eventually retaliated
in kind, such as the battle at Sulaimaniya on July 1, 1988, but conclusive evidence
remains elusive (Spiers 2010, 106-107).
Within a month of the ceasefire, reports circulated that Saddam Hussein
had authorized the use of chemical weapons against the Kurdish population in
northern Iraq. On August 25, 1988, the Iraqi army launched a massive offensive
to suppress the Kurdish insurgency, causing a mass migration of between 50,000
and 65,000 Kurds into south-eastern Turkey (Coleman 2005, 110; Spiers 2010,
108). Press reports indicated that they fled when faced with large-scale chemical
weapons attacks. In particular, the Iranian-occupied city of Halabja was
bombarded by the Iraqi Army with various chemical agents for three days, killing
between 5,000 and 8,000 Kurdish civilians (Ali 2001, 43-58; Leonard Doyle,
Donald Macintyre and Tom Wilkie. 1991). US estimates place the casualty count
at approximately 4,000 civilians killed and 10,000 injured (Spiers 2010, 107). Iraq
denied the charges.

20
Iraqi estimates put Iranian chemical casualties at 30,000 (Spiers 2010, 109).

44

On September 12, 13 nations requested that the UN investigate these
allegations. The UN requested access to the region in order to investigate, but
Turkey and Iraq both denied the requests (Coleman 2005, 111). Physicians for
Human Rights, who were stationed in the region and were caring for the refugees,
stated that Iraq did engage in aerial chemical weapons attacks which killed both
humans and livestock. Interviews with survivors in the camps and treatment of
their symptoms corroborated their allegations (Coleman 2005, 111).
Beginning in March 2011, the Assad regime in Syria became embroiled in
a civil war that eventually reached a stalemate. In the fall of 2012, Western
intelligence agencies noticed that the Assad regime was moving around its
chemical assets, potentially presaging a chemical strike (Eric Schmitt and David
E. Sanger. 2013). Several countries privately communicated warnings to the
Syrian government, halting the regime’s chemical mixing and bomb preparation
(Eric Schmitt and David E. Sanger. 2013). The US publicly warned the Assad
regime against use of their chemical weapons, first with Obama’s famous “red
line” speech and an additional warning in December 2012 (Anonymous 2012;
Peter Baker and Michael R. Gordon. 2012; Peter Baker. 2012). On March 19,
2013 the first attack using chemical weapons in Syria was alleged to have
occurred, killing approximately 24 people and injuring another 124 (Baum 2015,
137). Attacks occurred sporadically through the end of August, with the largest
occurring on August 21, killing 1,429 people and injuring approximately 2,170

45

others (United Nations Mission to Investigate Allegations of the Use of Chemical
Weapons in the Syrian Arab Republic 2013a; United Nations Mission to
Investigate Allegations of the Use of Chemical Weapons in the Syrian Arab
Republic 2013b; United States of America 2013). The US blamed this attack on
the Assad regime, citing intelligence that placed personnel from Assad’s chemical
weapons branch in the area and chemical filled munitions originating from
government-controlled areas of Damascus (Day August 27, 2013; United States
of America 2013, 2-3).
Many states reacted with shock and dismay that chemical weapons had
been used. Some states advocated punishment for the Syrian regime, including the
use of military force.20F
21
Others declined taking a stance on the issue beyond
disapproval that chemical weapons had been used. China and Russia blocked US
attempts in the UN Security Council to authorize the use of force. Tensions
between Russia and the US grew so strained that it appeared that the two
countries were on the brink of a physical confrontation over Syria. Due to a
combination of diplomatic skill and serendipity, Russia managed to broker a deal
by which the Syrian government agreed to surrender their chemical arsenal in
exchange for the US not militarily intervening in its civil war. The removal of
chemical weapons from Syria began in December 2013, but did not stop the use

21
These countries included South Korea, Germany, and the United Kingdom (Anonymous 2013a;
Anonymous 2013b; Lesley Clark and McClatchy Washington Bureau. 2013; Dempsey 2013).

46

of chemical weapons. Beginning in April 2014, crude chlorine bombs were used
against civilian populations within Syria. While the US has declined to place
blame, facts in evidence strongly suggest that the Assad regime is responsible
(Organization for the Prohibition of Chemical Weapons. Office of the Director-
General December 18, 2014; Hannah Lucinda Smith. 2015). On March 6, 2015,
UN Resolution 2209 demanded that use of chemical weapons cease, warning that
the Security Council would authorize the use of force if necessary; use of these
weapons by the regime has since abated (Anonymous 2015).
From the above cases of chemical weapons use in which use was well
known and acknowledged by the international community, it is clear that
international response to the use of chemical weapons is generally mild. When
action is taken, it has typically been a form of mild punishment (embargo,
surrender of arms). These are not actions typically expected of states reacting to a
taboo violation. The violation of a true taboo results in severe backlash from
society. Whether it be through shunning or punishment, open taboo violation
results in a redefinition of the violator as ‘other.’ To avoid contamination, society
reacts to prevent being associated with those acts and encouraging further
violation. The use of chemical weapons discussed above resulted in social
backlash, but in an incohesive and inconsistent manner. If a taboo violation had
occurred, exceedingly few states would have defended the violators; most would

47

have distanced themselves from those regimes or advocated for severe
punishment.

What It Should Be Called Instead
If the “chemical weapons taboo” does not qualify as such, then what
should the disinclination to use chemical weapons be called instead? While the
nonuse of chemical weapons should not be considered a taboo, it does qualify as
an international norm. More particularly, a prohibitive norm.
Specifically speaking, a norm is a standard of appropriate behavior for
actors with a given identity that channel and regularize behavior (Finnemore
1996, 22; Katzenstein 1996a, 3; Katzenstein 1996b, 5; Klotz 1995; Lapinski and
Rimal 2005, 129; Sunstein 1997, 40). Norms typically exist in two forms:
prescriptive and proscriptive norms. Prescriptive, or “thou shalt,” norms are
injunctions to behave in specific ways that that society deems must or should be
performed by its constituent individuals.21F
22
Examples include paying taxes,
reporting child abuse by mandatory reporters, and wearing clothing.22F
23

Proscriptive, or “thou shalt not,” norms are prohibitions from engaging in

22
These norms are also known as “injunctive” norms (Cialdini, Reno, and Kallgren 1990, 1015-
1026).
23
These examples are specific to contemporary American society. While these specific norms may
not apply to other areas of the world, the underlying mechanics for how their norms work is the
same.

48

behavior that is deemed socially undesirable.23F
24
Cross-dressing, murder, and theft
are examples of proscriptive norms.
Both prescriptive norms and proscriptive norms vary in strength,
depending on the weight society gives that particular norm. A prohibitory norm, a
proscriptive type of norm, forbids certain actions, places, or behaviors, and
enforces this ban because society believes it to be socially important. This
enforcement is accomplished through the use of formal social sanction
mechanisms, such as the legal system. In this way, prohibitory norms are not self-
enforcing.
Prohibitory norms are also not self-perpetuating; self-perpetuation requires
that society follows the rules of the prohibition a vast majority of the time. This
necessitates that most people never even consider violating the norm or that they
undergo an internal pro-con debate resulting in the decision to not violate social
rules. Frequent violations of prohibitory norms, however, reveal that these norms
are not self-perpetuating and rely on an active penal system to maintain their
social force. Good examples of prohibitory norms include murder and theft.
Murder and theft are more often than not punished in all societies, usually through
a criminal justice system. However, they do not qualify as a taboo because these

24
These norms are also known as regulative norms (Finnemore and Sikkink 1998, 887-917;
Muller 2013)

49

crimes are regularly committed, even though society chooses to impose sanctions.
They also do not function as ‘bright-line’ norms, per Tannenwald’s definition
discussed above, nor have uncertain or awful consequences. The consequences of
prohibitory norm violation are understood in advance, mainly because their
violation is enshrined in legal code.

The Differences Between a Prohibitory Norm and a Taboo While the distinction
between norm and taboo is a subtle one, the frequency of their violations play a
major role in differentiating the two. Norms, by definition, can withstand
violations because there is an external enforcement mechanism. As long as that
mechanism remains robust, social norms against certain behaviors remain in
effect. In contrast, taboos are rarely violated. Instances of violation are a
challenge to the integrity of the taboo; repeated violations tend to downgrade
taboos into lesser prohibitions on social behavior. Therefore, if a taboo has
survived, the number of instances of its violation are close to zero.
In addition to needing an external enforcement mechanism and violations
occurring more frequently than taboos, prohibitive norms also experience less
social outrage upon their violation. While the public might be “outraged” to learn
of a murder or robbery of a much-loved business, taboos experience another level
of response. Taboo violation tends to result in a visceral disgust in the violation

50

on the part of society (Jefferson 2009). This primal rejection of the violation and
the violator tends to result in calls for either avoiding or hurting behavior. The
violator is either cast out of the group wholesale, forbidden to return, or is subject
to physical violence or threats of violence (Crocker and Major 1989, 608-630;
Kerr and Levine 2008, 39-52; Kerr and Levine 2008, 39-52; Kurzban and Leary
2001, 187-208). While those who violate prohibitory norms are allowed to reenter
society, they may be required to engage in extensive rehabilitation or acts of
contrition. In contrast, those who violate taboos are rarely, if ever, allowed to
reconcile with society. If they do reintegrate, their previous behavior casts a
lingering pall on their relationships.

Why the Use of Chemical Weapons Needs to Be Explained The perception of the
frequency of the occurrence of unwanted social behavior is not only important for
those concerned with sociology or anthropology. The perception of the regularity
of the violation of standards of appropriateness plays a huge role in determining
one’s perceived individual risk of experiencing those violations. Taboo violation
approaches zero, meaning that the individual risk of being a victim of taboo
violation is also approximately zero (ex: probability that a cannibal will eat one
for lunch). In contrast, norm violation is more frequent. This means that the
perception of one’s individual risk is greater than zero (probability one will be a
victim of theft).

51

Correctly assessing a state’s risk for norm/taboo violation is incredibly
important. Since the consequences for violation can be dire, risk must be
accurately known in order to determine if preparation on the part of the
government is necessary in an attempt to ameliorate the effects of the violation
itself. Mislabeling norms as taboos can have serious consequences, as preparation
is generally deemed unnecessary (if the probability of taboo violation is
approximately zero, then it is more cost effective to spend resources elsewhere).
This places the state’s populace in greater danger of severe problems if the norms
(mislabeled as taboos) are violated. Failure to prepare could even threaten the
survival of the regime itself.
For example, if a state perceived a non-zero likelihood of experiencing a
chemical attack, it could take steps to lessen the severity of the situation, given
that it possessed the resources to do so. These measures would include training
their military on operating in a chemically contaminated environment, drafting
response protocol, and developing/improving chemical isolation gear. If the state
perceived a zero or near zero chance of chemical attack (taboo), it would have no
incentive to invest the millions of dollars needed to prepare. In the event of an
attack, that state would be virtually helpless as their enemies killed their military
forces with impunity. The country could be conquered by outside forces, deposing
and possibly killing those in power. By mislabeling the disinclination to use
chemical weapons as a taboo, states are rendered helpless against norm violations,

52

inflating body counts and the number of people who need intensive and expensive
medical care.
This hypothetical actually occurred during the Second Italo-Ethiopian War
(1935-1936) described above. Ethiopia faced the Italians entirely unprepared for
chemical warfare. Even though the Italians had forward-based chemical munitions
for years, the Ethiopian government did not invest in protective measures for its
troops and population. As a result, Italian chemical weapons devastated the
Ethiopian defense. With no possible defense, Ethiopia never stood a chance
against the military superior forces of Mussolini. It remained an occupied country
with its Emperor in exile until the 1941 liberation by the Allied forces during
WWII.
Since the consequences of unintentionally exaggerating the strength of the
international disinclination to use chemical weapons can be dire, more about the
factors surrounding the use of chemical weapons needs to be known. While case
studies may prove enlightening on individual cases, a more systematic approach
needs to be taken. With the number of acknowledged uses of chemical weapons at
least ten and allegations approximately quadruple that, something more
systematic appears to be at work. The number of cases is too high for a qualitative
study, but too small to do a statistical analysis. A hybrid method needs to be used.

53

Conclusion
This chapter advances our understanding of the state disinclination to use
chemical weapons. By examining what the literature means when utilizing this
term in light of empirical data, this chapter demonstrates that the term “taboo” is
misused when it is used to describe the non-use of chemical weapons by many
states.
The literature understands the chemical weapons taboo in an
anthropological and sociological sense. The taboo on the use of chemical weapons
is the normative belief that these weapons should not be used, resulting in a
change in reality if this conviction is violated. Additional features of this
phenomenon include self-perpetuation and self-enforcement. Adherence is
unthinking and perpetrated by the general lack of questioning whether this dictate
should be followed.
This chapter demonstrates that the state non-use of chemical weapons does
not qualify as a taboo, if one is adhering to the definition of the term. The
normative belief that chemical weapons should not be used has been violated
between 10 and 40 times since the end of WWI, in which the spirit of
international prohibitions was drastically shattered. Additionally, the international
response to “taboo” violation has typically been tepid, particularly in cases in
which violation is not in doubt.

54

This chapter proposes instead that the phenomenon of the state
disinclination to use chemical weapons should be labelled a prohibitive norm.
These norms forbid certain actions because society believes them to be important
and enforces them through formal sanctioning mechanisms, such as international
law. Prohibitive norms are not self-perpetuating, as they are violated on a far
more frequent basis than a taboo would.
Relabeling this phenomenon to more accurately describe its core attributes
and frequency of violation will greatly impact the perception of individual and
societal risk attached to the use of chemical weapons. If risk were more accurately
perceived and prepared for, many lives could be save by preventive measures.
The following chapter outlines the methodology of a fs/QCA study designed to
identify which conditions result in chemical weapons usage and restraint. By
identifying these conditions, the risk of experiencing chemical warfare can be
more accurately assessed and states are given a window of opportunity to avoid
the scenario all together.

55

Chapter 3
How Do We Know What We Know?

56

The only way to truly assess the risk of chemical weapons usage is to
examine cases of recent historical cases of use in a systematic manner. Once these
cases are identified and inspected, the conditions under which states use chemical
weapons can be readily identified. The aim of this chapter is to present the
methodology behind how this fs/QCA study will investigate the state use of
chemical weapons. First, a discussion of QCA’s background, strengths, and
drawbacks will be presented. fs/QCA will be introduced as a way to minimize
methodological drawbacks. Next, this authors adjustment of fs/QCA to produce
supersets will be proposed. The use of supersets will be proposed as a way to
identify conditions sufficiency in later fs/QCA analysis. Third, the
implementation of a coding scheme will be described. The coding rules will
include a description of case selection and the specific rules for assigning fuzzy
set scores to cases. Descriptive statistics of the included cases will also be
presented. Lastly, an explanation of the epidemiological method to assess which
cases are the drivers behind each analysis will be presented.

What Is QCA and How Is It Different from Normal Statistics?
QCA is a methodology that was introduced to the study of political
behavior and sociology in the 1980s by Charles Ragin. While relatively new to
the social sciences, QCA has epistemological underpinnings that reach back to the

57

foundations of scientific inquiry. The basis for QCA emerged in the works of
Hume and J. S. Mill as they tried to identify the causes of phenomena by
controlling the similarities and differences between variables potentially
explaining the outcome (Berg-Schlosser et al. 2009, 2; Hume 1758; Mill 1967
[1843]).
QCA is neither strictly qualitative nor quantitative; it is a hybrid of both
methodological schools. By using set theory and Boolean algebra, QCA is able to
systematically analyze many aspects of cases to identify which are the most
relevant. “By formalizing the logic of qualitative analysis, QCA makes it possible
to bring the logic and empirical intensity of qualitative approaches to studies that
embrace more than a handful of cases- research situations that normally call for
the use of variable-oriented, quantitative methods” (Ragin and Rubinson). By
staying grounded in the details of individual cases, QCA serves to bridge to gap
between small-n and large-n research by taking the best of both methodologies.
When using QCA, a case is scored in terms of set membership across a
number of different potentially sufficient and outcome conditions. Software is
then used to sort through patterns of conditions that produce the outcome in order
to identify common conditions.24F
25
In order to correctly calibrate the score of a case
in any given condition, in-depth knowledge of the case is required. The

25
The software used in this analysis is (Ragin, Drass, and Davey 2006)

58

combinations of conditions are then assessed using computer algorithms for
logical sufficiency and then simplified to identify the key components that appear
across the spectrum of cases.
The best work using QCA is highly transparent as to the scoring
requirements across the different components and as well as to why the individual
cases are scored as they are. This allows for a checking of borderline cases. By
altering the scoring, the weight each case exerts on the outcome can be assessed;
changes that do not contribute novel sufficient conditions or dramatically change
the results of the QCA analysis can then indicate that the analysis is robust
concerning small scoring deviations in the data. Because the scoring is highly
transparent, it is then possible to identify deviant cases and assess their weight
during the analysis. Not all deviant cases are potentially logically inconsistent,
and the identification of them can serve as ideal cases to examine for the
researcher to assess whether the QCA results hold up under closer scrutiny.
Strictly speaking, QCA cannot identify causal mechanisms, which is a
feature that has been critiqued in the literature.25F
26
All the method can do is
describe the conditions that are present or absent when the outcome of interest is

26
This is not a real critique of QCA, as QCA does not intend to explain the mechanisms behind
the causal process and it is a feature QCA shares with statistical analysis. For more reading on the
topic, see (Abbott 1995, 93-113; Abbott 2001; George and Bennett 2005; Gerring 2005, 163-198;
Goldthorpe 1997, 1-26; Mahoney 2003, 337-372; Mahoney 2004, 459-489; Pierson 2003, 177-
207; Rueschemeyer and Stephens 1997, 55-72; Stephens 1998, 22-25).

59

observed and not observed (De Meur, Rihoux, and Yamasaki 2009, 160). Because
QCA cannot determine by which mechanism the identified conditional
configurations led to a specified outcome, the closest QCA can get is identifying
what are known as “risk factors” in epidemiology. A risk factor in epidemiology
is a variable that is statistically associated with an increased chance of disease or
injury as the outcome. Risk factors, or combinations of them, are purely
correlational but can indicate a complicated or even convoluted causal
relationship between the variables of interest and the outcome.26F
27
Factors
associated with a decreased risk for a particular outcome are known as “protective
factors,” and correlate to the QCA conditions that are associated with the absence
of a particular outcome. Once QCA has identified the factors involved in
producing the outcome, the researcher must then reexamine the cases to determine
the causal sequence between the risk factors and the outcome in question, much as
should be done with statistical methods.
Another critique of QCA is that the method only deals with dichotomized
variables. This is an important drawback, as this dichotomization can lead to the
loss of large quantities of information and charges of data manipulation and
measurement problems (Bollen, Entwisle, and Alderson 1993, 343; Goldthorpe

27
For example, age is a known risk factor for death. While increased age does not itself kill
people, it is a marker for underlying problems that are closely correlated with increased age, such
as decreasing efficiency of DNA repair mechanisms, decreasing organ function, and increased
immune susceptibility to infection.

60

1997, 7). While oversimplification is a methodological evil that should be
carefully avoided, simplification methods are a necessary feature of scientific
inquiry as the researcher is reducing the complexity of real life into controlled
chaos (King, Keohane, and Verba 1994, 42). As a form of simplification,
dichotomization allows research to go beyond the gradualist perspective and
forces the researcher to make a clear threshold (De Meur, Rihoux, and Yamasaki
2009, 151). Because the threshold in QCA is so precise, it serves as an indicator
of the validity of the condition (De Meur, Rihoux, and Yamasaki 2009, 151). In
order to minimize the loss of data in this project, a variant of QCA was selected,
fuzzy-set QCA (fs/QCA).27F
28
Instead of using a dichotomous scoring scheme,
fs/QCA scores set membership in conditions on a scale from [1] to [0]. In this
threshold system, scores higher than [0.5] are judged to be varying degrees of in
the set with [1] being completely in the set. Scores lower than [0.5] are considered
varying degrees on being out of the set, with [0] being completely out of the set.
The midpoint, [0.5], is not used as it represents the point of maximum ambiguity;
the researcher is unable to tell whether a case is in or out of the set.

28
There is significant uncertainty in judging whether or not the use of chemical weapons in fact
occurred. This is due to a number of reasons, among which the lack of laboratory identification of
substances in question and unbiased evidence that a particular entity is responsible for an attack
play key roles. Therefore, the decision was made to use fsQCA to preserve this sense of
uncertainty while conducting QCA analysis and avoid messy arguments about whether different
cases should be scored as definitely occurring or not occurring.

61

What can QCA Offer That Traditional Statistical Methods Cannot? QCA offers
researchers three key elements of analysis that traditional statistical methods are
incapable of delivering. Firstly, QCA assumes that multiple conjectural causation
is present in the data. Multiple conjectural causation is the idea that the causes of
outcomes are not single factors in isolation; instead outcomes are understood to be
caused by the presence of several factors, each of which are only sufficient to
cause the outcome in the presence of the others (Schneider and Wagemann 2012,
78).
Upon initial inspection of the cases, it became clear that there was no
single factor that could explain the use of chemical weapons. Not only were the
alleged users of chemical weapons not restricted to a single continent, these
agents were used in different types of conflicts by both poor and wealthy
countries in order to achieve different results. This led to the suspicion that the
decision to use chemical weapons could be more complicated than would
typically be allowed for by statistical methods. Statistical interaction terms would
not be helpful in illuminating the conditions under which chemical weapons are
used as it is likely that more than two conditions would need to interact to
produce the outcome. Higher-order terms are virtually absent from the statistical
literature, as this model puts high pressure on both the data itself and the
interpretive skills of the researcher (Brambor, Clark, and Golder 2006, 63-82;
Kam and Franzese 2007).

62

Secondly, QCA allows for and embraces the presence of equifinality.
Equifinality is the presence of “different, mutually non-exclusive explanations for
the same phenomena” (Schneider and Wagemann 2012, 78). Otherwise known as
the presence of “multiple causal pathways,” equifinality allows for the fact that
there are different ways in which to achieve the same situations (De Meur and
Rihoux 2002). Upon initial inspection of the data, it became apparent that not
only are the factors behind the decision to use chemical weapons different in each
case, no two decisions were similar enough to comfortably utilize statistical
analysis. While several instances occurred during civil wars, they were
perpetrated by countries with significantly different backgrounds facing different
political and military situations. This led to the suspicion that multiple pathways
toward the use of chemical weapons have been at work.
The presence of equifinality in statistical data presents a large problem of
researchers, as the method rests on assumptions of unifinality and additivity
(Schneider and Wagemann 2012, 86). The different elements of a linear
regression model are added together to reach a single result that explains the
outcome. Different variables are assumed to have the same incremental effect on
the desired outcome across all cases and the outcome is expected to not be
affected by the values of other causally relevant variables (Berg-Schlosser et al.
2009, 8). Therefore, statistical analysis runs into significant problems when

63

equifinality is present because it tends to reduce the significance of the individual
parts of the model, leading to a null outcome.
Lastly, QCA is a unique tool that allows for the examination of
asymmetric causation in the data. Causal asymmetry is present when the causes of
the outcome in question and the causes of the non-occurrences of the phenomenon
are different (X  Y ≠ ~X  ~Y). Upon inspection of the cases, it became clear
that some of the preliminary conditions associated with chemical weapons use
were also present in countries that never used chemical weapons. This led to the
suspicion that the reasons why chemical weapons were utilized were
fundamentally different from why they were not used. Additionally, orthodox
assumptions about reality of the chemical weapons taboo preventing chemical
weapons use were insufficient as the “taboo” was present in cases of both use and
non-use.
In addition to unifinality and additivity, linear regression rests on the
assumption of causal symmetry. While there has been an attempt to account for
asymmetric causation in statistical analysis, it is only applicable to situations that
are devoid of logical contradictions and situations in which empirical data is
present for all logically possible combinations of independent variables (Clark,
Gilligan, and Golder 2006, 311-331; Schneider and Wagemann 2012, 87). At the
very least, the assumption would need to be made that all missing information is

64

substantively irrelevant. While these multivariate tests for asymmetric hypotheses
may work in theory, they are unlikely to work in practice (Schneider and
Wagemann 2012, 87).
Statistical methods face severe challenges in the presence of multiple
conjectural causation, equifinality, and asymmetric causation, largely due to the
underlying assumptions built into the method. While QCA does not face these
same problems, it is vulnerable to a problem just as difficult. Because of the
mathematics used to analyze cases systematically, QCA is particularly vulnerable
to what is known as the “problem of limited diversity.”

The Problem of Limited Diversity
The problem of limited diversity is a feature of all naturally occurring
social phenomena, and is one that is particularly apparent in QCA because of its
utilization of logical truth tables. A problem of limited diversity exists when there
are logical remainders in social scientific data. A logical remainder is a set of
logically possible combinations of conditions for which there is either no or
insufficient empirical data (Schneider and Wagemann 2012, 151). An example of
such a combination is an instance in which all indicators of wealth are high and
the country’s economy is purely agrarian, completely unindustrialized, and only
utilizes 14
th
century technology. While logically possible, this particular

65

combination is not empirically observed. How is a social scientist able to draw
inferences from this case? While it can be assumed that the lack of data is an
indicator of the impossibility of those condition leading to the outcome, this
assumption becomes tenuous at best when there are large numbers of similar
logical remainders.
QCA methodology compounds the problem of limited diversity because it
has the potential to quickly increase the number of logical remainders. With the
introduction of each condition, the number of possible logical combinations of
conditions increases by 2
k
, with k being the number of conditions. This poses a
rather large problem for current practice in social science. The current procedure
in statistical analysis of selecting a few variables to focus on (ex: GDP, level of
democracy, party affiliation) and then adding a raft of control variables to deal
with confounding is simply untenable in QCA. For example, a QCA analysis
involving four conditions results in a total of 16 total possible condition
combinations. While not an overwhelming number of combinations, it is likely
that at least one row does not contain an empirical observation. Adding four
additional “control” conditions to deal with things such as age, sex, or location
increases the number of possible logical combinations to 256. Adding five
controls increases the number to 512. Meanwhile, the number of combinations
with empirical observations remains relatively the same, leading to serious
problems regarding logical inference.

66

QCA deals with the problem of having excessive amounts of logical
remainders through two joint methods, the first of which is to limit the number of
conditions it analyzes at a single time. This decreases the total amount of possible
logical combinations, increasing the odds that empirical observations are present
for the majority of them. The second method to deal with limited diversity was
proposed by Ragin. Called the “Standard Analysis,” this method produces three
QCA solutions varying on the number of assumptions that had to be made about
the missing data.28F
29
This variety of mathematical solutions is necessary because
each solution has carries a risk of distorting the data. By assuming different things
about data not in evidence in a controlled and methodical manner, the overall
picture of the data becomes more accurate.
Further complicating the problem of limited diversity is the fact that
different QCA runs using different conditions (ex: swapping out one condition for
the other) produces results that are not comparable. As QCA specifically looks for
interaction between conditions to develop sufficiency implications, missing
conditions can seriously weaken the overall analysis by being incapable of

29
These solutions are the conservative solution (no assumptions about logical remainders), the
most parsimonious solution (using all simplifying assumptions), and the intermediate solution
(using only easy counterfactuals). The conservative solution does not make sufficiency
assumptions regarding the outcome, and is a subset of all other logically minimized solutions. The
most parsimonious solution is the solution set that contains the least number of logical operators.
The intermediate solution is formed by dropping single conditions that do not appear in the most
parsimonious solution and are in line with directional expectations from the conservative solution
(Schneider and Wagemann 2012, 175).

67

making an informed decision on most of the possible condition combinations.
Therefore, one cannot simply split the conditions under consideration across a
range of QCA analyses.

Supersets: A Methodological Change for Unknown Causes
In this project, I propose a third method to deal with the problem of
limited diversity to be used in conjunction with the above two, particularly in
scenarios where the proposed causes of a phenomenon are not well known and a
number of conditions will need to be examined. In addition to limiting the number
of conditions to the absolute minimum and utilizing the Standard Analysis, a
social scientist can utilize the properties of condition sets to collapse them into a
single condition. This would involve creating logical supersets.
Sometimes called an inclusion, a superset is any set that contains all the
elements of a smaller set. One example would be the superset “Colors.” Inside
this set is every possible color that could ever exist, including all frequencies of
the electromagnetic spectrum. Another superset could be one called “Women.”
Inside Women are all possible conceptions of womanhood including an additional
set labelled “Pregnant.” Figure 1 is an example of a simple superset.29F
30

30
All Figures can be found in Appendix A

68

In logical terms, A is a superset (⊇) of B if A encompasses all elements of
B.30F
31
A would then be written A= (B). Additional elements of the superset could
then be added on if different categories were introduced (ex: A= (B, Red, 4,
rainbows, female)). Supersets can be defined to include any value one could
choose, including other sets, colors, dates, etc. This property makes it a valuable
simplification tool for fs/QCA. For instance, if one wanted to do an fs/QCA
analysis using six different conditions, three of which share a common logical and
theoretical superset (ex: inter-state wars, extra-state wars, and intra-state wars are
all different types of war). By substituting the superset for the three individual
conditions (ex: at war) into the fs/QCA analysis, one can run an analysis using
only four conditions. This would drastically diminish the problem of limited
diversity.
Example: Conditions1: A, B, C, D, E, F; 2
6
logical combinations
𝑆 ⊇ (A, B, C)  S can be substituted in for A, B, and C
Conditions2: S, D, E, F; 2
4
logical combinations, 48 less logical
combinations
While the use of supersets makes intuitive logical sense, the problem is
how to assign a value to the superset. Since it is constructed purely from subsets
and there are no empirical observations about the superset itself, one cannot

31
Using the above example, the superset Women encompasses all elements of Pregnant because
there isn’t a case in which one is not be a woman and pregnant (ex: pregnant men). However, it is
allowable that one can be a Woman and not pregnant (within the larger circle A but not in B).

69

simply rely on empirical data for clues on how to assign values. One has to look
at how the superset was constructed.
If we define the superset S to include subsets A, B, and C (see Figure 2), S
= (A, B, C). For each case, A, B, and C have numerical values representing the
degree to which a given case belongs to that given set. Assigning a numerical
value to superset S is the same as assessing the degree to which a given case
belongs to S. In other words, the value of S is the degree to which the case
belongs to all three sets simultaneously. In logical terms, the value of S for any
case is then equal to A*B*C.31F
32
Because this operator requires that there be
membership in all three cases, A*B*C is equal numerically to the case’s lowest
membership in any of the constituent sets.32F
33

The value of the superset cannot be a result of the other logical operator,
A+B+C. This is because the case membership in S would then be equal to the
maximal individual membership in either A OR B OR C, regardless of the values
in the other two. For example, let case membership in A =1, B = 0, and C = 0. If
the membership in S was calculated by A+B+C, then the highest value of the trio
(1) would be used as the overall value of a case in S. This is the same as saying
that a case has full membership in the superset, a statement logically inconsistent

32
Read A AND B AND C using the Boolean AND function.
33
The logical AND works on the logical property of the “weakest link in the chain.” In this
allegory, the strength of a chain in general is equal to the breaking strength at which its weakest
link snaps.

70

with knowledge that the case has no membership in two of the superset’s
components.
The utilization of superset would be ideal to initially assess a large number
of conditions for sufficiency without sacrificing analytical power to the problem
of limited diversity. In fs/QCA and QCA analysis, conditions are said to be
sufficient if they are scored more in the outcome set than the proposed condition
set. This translates to case scores above the line X=Y. Supersets use the minimum
value of the constituent X components as the overall score. This value is the
furthest to the left on an XY graph, and is the most likely value to be consistent
with sufficiency conditions, as is demonstrated in Figure 3. In this figure, the
potential values of a superset across its three components are highlighted in red.
Two of the components violate the sufficiency condition of being above the line
X=Y, but the component closest to the Y axis does not. When aggregated, the
superset takes the value of that component, indicating sufficiency. Because of
how the superset is constructed, it favors the conditions being sufficient to result
in the outcome. Therefore, supersets can be used to initially check for sufficient
conditions without resorting to a more detailed analysis.

71

Chemical Weapons Usage: Who Are We Talking About?
The conditions under which chemical weapons are used and not used are
yet to be understood. In order to begin the analysis and identify which conditions
are most important for the use of chemical weapons, a cross-national dataset was
constructed. In this dataset are all states since the end of World War I (November
11, 1918) through December 201433F
34
that were accused of possessing either
chemical weapons or a chemical weapons development program.34F
35
This includes
the following:
 States that possess captured munitions belonging to another state,
 States conducting research on chemical weapons, particularly for military
purposes,

34
Post WWI was chosen because the international norm prohibiting the use of chemical weapons
was truly formed during the fighting of WWI. Prior to WWI, there were two agreements by
European powers forswearing the use of chemical weapons (the Strasbourg Agreement (1675) and
the Hague Convention of 1899)). These treaties, which some states argued were not technically
broken, were completely ineffective in preventing the use of chemical agents during WWI. After
the war, returning soldiers and an overly fearful public helped rebrand chemical weapons as an
unfair method of warfare, resulting in the creation of an international norm proscribing the use of
these weapons (Price 1997).
35
The definition of what does/does not constitute a chemical weapon has been the subject of
intense debate since the beginning of the 1900s. In order to maintain a systematic definition of
what constitutes a chemical weapons and retain a focus on the use of chemicals utilized to poison
human beings, this study will use the definition of a chemical weapon found in the Chemical
Weapons Convention (CWC), which entered into force in 1997. Previous treaties banning the use
of chemical weapons neglected to identify exactly what qualifies as such weapons, and are
therefore unsuited to use for systematic study. Herbicides are specifically ignored in the CWC
definition for political purposes. The lack of inclusion of state use of herbicides in in this project
should not be interpreted as this author’s approval of the use of these compounds in conflict or the
states that are accused of their use. Additionally, the possession of some historical chemical
weapons (i.e. chlorine) are not controlled for by the CWC due to their usefulness in widely used
multi-industry processes. As nearly every state with industrial capabilities possess some stock of
these chemicals, and they have only been used in a handful of occasions since WWII, only the use
of these chemicals as weapons will be included in the database.

72

 States purchasing Schedule 1 substances or chemical weapons from
another state,
 State with an active chemical weapons development program or still
possess the leftovers or forgotten munitions from a mothballed program,
 States that do not declare Schedule 2 substances, and
 States that had chemical weapons abandoned on their territory.35F
36

States that were accused of using chemical weapons are de facto in the
dataset, as one must possess chemical weapons in order to use them. Additionally,
the years in which states used non-lethal chemical weapons in ways forbidden by
the Chemical Weapons Convention will also be included.36F
37
Not all countries will
have data entries in all years as states acquired chemical weapons at different
times. The data is segmented in five year increments, with the exception of 1918-
19. This segment constitutes its own data section. A five state-year increment was
chosen for the following reasons:
 The data is not coded by country (with a single entry for each state)
primarily in order to increase the number of observations of the data.
Additionally, there are multiple states that used chemical weapons in
several conflicts. By lumping all the events into a single observation, it
ignores the fact that a state might have different reasons to use chemical
weapons for each conflict. This would result in significant loss of data.

36
This is included under the assumption that if pressed, a state could retrofit these munitions in
order to use them in their own defense.
37
In order to identify the conditions under which states choose to use chemical weapons, the
dataset is being restricted to instances in which the orders to use chemical weapons come from the
upper echelons of the state military and/or political establishment. Specifically excluded are
instances in which there was an accidental release of chemical weapons (even during warfare) and
decisions made by local commanders to use chemical weapons without the knowledge of their
superiors.

73

 Due to the fact that conflicts in which chemical weapons are used often
span multiple years, the data is not coded by state-year. If the data were to
be coded by state-year, it would inappropriately assume that the decision
to use chemical weapons in any given year is independent from previous
usage. It is logical to assume that once the decision to use chemical
weapons in a given conflict has been made, it is not reassessed on a
calendar-year basis.
 Five year increments are long enough to include all years of most conflicts
within a single entry. This reduces the effect of the assumption above and
allows chemical weapons usage to be assessed in a holistic manner
regarding the course of the conflict.
 Segmenting the data by country-decade presents problems, as it has the
potential of condensing multiple conflicts and instances of chemical
weapons usage into a single data point. It would then inappropriately
confound cause and effect, as there could be different reasons for each
instance.
 In each instance, the assessed conditions examined for sufficiency for the
use of chemical weapons will be required to have occurred before the use
of these weapons. This is done in order to avoid ex post facto causation. In
the event that chemical weapons were not used during that period, the
conditions will be assessed on presence or absence.

The data is segmented into five year increments in an effort to include all
years of a conflict within a single entry. For those conflicts that encompass
multiple data segments, the segments will be coded individually. A list including
all the countries alleged to have used chemical weapons is included in Table 1.37F
38

Included in this table are the years in which it is alleged that chemical weapons
were used and by which state. The table also includes where the chemical

38
All Tables can be found in Appendix B.

74

weapons were allegedly used and in which conflict. Lastly, information about the
acceptance status in the literature is included. This indicates whether current
literature believes the chemical weapons to have occurred, not occurred, and
whether there is debate over the occurrence. Most of the entries are contested in
some manner.
Due to the nature of chemical weapons programs, however, there can be
significant uncertainty as to their exact dates of existence. In this instance, the
best effort will be made to estimate when the state was no longer in possession of
restricted substances. If there is no evidence that the termination of a well-
established chemical weapons program has been terminated, it will be assumed
that the program is still in existence and the state is still in possession of chemical
weapons.38F
39
In both cases, a note will be included in the file explaining the
uncertainty.

Case Summary Statistics In order to gain a better understanding of the cases
outlined above, some basic summary information was collected. In order to
identify whether or not the majority of cases of chemical weapons use were
conducted by the same state, a histogram was constructed. The number of times a

39
The assumption is that states would be unwilling to destroy materials for which they have paid a
significant sunk cost and receive nothing in return for their destruction.

75

state was accused of using chemical weapons was recorded in Figure 4. From this
graph, it is clear that a single state is not responsible for the majority of the
approximately 43 times chemical weapons were alleged to have been used. A
quarter of the instances in which chemical weapons were claimed to have been
used were perpetrated by states who were only accused of using them that one
time.
On average, a state that was accused of using chemical weapons did so in
1.8 instances. That state was also far more likely, on average, to use chemical
weapons outside the state than inside the state; states accused of using chemical
weapons used them outside of the state an average of 1.4 times and inside the
state 0.5 times. A little less than half of the states experienced chemical weapons
use at the hand of an outside power. A commonly asserted position explaining
why chemical weapons are not used is that states that have experienced chemical
weapons use are unlikely to impose that type of punishment on others. This does
not seem to be an accurate truism, as about half of the cases in the database
experienced chemical weapons and perpetrated it. Instead, a casual examination
of the data suggests a generational effect, with generations outside living memory
of chemical weapons use using these agents themselves. See Table 2 for a
summary of the breakdown of this information.

76

Further probing of the relationship between those suffering chemical weapons
use and those perpetrating it was conducted. A very basic linear regression was
run, focusing on both overall chemical weapons use and location of usage,
whether inside or outside the state. It was concluded that a relationship between
the two does not exist, as the null hypothesis was unable to be rejected in each
case. See Table 3 for summary statistics.
In order to further understand how frequently chemical weapons are used in
conflict by states, the number of instances in which chemical weapons were used
was compared to the number of conflicts the states participated in. The states
included in the dataset were participants in 177 conflicts meeting the inclusion
criteria for the Correlates of War (COW) Project. In 21.5% of these conflicts, a
state was accused of using chemical weapons.39F
40
Of those states that were accused
of using chemical weapons, they are accused of this activity in 37.8% of their
conflicts meeting the COW inclusion criteria. While not a commonly used type of
weapon, it does appear that chemical weapons have been utilized by states far
more frequently than is commonly assumed.
In order to understand whether or not a single region was responsible for
the majority of chemical weapons usage, a graph depicting chemical weapons use

40
For this calculation, both the recent Afghanistan and Iraq invasions and occupations were
considered a single conflict. Nine additional incidents of chemical weapons use were reported, but
did not occur in conjunction with a COW qualified conflict. As such, these incidents were not
included in the calculations.

77

by region was created. Figure 5 breaks down chemical weapons use by region and
compares it to how many times a state in that region was accused of using
chemical weapons. It also records how many times a state in that region suffered a
chemical weapons attack. States in Africa, Asia/Pacific, and the Middle East are
the three regions that have seen the most chemical weapons usage, with Africa
experiencing double the number of attacks than perpetrations. Contrastingly,
Europe has used chemical weapons far more than it has suffered their usage.
European countries have been accused of using chemical weapons 16 times since
WWI, and have only experienced a chemical attack on its soil 3 times.
Asia/Pacific and the Middle East have seen roughly the same number of users and
victims. The only region untouched by chemical weapons is South America; no
state on that continent has even been accused of developing, possessing, or using
these agents. The biggest take-away from Figure 5 is that there does not seem to
be an overwhelming story of a single region always using chemical weapons
against another region. While Europe is far more likely to be a chemical weapons
user, no region is entirely innocent, except for South America.
The split between chemical weapons user and sufferer is further
illustrated in Figure 6. This chart identifies the number of times a given state has
both had chemical weapons used against them and used chemical weapons against
others. As evidenced by this figure, there is some overlap between those who
have used and suffered chemical weapons usage (ex: China, Iraq). However, some

78

of the most egregious chemical agent users have not had chemical weapons used
against them since WWI. These include France, Spain, the USSR, and the United
Kingdom.
In addition to identifying which states are using chemical weapons,
another important piece of information to identify is whether or not these states
are using chemical weapons against their own people (in which it can be claimed
that chemical weapons use is an internal issue) or against opponents outside the
state (in which chemical weapons becomes an international issue). Figure 7
illustrates this division. A decent number of states have only been accused of
using chemical weapons within their borders (ex: China, Cuba, Greece, Israel,
Rhodesia, and Syria). The only three states that have been accused of using
chemical weapons both inside and outside their borders are Iraq, Spain, and the
USSR. The majority of the time, a state will be accused of using chemical
weapons inside its borders or outside exclusively.

How Can We Tell the Difference Between Users and Abstainers?
While the above summary statistics are informative, they do not reveal a
clear distinction between those who are alleged to have used chemical weapons
and those that have not. This calls for a more in-depth analysis utilizing fs/QCA.

79

fs/QCA operates based on principals found in set theory. While the notion
of a “set” is not very well known in social science methodology, a good part of
human conceptual reasoning is based on the implicit idea of sets (Mahoney 2010).
In this project, a set is understood in terms of concepts. These concepts are then
defined in terms of “boundaries that define zones of inclusion and exclusion” and
“[c]ases are measured according to their fit within the boundaries of the set”
(Mahoney 2010, 7, 2). In this way, sets operate as “data containers” and cases are
judged by whether they fit the concept or not (Sartori 1970, 1039).40F
41

Because the principles of a set are different from measurements used in
traditional political research, the numerical descriptors of set membership also
have different meanings. Instead of probabilities, the numbers associated fs/QCA
indicate membership within a given set.41F
42
In traditional, or crisp-set QCA, these
qualitative points are no membership in the set [0] and full membership in the set
[1]. fs/QCA requires the definition of three anchor points: full set membership [1],
full non-set membership [0], and indifference [0.5] (Schneider and Wagemann
2012, 28). The membership scale in fs/QCA can then be further divided in order

41
An alternative definition of “set” is also defined by Mahoney as “a mental representation of an
empirical property” in which cases are measured “according to the extent to which they are in
possession of the represented property” (Mahoney 2010, 2). This author agrees with the implicit
assertion by Schneider and Wagemann that the first definition is more appropriate (Schneider and
Wagemann 2012, 24).
42
Several scholars make the mistake of reading fuzzy set membership as probabilities. For more
on this discussion, see (Altman and Perez-Linan 2002, 91; Eliason and Stryker 2009, 102-146;
McNeill and Freiberger 1993, 185ff; Schneider and Wagemann 2012, 31; Zadeh 1995, 271-276).

80

to indicate additional degrees of uncertainty, should the occasion warrant.42F
43
For
the purposes of this project, the following membership scheme has been adopted:
full non-membership [0], mostly not a member [0.2], more not a member than a
member [0.4], more a member than not a member [0.6], mostly a member [0.8],
and full membership [1].
As discussed above, supersets will feature prominently in this analysis.
While there is no empirical observation correlating to a case membership in the
superset, membership can be constructed using the constituent conditions. For this
analysis, membership in a superset will be calculated by using the logical AND
function on the constituent conditions. fs/QCA analyses using both the superset
and subsets will be conducted in order to assess how well the superset functions
as an indicator of potential sufficiency.

Conditions Under Consideration The conditions investigated in this analysis have
been grouped together based upon their theoretically relevant superset properties,
for a total of three supersets. These supersets include Military Vulnerability (V),

43
The intervals between the fsQCA scores do not need to be identical. It is acceptable to use
membership scores such as [0], [0.4], [0.6], and [1] if theoretical considerations justify it. Fuzzy
scales should never be read as either a continuous or ordinal scale. Reading the scale as a
continuous scale downplays the qualitative difference between scores above and below [0.5],
which is the essential feature of fuzzy sets. Ordinal scales rank order their constituent parts; fuzzy
sets indicate differing degrees of belonging to the concept and moving between intervals means
different things depending on which side of [0.5] you are on (Schneider and Wagemann 2012, 29-
30).

81

Political Vulnerability (P), and Conflict Location (L). In addition to the three
supersets, there are also two outcome conditions (W, R).
When represented graphically, fs/QCA produces a scatterplot between the
origin and the maximum X and Y values of [1]. The membership of a given case
in the configuration of conditions is plotted along the X axis, while membership
in the outcome of interest is plotted along the Y axis. A constellation of
conditions is said to be sufficient to result in an outcome when all the values of
the cases lie on or above the line Y=X. This graphically represents the concept
that whenever a group of conditions is present, the outcome is also present (the
definition of sufficiency).43F
44
Perfect sufficiency results in all points lying on the
Y=X line, but is a phenomenon rarely seen outside artificially created datasets. A
grouping of conditions is said to be necessary for the outcome if the membership
scores in the conditions are greater than or equal to the outcome. This results in a
scatterplot with cases lying on or below the line Y=X. This graphically represents
the concept of whenever an outcome is present, the conditions of interest are also
present (the definition of necessity).44F
45
Scatterplots in which the cases are
distributed across the space are judged to represent constellations of conditions

44
Cases in the bottom-right quadrant are those in which the conditions occur and the outcome
does not, and is not allowed.
45
Cases in the upper-left quadrant are those in which the conditions did not occur and the outcome
did, and is not allowed.

82

that are neither necessary nor sufficient, as the outcome occurs without regard to
the presence or absence of the conditions.

Military Vulnerability (V) There are three components to Military Vulnerability.
The first subset, Regime Survival (T), posits that a perceived threat to regime
survival is sufficient for the state to use chemical weapons to neutralize the threat.
If the choice is either perish or persist, regimes will choose to use their chemical
weapons in order to persist. Analogues to this situation can be found in the
nuclear proliferation literature, and is at the core of realist perspectives on nuclear
weapons (Beaton and Maddox 1962; Dunn and Kahn 1976; Jo and Gartzke 2007,
167-194; Sagan 2011, 225-244).
In the Regime Survival condition, a case is scored for set membership
from [0] (the state does not perceive a threat to the regime) to [1] (a severe threat
to regime survival is perceived to exist). The set membership intervals between
full and no membership correspond to the following qualitative pegs: [0.2]
corresponds to the state perceiving an extremely minor threat to regime survival,
[0.4] corresponds to the state perceiving a minor threat to regime survival, [0.6]
corresponds to the state perceiving a moderate threat to regime survival, and [0.8]
corresponding to the regime perceiving a major threat to regime survival. For

83

more information on precisely on how these membership levels are coded, see
Table 4.
The Power Asymmetry (A) condition argues that if there is a large power
asymmetry during a conflict between combatants, the state on the losing end will
use their chemical weapons. If a country feels that it is vastly outmatched in either
conventional or unconventional weaponry, that state may use their chemical
weapons in order to correct the power differential and return to an equal footing
with its opponent. This is a traditional component of the nuclear proliferation
literature, and has often been cited as the reason for chemical weapons
proliferation in the Middle East (balancing against the Israeli nuclear threat)
(Bahgat 2005, 401-424; Jo and Gartzke 2007, 167-194; Potter 1982; Quester
2005).
For the subset Power Asymmetry condition, a case is scored based the
power differential between itself and its strongest component in a conflict, with
[1] corresponding to the state’s most powerful opponent being a nuclear power
while the state is not and [0] corresponding to the state its most powerful
opponent being roughly even in power, the state being the more powerful of the
two, the state being a nuclear power, or the state not being an armed conflict. The
varying levels of set membership between the maximal and minimal values
should be taken to mean as follows: [0.2] as the state being slightly less powerful

84

than its most powerful opponent or both states being nuclear powers, [0.4] as
neither state being a nuclear power, but the state’s opponent is slightly more
powerful, [0.6] as neither state being a nuclear power, but the state’s opponent is
clearly militarily superior, and [0.8] as neither state being a nuclear power, but the
state’s opponent has an overwhelming military superiority. For more information
on exactly how these membership scores will be coded, see Table 4.
The last Military Vulnerability subset, Stalemate (S), argues that if a
chemical weapons possessor state finds itself in a stalemate during a war that is
sufficient for the state to use its chemical arsenal to gain an advantage. Using
chemical weapons unexpectedly and on unsuspecting/unprepared forces can result
in brief periods when the utilizer has the upper hand. In order to manufacture such
a window of opportunity during a conflict, states may use their chemical weapons.
For Stalemate, a case is scored based on the state’s perception of its
forward progress in winning a conflict, with [1] corresponding to the state being
in an armed conflict, but there have been periods where forward progress has been
stalled for several months to years and [0] corresponding to the state not being in
an armed conflict. The varying levels of set membership between the maximal
and minimal values should be taken to mean as follows: [0.2] as the state being in
an armed conflict in which it has made constant progress towards winning, [0.4]
as the state being in an armed conflict, and while the state is mostly winning,

85

there are short periods of stalling, [0.6] as the state being in an armed conflict, and
forward progress in winning stalls for up to two weeks at a time, and [0.8] as the
state being in an armed conflict, and forward progress in winning stalls for up to a
month at a time. For more information on exactly how these membership scores
will be coded, see Table 4.

Political Vulnerability (P) There are three components to Political Vulnerability.
The first condition, Credible Commitment (C) argues that states that lack a
credible nuclear security commitment from a more powerful state is sufficient for
a state to use chemical weapons in a conflict. If a state is threatened and has no
powerful ally to rely upon, the state is more likely to use its chemical arsenal to
guarantee its security. This logic is echoed in the nuclear proliferation literature,
and is often cited as a compelling reason to extend a “nuclear umbrella” in order
to prevent the spread of nuclear weapons to countries like South Korea (Jo and
Gartzke 2007, 167-194).
For the subset Credible Commitment condition, a case is scored based the
strength of its alliances while it is in a conflict, with [1] corresponding to the state
lacking an alliance with any country and lacking a nuclear program [0]
corresponding to the state having a nuclear defense pact with a stronger state, the
state being a nuclear power or a regional hegemon, or the state not being in an

86

armed conflict. The varying levels of set membership between the maximal and
minimal values should be taken to mean as follows: [0.2] as the state having a
defense pact in place with a major power or the state being a near nuclear power,
[0.4] as the state possessing a non-aggression treaty with states not in the conflict
or the state having a small nuclear program, [0.6] as the state having an entente
understanding with states not in the conflict and the state not having a nuclear
program, and [0.8] as the state having a neutrality agreement with states not in the
conflict and the state not having a nuclear program. For more information on
exactly how these membership scores will be coded, see Table 5.
The Ostracization (O) condition claims that states that are either ignored
or ostracized by those in power are more likely to use their chemical weapons in a
conflict because there is little reason not to. Allies cannot restrain this behavior by
definition and there is little reason to adhere to the social norm prohibiting its use
as no one is watching to enforce it. In the nuclear proliferation literature,
acquiring nuclear weapons also serves to signal the regime’s vitality and power as
a secondary function, although this is less applicable to chemical weapons (Betts
1977, 157-183; Jo and Gartzke 2007, 167-194; Quester 1973; Rosen 1975, 157-
184).45F
46

46
This is due to the fact that chemical weapons are seen as “dirty” and illegitimate means of
warfare.

87

In the Ostracization condition, a case is scored for set membership from
[0] (the state is a major player in the international system) to [1] (state is
ostracized by the international community, has little trade or diplomatic contact).
The set membership intervals between full and no membership correspond to the
following qualitative pegs: [0.2] corresponds to the state being a moderate player
in the international system and generally considered a contributing/productive
member of the system, [0.4] corresponds to the state being a small player in the
international system and is generally considered unimportant, [0.6] corresponds to
the state being on the fringes of the international economic system and generally
considered insolvent or unimportant, and [0.8] corresponding to the state being at
the fringes of the international economy and generally considered problematic or
engaged in bad behavior. For more information on precisely on how these
membership levels are coded, see Table 5.
The last subset condition of Political Vulnerability, Democratic Regime
(D), posits that a state possessing a democratic government is sufficient for a state
to use their chemical weapons. Studies into nuclear proliferation have discovered
that democracies are slightly more likely than anocracies to acquire nuclear
weapons. This could be due to the fact that democracies are more vulnerable
internal pressure to nuclearize, particularly if it promises increased popularity
among nationalist populations (Fuhrmann 2009, 36; Jo and Gartzke 2007, 179;
Kroenig 2009, 172; Sagan 2011, 237; Singh and Way 2004, 864).

88

For Democratic Regime, a case is scored based on its perception as a
democracy, with [1] corresponding to the state being a consolidated democracy
and [0] corresponding to the state being a hereditary monarchy. The varying
levels of set membership between the maximal and minimal values should be
taken to mean as follows: [0.2] as the state being considered an autocracy, [0.4] as
the state being considered a more authoritarian anocracy, [0.6] as the state being
considered a more democratic anocracy, and [0.8] as the state being considered a
democracy. For more information on exactly how these membership scores will
be coded, see Table 5.

Conflict Location (L) There are two components to Conflict Location. The first
subset, Foreign Territory (E), posits that state participation in a conflict outside its
home territory is sufficient to allow the state to use its chemical weapons against
its opponent. The use of a chemical agent results in widespread contamination of
both objects and earth for extended periods of time. This could result in a
backlash from one’s citizenry and a decrease in economic activity as their homes
and property become inaccessible. However, chemical contamination in another
country’s territory will not affect one’s citizens, meaning that states are more
likely to use chemical weapons when they are not fighting in their own territory.

89

For the Foreign Territory condition, a case is scored for set membership
from [0] (state is not in an armed conflict) to [1] (state is involved in an armed
conflict in which all the fighting is in another state). The set membership intervals
between full and no membership correspond to the following qualitative pegs:
[0.2] corresponds to the state being involved in an armed conflict in which the
occasional battle is fought outside the pre-war boundaries of the state, [0.4]
corresponds to the state being involved in an armed conflict in which a small
number of battles are fought outside the pre-war boundaries of the state, [0.6]
corresponds to the state being involved in an armed conflict in which the fighting
is conducted half in another state and half within the state’s prewar boundaries,
and [0.8] corresponding to the state being involved in an armed conflict in which
the fighting is mostly conducted in an outside state. For more information on
precisely on how these membership levels are coded, see Table 6.
The second Conflict Location subset, Global South (G), proposes that
fighting in a conflict in the Global South is sufficient for a state to use chemical
weapons. States in the Global North wield the majority of the economic and
military power internationally and are primarily concerned with each other’s
foreign policies. If a state (including those in the North) were to use chemical
weapons in the Global South it would most likely be ignored; there is little
political or financial gain from raising a ruckus about it. Therefore, states in the
Global South are much more likely to see chemical warfare on their soil.

90

For Global South, a case is scored based on its perception as a democracy,
with [1] corresponding to the state being involved in an armed conflict in a
country considered to be in the Global South or 3
rd
World, and it is a relatively
unimportant state and [0] corresponding to the state not being involved in an
armed conflict. The varying levels of set membership between the maximal and
minimal values should be taken to mean as follows: [0.2] as the state being
involved in an armed conflict in a country considered to be in the 1
st
or 2
nd
World,
Global North, a permanent member of the UNSC, or in a wealthy state, [0.4] as
the state being involved in an armed conflict in a non-aligned country, a non-
permanent member of the UNSC, or relatively moderately wealthy state, [0.6] as
the state being involved in an armed conflict in a country considered to be in the
Global South or 3
rd
World which is a member of OPEC, and [0.8] as the state
being involved in an armed conflict in a country considered to be in the Global
South or 3
rd
World which is not a member of OPEC. For more information on
exactly how these membership scores will be coded, see Table 6.

Outcome Conditions (W, R) The outcome conditions assess the ability of a case to
be classified as a member of the outcome of interest. The first outcome, Chemical
Weapons Use (W), is the set of cases in which a state is believed by the literature
to have used chemical weapons. A case is scored on that basis, with [0]

91

corresponding to no accusation of state chemical weapons usage and [1]
corresponding to the state being believed to have used chemical agents beyond
any doubt. The varying levels of set membership should be taken to mean as
follows: [0.2] as there being accusations of state chemical weapons use, but there
is little to no consensus about the veracity of the claim or those chemical weapons
allegations that are later proven to be false, [0.4] as there being accusations of
chemical weapons use but there is no consensus in the literature over whether the
events in fact occurred, [0.6] as there being accusations of chemical agent use and
international attempts to verify use that judge it was more likely than not, and
[0.8] as there being accusations of chemical weapons use and it is generally
accepted that chemical weapons have been used. For more information on exactly
how these membership scores will be coded, see Table 7.
The second outcome of interest Non-Lethal First (R) is the set of cases in
which the literature believes that the state used non-lethal chemical agents and
then progressed to using lethal chemical weapons. This is an important distinction
historically, there have been countries that were only accused of using non-lethal
chemicals as weapons and those that were accused of using both lethal and non-
lethal chemicals. The use of non-lethal chemicals as weapons either constituted a
grey area in international law or were permitted throughout much of the period
after 1918. Why states progressed from a using non-lethal chemicals to lethal
chemicals is not understood. Since 1992, the use of riot control agents is legal for

92

internal policing only under the Chemical Weapons Convention; any use in a
militarized dispute is illegal (United Nations General Assembly 1992). If the risk
factors associated with a transition from non-lethal to lethal chemical weapons
could be identified, the use of non-lethal agents could serve as a warning of bad
things to come, potentially giving the international community the time it needs to
head off future chemical weapons use.
A case is scored for set membership from [0] (no accusation of state
chemical weapons usage) to [1] (state is believed to have used non-lethal and then
lethal chemical weapons, beyond a doubt). The set membership intervals between
full and no membership correspond to the following qualitative pegs: [0.2]
corresponds to there being accusations of non-lethal chemical weapons use, but
there is little consensus on the use of non-lethal chemical weapons and former
accusations that were later proven to not have occurred, [0.4] corresponds to there
being accusations of non-lethal chemical weapons use that are generally accepted
by the international community to have occurred, [0.6] corresponds to there being
accusations of lethal and non-lethal chemical weapons use, but while it is
generally accepted that non-lethal usage has occurred there is a lesser consensus
on the use of lethal chemical weapons, and [0.8] corresponding to accusations of
non-lethal and lethal chemical weapons use and it is generally accepted that both
lethal and non-lethal chemical weapons have been used. For more information on
precisely on how these membership levels are coded, see Table 7.

93

Assessing the Driving Cases
In addition to assessing which risk factors are associated with chemical
weapons usage, an understanding of the which cases are driving the phenomenon
needs to be developed. To do so, the configurations of risk factors identified by
fs/QCA as being sufficient for chemical weapons use to occur will be used. Using
epidemiological methods, the risk will be assessed in the following manner.
Attack ratios for the various condition configurations will be calculated.
This ratio assesses the proportion of an outcome that can be attributed to a given
condition (or cluster of conditions). Using this method, it is possible to evaluate
which clusters of conditions contribute more to the risk of using chemical
weapons than others, teasing out the condition constellations of most import.
𝐴𝑡𝑡𝑎𝑐𝑘 𝑅𝑎𝑡𝑖𝑜 =
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑐𝑎𝑠𝑒𝑠 𝑜𝑓 𝑠𝑡𝑎𝑡𝑒 𝑢𝑠𝑒 𝑜𝑓 𝑐 ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝑤𝑒𝑎𝑝𝑜𝑛𝑠 𝑓𝑜𝑟 𝑎 𝑔𝑖𝑣𝑒𝑛 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝑐𝑜𝑛𝑓𝑖𝑔𝑢𝑟𝑎𝑡𝑖𝑜𝑛 𝑡𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑡𝑎𝑡𝑒𝑠 𝑡 ℎ𝑎𝑡 𝑝𝑜𝑠𝑠𝑒𝑠𝑠 𝑐 ℎ𝑒𝑚𝑖𝑐𝑎𝑙 𝑤𝑒𝑎𝑝𝑜𝑛𝑠

The calculation of an attack ratio for the clusters of conditions associated with
chemical weapons use is possible because these conditions are similar to many
instances of acute infectious disease outbreaks. In these outbreak investigations,
the source of the infection is discovered using attack rates and attack ratios. Once
the attack ratio for all the condition combinations are calculated, they are divided
by the lowest present attack ratio. The highest comparative attack ratio(s) for a

94

given variable configuration is the one considered mostly responsible for the
usage of chemical weapons (Hennekens and Buring 1987).

Conclusion
This chapter presented the methodology behind how this fs/QCA study
will investigate the state use of chemical weapons. First, a discussion of QCA’s
background and unique analytical style was presented. Its prioritization of
multiple conjectural causation, equifinality, and causal asymmetry make this
method uniquely suited to examining the state use of chemical weapons. This is
doubly so when the high degree of uncertainty surrounding chemical weapons
programs and chemical weapons usage is factored into the decision on which
method to select.
This chapter also introduced an innovation to current fs/QCA
methodology to allow for the screening of many different potentially sufficient
conditions without exacerbating the problem of limited diversity. This new tool is
especially useful for pilot studies, or those research projects in which the true
drivers of phenomena are unclear. By using the Boolean AND function to create
artificial supersets of theoretically linked concepts, researcher can screen rafts of
conditions for those who promise to be sufficient in later analysis.

95

Prior to conducting the fs/QCA study, this chapter presented the coding
scheme for inclusion in the database as well as fs/QCA calibration. Summary
statistics were provided for those states included in the dataset, demonstrating that
a more systematic examination with fs/QCA was needed. The coding scheme for
this fs/QCA analysis was presented, including the individual conditions, their
supersets, and details on calibration.
Lastly, the method used to assess which cases drive the fs/QCA results
was introduced. Borrowed from epidemiology, the comparison of attack ratios
promises to help identify problems with fs/QCA analyses, find important cases for
later case study, and detect configurations of cases that might be at higher risk of
using chemical weapons due to their nature as significant cases.

96

Chapter 4
Expectations and Surprises

97

Understanding why states choose to use and refrain from using their
chemical weapons is an important task for international politics, one that has yet
to be undertaken. In order to fill this gap, this chapter has two purposes. Firstly,
this chapter will undertake an fs/QCA analysis of the hypothesized conditions
associated with chemical weapons use and abstention between 1918 and 2014. An
additional analysis will be undertaken to understand a special case of chemical
weapons use: the state decision to change from using non-lethal chemical
weapons to a more lethal variety. The second purpose of this chapter is to assess a
novel contribution to fs/QCA methodology. Taken in concert with the standard
analysis, this assessment will evaluate the use of supersets as indicators of later
subset sufficiency in fs/QCA analysis in order to combat methodological
problems facing studies in which the potential causes are unknown or
understudied.
To accomplish these goals, this chapter is laid out in the following
manner. First, an fs/QCA analysis of the conditions underlying the use of
chemical weapons from 1918-2014 will be conducted, paying particular attention
to test for necessary conditions as well as sufficient ones. Separate analyses of
different segments of the dataset will be conducted, as preliminary fs/QCA testing
revealed significant problems within the dataset. Next, an fs/QCA analysis of the
conditions underlying the non-use of chemical weapons will be undertaken,
paying particular attention to the split in the database identified in previous

98

fs/QCA testing. Attention will be paid to identifying necessary conditions as well
as sufficient ones. Thirdly, an analysis of which cases are driving the phenomena
of both chemical weapons use and abstention will be undertaken, incorporating
methodology adapted for fs/QCA from epidemiology. Following this discussion,
an analysis of the performance of supersets as predictors of subset sufficiency will
be conducted as a proof of concept test. Special attention will be given to how
well the supersets predict the sufficiency of the absence and presence of
conditions, particularly when both are present in the fs/QCA analysis. Lastly,
overarching themes identified by the fs/QCA analyses and superset testing will be
discussed.

Is The Use of Chemical Weapons Consistent from 1918 To 2014?
Table 8 contains the fuzzy set scores for the two outcomes and eight
conditions for the states that have been accused of using chemical weapons.46F
47

Results are presented in equation form, with letter abbreviations for both the
outcomes and conditions. A complete listing of each abbreviation and their
corresponding condition is included in Table 8. Capital letters indicate the
presence of a condition, while lower case letters indicate an absence (i.e. T

47
It should be noted that although accusations of chemical weapons use occurred in 43 instances,
the dataset contains 54 cases of chemical weapons use. This is due to the fact that use in some
cases spread over two or more date “bins.”

99

indicates a high challenge to regime survival and t indicates a low-to-no challenge
to regime survival).47F
48
Three sets of conditions later found to be significant for
either non-use or use of chemical weapons were also included in Table 8 as
examples.
The results of this fs/QCA analysis will be reported using both graphical
representation and descriptive measures such as consistency and coverage scores
(Ragin 2006, 291-310). Consistency evaluates the degree to which a subset
relation has been approximated, while coverage evaluates the empirical relevance
of a consistent subset (Ragin 2006, 291-310). In other words, consistency
measures the degree to which a condition (or combination of conditions) is
associated with the outcome of interest, and is roughly analogous to the idea of
significance in statistical modeling (Thiem 2010, 6). In cases of perfect
consistency, all cases in which condition X appear will also have the outcome of
interest (i.e. all cases of pregnant humans will have a sex “outcome” of female).
As most conditions in fsQCA studies are only consistent to a limited degree,
consistency is rated within a range between [1] (perfectly consistent) and [0]
(perfectly inconsistent), with higher scores indicating that a condition is
associated with the outcome a greater percentage of the time.

48
A more detailed scheme on how these conditions were coded can be found in Appendix A.

100

Coverage is an fsQCA measure that evaluates the empirical relevance of a
given set of conditions to the overall outcome, and is somewhat equivalent to the
R
2
measure in statistical models (Legewie 2013; Thiem 2010, 6). Coverage
evaluates empirical relevance by evaluating the degree to which set membership
in a condition covers the cases’ set membership in the outcome, and is scored
between [0] and [1] (Ragin 2006, 292).48F
49
High coverage points to a non-trivial
condition regarding the outcome, whereas a low coverage score can indicate that
that condition is relatively unimportant to establishing overall sufficiency (Thiem
2010, 5-6). Coverage will only be interpreted for consistent results. Coverage for
inconsistent results in not considered a meaningful indicator of empirical
relevance (Schneider and Wagemann 2007, 209). Consistency and coverage are
calculated using the software fs/QCA 2.5 (Ragin, Drass, and Davey 2009).

Necessary conditions for the outcomes ‘chemical weapons use’ and ‘non-lethal
weapons first’ Schneider and Wagemann recommend that conditions only be
considered necessary if their consistency scores are very high (Schneider and
Wagemann 2007). In order to make ensure that different conditions are truly
necessary for the outcome, a threshold consistency of 0.90 for accepting a

49
“Solution coverage” is used to refer to the overall degree to which set membership in all of the
component conditions covers the cases’ set membership in the overall fsQCA solution and is
reported separately, along with the overall consistency of the solution.

101

condition as necessary was selected. Table 9 contains the results of an analysis of
the necessary conditions for the outcome ‘chemical weapons use.’ As can be seen
in Table 9, all consistency scores are far below the 0.90 threshold, indicating that
none of the conditions are in themselves necessary for states to use chemical
weapons. Table 10 contains the results of an analysis of the necessary conditions
for the outcome ‘non-lethal weapons first.’ As with Table 9, none of the
consistency scores reach the 0.90 threshold.

Sufficient conditions for the outcomes ‘chemical weapons use’ and ‘non-lethal
weapons first’ Table 11 contains the results of the analysis of the sufficient
conditions for the outcome ‘chemical weapons use.’ Both the consistency (0.500)
and coverage (0.204) values are low.49F
50
This indicates that the solution only covers
a small proportion of the cases and there is significant inconsistency in the
outcome of the condition combinations. Since they are low and do not meet a
minimum sufficiency of 0.70, these results will not be interpreted.

50
For this, and each subsequent analysis, the intermediate solution was used unless noted. There
are three solutions generated in each fsQCA analysis: a conservative solution in which no
assumptions about logical remainders are made, an intermediate solution in which only easy
counterfactuals are used, and the most parsimonious solution in which all simplifying assumptions
are used. The conservative solution did not reduce the solution sufficiently and the most
parsimonious solution resulted in, this author’s opinion, overly simplified and insufficiently
nuanced results.

102

Figure 8 presents the results graphically. For a condition or a combination
of conditions to be considered sufficient, all cases should be located around or
above the bisecting line (Ragin 2000, 236). This result confirms the significant
variation in outcome indicated in the numerical results. Instead of clustering
above the bisecting line, there are a fair number of cases in the lower right corner,
which are severe challenges to the supposition that the conditions T*a*S*C*G +
T*S*C*d*G are sufficient to result in chemical weapons use. An additional
challenge to the validity of the results arises when the two sets of conditions are
examined. Instead of representing actual cases, these constituent conditions are
actually logical remainders- lines on the truth table without an actual case
assigned to it. This is an indicator that the results should not be trusted.
Table 12 contains the results of the analysis of the sufficient conditions for
the outcome ‘non-lethal weapons first.’ Both the consistency (0.470) and
coverage (0.261) values are very low, although the coverage is slightly better than
for the outcome ‘chemical weapons use.’ As with Table 11, this indicates that the
solution also only covers a small proportion of the cases and there is also
significant inconsistency in the outcome of the condition combinations. Figure 9
presents the results of graphically. As with the results of in Figure 8, this result
confirms the significant variation in outcome indicated in Table 19. There are a
fair number of cases in the lower right corner, and only a few cases in the middle
have changed from Figure 8.

103

The lack of adequate sufficient results for the use of chemical weapons
and the use of non-lethal weapons from 1918 to 2014 indicates that there is at
least one underlying problem with the distribution of cases across the potential
combinations of conditions. One method to identify if this is the case is to stratify
the data. From an inspection of the cases, it appears that the profile of what state
uses chemical weapons changes after WWII. Prior to this event, the majority of
states that used chemical weapons reside in Europe. These states tended to use
chemical weapons outside Europe and North America. After WWII, the pattern
shifts. States outside of Europe were primarily responsible for chemical weapons
use, although the United States used chemical weapons in two conflicts.
While more research is needed to determine why this shift occurred, it is
likely that it had to do with the newest innovation in weapons of mass destruction:
the nuclear warhead. The power of nuclear weapons was introduced at the end of
WWII in the Pacific theater. Never before had a country possessed the ability to
literally wipe an entire city off the map with a single warhead. Neither chemical
nor biological weapons are designed to do that. This opened up a new realm of
possibilities to the military planner and political decision maker. Nuclear weapons
became the newest, most “sexy” toy for the military arsenal, and older
technologies were sidelined in order to provide for nuclear funding and
development. It is unlikely that it is a coincidence that the members of the core

104

abandoned their use (for the most part) of chemical weapons, leaving non-nuclear
weapons states to use chemical weapons after the Second World War.

1918-1945 Dataset: Necessary and Sufficient Conditions for ‘chemical weapons
use’ and ‘non-lethal weapons first’ For the next set of analyses, the data set was
split in two: all cases prior to 1945 constitute the “1918-1945 Dataset” and those
occurring after constitute the “1945-2014 Dataset.”50F
51
The same fs/QCA analyses
previously conducted were run for each dataset.
Table 13 contains the results of an analysis of the necessary conditions for
the outcome ‘chemical weapons use’ in the 1918-1945 dataset. As can be seen in
Table 13, all consistency scores are far below the 0.90 threshold, indicating that
none of the conditions are in themselves individually necessary for states to use
chemical weapons. The use of chemical weapons in the Global South is close to
being necessary, but does not quite meet the requirements.
An analysis for the necessity of different combinations of conditions can
also be conducted. However, these analyses should only be performed if there are
good theoretical arguments for combining the conditions (Schneider and

51
There is one exception to this rule: Japan 1945. Since Japan was still engaged in using chemical
weapons in China during WWII, the decision was made to keep that case with the rest of the
WWII Japanese cases in the 1918-1945 dataset.

105

Wagemann 2007, 59). A good theoretical argument applies for combining the
degree of threat to the state and with the location in which a conflict is fought.
Use of chemical weapons can be expected (according to the hypotheses) by states
that are experiencing severe challenges to the viability of the regime (T and A) in
places of the world that are relatively unimportant (G). The state experiences a
high incentive to act with the expectation that there will be little push back from
members of the international system if it does so. As Table 13 reveals, this is in
fact the case. The consistency score in the case of T+A+G is 1.000, making this
combination of conditions a necessary condition for the use of chemical weapons.
Figure 10 displays this relationship graphically. In order to consider a
condition as necessary for the outcome, all cases need to be located around or
below the bisecting line (Ragin 2000, 215). In Figure 10, all cases lie either on the
bisecting line or along the x and y axis. This means that while a severe challenge
to regime viability in the Global South is necessary for states to use chemical
weapons, it is not itself sufficient. Some countries have high levels of challenge to
the regime in the Global South but still choose not to use their weapons (bottom
right corner).
Table 14 contains the results of an analysis of the necessary conditions for
the outcome ‘non-lethal weapons first.’ While most of the conditions do not meet
the requirements for necessity, the condition Global South (G) reaches necessary

106

consistency levels (0.906). For the time period 1918-1945, a conflict in the Global
South was necessary for states to escalate their use of chemical weapons from
non-lethal to lethal. This corroborates observations of the historical record.
During the interwar period, Italy and Japan extensively tested various delivery
systems for both lethal and non-lethal agents (Del Boca 1969; Hilmas, Smart, and
Hill 2008, 9-76; Rochat 2005, 37-46; Sbacchi 2005, 47-56; Spiers 2010;
Stockholm International Peace Research Institute 1971).
Table 15 contains the results of the analysis of the sufficient conditions for
the outcome ‘chemical weapons use’ for the 1918-1945 dataset. The consistency
of the solution (0.727) meets the threshold for sufficiency, although the solution
coverage is low (0.229). This indicates that while the conditions identified below
consistently result in the use of chemical weapons, it only applies to a small
segment of the cases. In addition, only two components of the solution
(T*A*S*c*d*G and T*A*C*D*E*G) meet sufficiency conditions. These
combinations of conditions correspond to: a) non-democratic states with security
commitments from a more powerful state that perceive severe challenges to the
regime by a more powerful opponent in the Global South and b) democratic states
without security commitments that perceive severe challenges to the regime by a
more powerful opponent fighting outside their territorial borders in the Global
South are sufficient for the state to use chemical weapons.

107

The first solution component (T*A*S*c*d*G) describes the Soviet
Union’s use of chemical weapons in 1921.51F
52
While it is alleged that the
Bolsheviks used chemical weapons on three occasions to suppress uprisings, only
the incident in the Tambov region has been substantiated. Rebel bands were
hiding in forested areas, and the army used poison gas to kill everything hiding
there (Fedorov 1994; Johnson, Larsen, and Meek 2015, 7-16; Pozdnyakov 1956;
Spiers 2010; Stockholm International Peace Research Institute 1971; Waitt 1922;
Werth 1999, 108-131). The second solution (T*A*C*D*E*G) describes the
Japanese use of chemical weapons in the Second Sino-Japanese War from 1937-
1939.52F
53
Japanese army gas units were sent to Manchuria in 1937, and initial
attacks used non-lethal gases. Attacks using chemical weapons quickly increased
in 1938, with lethal gases being deployed against the underprepared Chinese in
support of ground operations. Despite evidence presented to the League of
Nations by the Chinese, nothing of substance was done to halt the Japanese use of
chemical weapons (Anonymous 1932, 361-372; Anonymous 1937a; Anonymous
1937b; Anonymous 1937c; Anonymous 1937d; Anonymous 1937e; Anonymous
1938, 378-380; Anonymous 1938, 878-881; Anonymous 1938a, 863-865;
Anonymous 1938a; Anonymous 1938b, 863-865; Anonymous 1938b;
Anonymous 1942a; Anonymous 1942b; Anonymous 1942c; Anonymous 1943a;

52
Asymmetry is scored highly in this case because the United Kingdom was the USSR’s opponent
in the Civil War at this time.
53
Japanese use of chemical weapons continued until their defeat in 1945.

108

Anonymous 1943b; 5250th Technical Intelligence Company 1946; Chi-tsai 1938,
666; Chi-tsai 1938, 667-668; Chi-tsai 1938, 665-666; Hilmas, Smart, and Hill
2008, 9-76; Joy 1997, 87-109; Spiers 2010, 76-77; Stockholm International Peace
Research Institute 1971).
Several things are of interest in this solution. Firstly, the regime survival
(T), power asymmetry (A), and Global South (G) conditions occur in all of the
components. This corroborates the finding that T+A+G was necessary in Table
13. Secondly, as with the overall analysis, the degree of state ostracization (O)
does not appear in any component of the solution. This indicates that this
condition is irrelevant for the outcome of interest, at least in the interwar period.
Thirdly, the directionality of the conditions only partially meets the expectations
put forth by the hypotheses. According to the hypotheses, the presence of these
conditions (upper case letters) is associated with chemical weapons use. Two of
the component conditions meet this expectation (T*A*C*D*E*G and
T*A*S*D*E*G). The other two components, T*A*S*c*d*G and T*A*S*c*E*G,
incorporate the absence of those conditions.
Table 16 contains the results of the analysis of the sufficient conditions for
the outcome ‘non-lethal weapons first’ for the 1918-1945 dataset. The consistency
of the solution (0.714) barely meets the threshold for sufficiency, although the
solution coverage is very low (0.189). Of the two components of the solution,

109

only one meets sufficiency requirements (consistency 0.800): a democratic
country without security commitments that perceives a threat to the regime by a
stronger power in a conflict not in its territory in the Global South is sufficient for
the state to escalate its use of chemical weapons from non-lethal to lethal. A
graphical representation of the results is included in Figure 11. Both conditions
represent the Japanese use of chemical weapons against China, first in the Second
Sino-Japanese War and then in World War II.
Firstly, the Global South (G) condition occurs in both components of the
solution, corroborating the finding that G was necessary in Table 14. Secondly, as
with the analysis of chemical weapons use in the 1918-1945 dataset, the degree of
ostracization in the international system seems to be irrelevant. It is not included
in either of the solution components. Lastly, unlike the above analysis, this
solution is consistent with the expectation put forth in the hypotheses. Both
components indicate that the presence of the conditions is associated with the
outcome.
Compared to the fs/QCA analysis using the entirety of the data, the results
of the fs/QCA analyses of the 1918-1945 yield interesting results. Firstly, none of
the conditions identified by the aggregate analysis are found in the stratified
results. This indicates that the cases after the Second World War look different
than those in the interwar period. This strengthens the cases that the dataset

110

should be split along those lines. Secondly, as with the aggregate fs/QCA
analysis, the use of chemical weapons and escalation of use look very similar in
this time period. The two solution components found in the ‘non-lethal weapons
first’ analysis are also components in the ‘chemical weapons use’ solution. This
indicates that fs/QCA cannot differentiate between the different conditions that
lead on one hand to the use of chemical weapons in general and those states that
start with non-lethal weapons and then escalate to lethal. Exactly why this is the
case shall be explained later in Tables 37 and 39.
Now that fs/QCA analyses have been run for the use of chemical weapons
and escalation of chemical weapons use in the 1918-1945 dataset has been
conducted, the same set of analyses was run for the post-WWII period.

1945-2014 Dataset: Necessary and Sufficient Conditions for ‘chemical weapons
use’ and ‘non-lethal weapons first’ An evaluation of the necessity and sufficiency
of different conditions to result in chemical weapons use and escalation from non-
lethal to lethal chemical weapons use between 1945 and 2014 was conducted.
Table 17 contains the results of an analysis of the necessary conditions for the
outcome ‘chemical weapons use.’ As can be seen in Table 17, all consistency
scores are far below the 0.90 threshold, indicating that none of the conditions are
in themselves individually necessary for states to use chemical weapons. The use

111

of chemical weapons when the regime perceives a threat to its survival is close to
being necessary, but does not quite meet the requirements. Table 18 contains the
results of an analysis of the necessary conditions for the outcome ‘non-lethal
weapons first’ in the 1945-2014 dataset. As with the use of chemical weapons,
none of the conditions meet the threshold to be considered necessary conditions.
Perception of threat to regime survival also approaches being considered a
necessary condition, but falls short of the 0.90 consistency requirement.
Table 19 contains the results of the analysis of the sufficient conditions for
the outcome ‘chemical weapons use’ in the 1945-2014 dataset. The consistency of
the solution (0.786) meets the threshold for sufficiency, while the solution
coverage is very low (0.120).53F
54
This solution corresponds to systemically
important non-democratic states without security commitments that perceive
threats to their regime by an equally powerful opponent in a stalemated conflict
within their territory in the Global South. This solution represents the case of Iran
during the Iran-Iraq War. There are some reports suggesting that during this
period Iran employed chemical agents on a limited scale between 1984 and 1988.
It is believed that initial attacks consisted of firing captured Iraqi munitions at
Iraqi forces. By the end of the war, however, Iran reported employed domestically
produced chemical munitions. A review of open source literature has failed to

54
The complex solution is the only significant solution for chemical weapons use after 1945. Both
parsimonious and intermediate solutions fall far below the 0.70 consistency requirement.

112

conclusively verify these claims, however (Anonymous 1991b; Anonymous
1991a, 26-27; Ali 2001, 47-48; Black and Pearson 1993, 584-587; Black et al.
1994, 301-321; Burck and Flowerree 1991, 237-245; Center for Nonproliferation
Studies. Monterey Institute of International Studies 2015b; Eisenstadt 1998, 2;
Herby 1992, 25; Hilmas, Smart, and Hill 2008, 78; Hogendoorn 1997, 37;
Robinson and Goldblat 1984; Tucker 2006, 251-253; United States of America.
Department of Defense January 2001).
Table 20 contains the results of the analysis of the sufficient conditions for
the outcome ‘non-lethal weapons first.’ The consistency of the solution (0.643)
does not meet the threshold for sufficiency and the solution coverage is very low
(0.136).54F
55
Since this solution does not meet the requirements for sufficiency, there
is an essentially null finding for this fs/QCA analysis and the results will not be
interpreted.
These fs/QCA analysis indicate that there is a significant problem lurking
in the 1945-2014 dataset. The difficulty is that almost all truth table rows have
low consistency scores; in other words, for each row, there are significant
numbers of cases that exhibit the outcome and significant numbers of cases that

55
This is the only significant solution for an escalation from non-lethal to lethal chemical weapons
use after 1945. Both parsimonious and intermediate solutions fall far below the 0.70 consistency
requirement.

113

do not. The only truth table row that has sufficient consistency contains both
Iranian cases from the Iran-Iraq War. This could be due to two possible reasons.
Firstly, it is possible that the hypotheses proposed in this project do not
capture the underlying condition(s) that lead to chemical weapons use after the
Second World War. States could use chemical weapons for reasons not
extrapolated from nuclear theory; perhaps the role of military advisors in the
upper echelons of power play a larger role than expected. Perhaps the timing of
chemical weapons usage matters; perhaps countries learned about how to get
away with chemical weapons use by observing the cases that came before them
and altered their strategies based on previous success and current demands.
Additional work needs to be done to identify if in fact there are outside factors
that played a significant role in allowing states to use chemical weapons.
Secondly, the reason consistency and coverage scores for chemical
weapons use were so low could be due to the individuality of each state decision
to use chemical weapons. Instead of some common underlying factors or
tendencies, states could choose to use their arsenals for any of a myriad of
reasons, each heavily dependent on context. The particular mix of factors in each
state decision could be wildly different than the same decisions made by other
states. More work needs to be done to closely look at the conditions under which
states in this time period chose to use their chemical arsenals. Only through

114

careful case study could it become clear if conditions were a major deciding
factor in the state use of chemical weapons.

Differing results in the two datasets: what does this mean? Comparing the fs/QCA
results across the two datasets yields one major observation. It is abundantly clear
that there are consistent fs/QCA solutions for the 1918-1945 dataset and not for
the 1945-2014 dataset. Something is present in the 1945-2014 dataset that was not
present in the interwar period. Based on the historical record, the presence of
international bans on the use of chemical weapons cannot be a factor, as the same
ban (namely the Geneva Protocol) was present in both time periods. The
Chemical Weapons Convention was not signed until 1993. Of the five countries
(India, Iraq, Sudan, Syria, and Serbia) that were accused of using chemical
weapons after the treaty signing, only India had signed the treaty prior to the
accusations, meaning that the other four states were not violating their Chemical
Weapons Convention treaty obligations by “using” chemical weapons.55F
56

The clearest defining feature of the 1945-2014 age that clearly divides it
from the interwar period is the introduction of nuclear weaponry into the
international system. While the initial, knee-jerk reaction is to assume that the

56
They would have been violating their obligations under the Geneva Protocol, but there were
significant problems with recognizing violation and adherence to that agreement. It essentially
devolved into a non-first-use treaty with no enforcement mechanism.

115

introduction of nuclear weapons changed everything for chemical weapons, a
clear explanation as to why this should be so needs to be fully explored. There are
reasons why this new technology should have little impact on previous weapons
of mass destruction (lower cost/technology requirements make chemical weapons
more attractive for poorer or less developed countries, destructiveness of
thermonuclear weaponry means that they really can’t be used to any tactical
advantage, and lack of treaty enforcement beyond a public relations black eye
means that there is little cost to their use). However, nuclear weapons really did
change the global strategic landscape (use of chemical or biological weapons
could invite retaliation through nuclear means, defense spending is redirected to
the latest technology leaving a rusting and increasingly inoperable chemical
weapons capability, and a general preoccupation with nuclear weapons as a
military panacea pushes the use of chemical weapons out of consideration for
many military operations).
Since elements of the solutions identified in the 1918-1945 dataset do not
recur in the 1945-2014 dataset, this is evidence that the mechanisms that drove
chemical weapons use and escalation in the interwar period were not responsible
for the same outcome after World War II. While a clear picture as to why states
use chemical weapons in this period did not emerge, at least one set of conditions
can be ruled out. This observation helps shape future inquiry by eliminating some

116

easy mechanisms from the list of possible conditions that are associated with
chemical weapons use.

Why Do Countries Abstain from Using Chemical Weapons?
Now that fs/QCA has identified the conditions under which states will use
and escalation the use of chemical weapons in the interwar period and laid bare
problems in the post-WWII period, a bigger, perhaps more salient question needs
to be asked. Under what conditions to states refrain from using their chemical
weapons? Unlike in traditional regression, this question cannot be answered
simply by inverting the explanatory conditions. Since fs/QCA is used when causal
asymmetry is suspected, a separate set of analyses is required to understand the
absence of a phenomenon. To do this, the outcome of interest (i.e. “use of
chemical weapons’) is transformed into its complement “non-use of chemical
weapons” by the following formula ~A=1-A. The complements are now the
outcome of interest, and are inputted into the fs/QCA analysis.

Necessary conditions for the outcomes ‘non-use of chemical weapons’ and ‘non-
use of non-lethal weapons first’ Table 21 contains the results of an analysis of the
necessary conditions for the outcome ‘non-use of chemical weapons.’ Since the
hypotheses about the roots of chemical weapons use were predicated on the

117

presence of those conditions, the absence of the outcome should be associated
with the absence of those conditions. Therefore, instead of the presence of those
conditions, necessity tests were conducted using the complements of those
conditions. As can be seen in Table 21, most consistency scores fell below the
0.90 threshold, indicating that those conditions are in themselves not necessary
for states to refrain from using chemical weapons. However, the lack of security
commitment (c) barely met the necessity threshold, with a consistency score of
0.902. As the coverage for this condition was also very high (0.902), this
condition is likely to be found in each component of the sufficiency solution.
Table 22 contains the results of an analysis of the necessary conditions for
the outcome ‘non-use of non-lethal weapons first.’ As can be seen in Table 22,
most of the consistency scores are below the 0.90 threshold, indicating that those
conditions are not in themselves necessary for states to refrain from escalating
their use of chemical weapons. As with the non-use of chemical weapons,
however, the lack of security commitment (c) is also considered a necessary
condition with a very high coverage score (0.934).

Sufficient conditions for the outcomes ‘non-use of chemical weapons’ and ‘non-
use of non-lethal weapons first’ The results of the analysis of the sufficient
conditions for the outcome ‘non-use of chemical weapons’ for the 1918-21945

118

dataset are displayed in Table 23. The overall consistency score (0.877) is high,
while the coverage score (0.574) indicates that the results cover a good section of
the cases. Three combinations of conditions lead to the non-use of chemical
weapons: democracies that are important to the international system with security
commitments fighting in conflicts in which they are at least as powerful as their
opponent, regimes with security commitments that do not perceive challenges to
their survival fighting in conflicts not in their territory (in the Global North) in
which the conflict progresses steadily in their favor and they are at least as
powerful as their opponent, and non-democracies with security commitments
fighting in conflicts not within their territory (in the Global North) in which the
conflict progresses steadily in their favor and they are at least as powerful as their
opponent. Each of the individual consistency scores are very high, although the
coverage scores tend to be low, indicating that they cover small sections of the
dataset.
This result reveals some interesting regularities and surprising anomalies.
Consistent with hypothetical expectations are the first, third, and fourth
combination of conditions. According to the hypotheses, it was expected that low
levels of challenge to regime survival will not result in chemical weapons use.
Autocracies fighting not in the Global South would also be consistent with
expectations, as theory predicted that chemical weapons use would result from
conflicts among democracies and on the periphery. Lastly, conflicts that progress

119

smoothly and involve states that have robust security agreements with major
powers is also in line with the expectations of theory.
The above result also revealed some unexpected deviations from theory.
Despite expectations of chemical weapons use, those states that are ostracized in
the international system do not use chemical weapons. Even though these states
have nothing to lose, they refrained from using their weapons. Democracies
experiencing stalemates in their conflicts in the Global South also refrained from
using their chemical arsenal, despite good reasons to do so. Something must be
preventing these states from using their chemical weapons.
Democracies fighting within their own territories is a semi-surprising twist.
While theory predicted that democracies would use chemical weapons, it also
states that chemical weapons use is more likely outside the state. This is a useful
observation for those concerned with predicting use of chemical weapons. The
final condition combination complements this finding. Even if a democratic
power is in a conflict in the Global South, the state will not use chemical weapons
if they are at least as equally as powerful as their opponent. This observation is
counter to theory, as it expects that drastically weaker states are more likely to use
their weapons to even the score. Even the chance to gain an edge on an evenly
matched opponent does not entice democracies fighting in the Global South to use
their weapons.

120

The results of the analysis of the sufficiency conditions for the outcome
‘non-use of non-lethal weapons first’ are displayed in Table 24. As with the non-
use of chemical weapons, the parsimonious solution was selected to maximize the
overall consistency and sufficiency scores. Both the consistency (0.945) and
coverage (0.945) are very high and are slightly higher than those for ‘non-use of
chemical weapons.’ Four combinations of conditions lead to the non-use of
chemical weapons: a democratic regime; a conflict not fought in the Global
South; a conflict fought within a state’s territory; and a conflict that smoothly
progresses in the state’s favor. Each of the individual consistency scores are high,
although the coverage scores tend to be low. This indicates that there are clusters
of cases explained by each condition combination fairly exclusively. Figure 12
represents these results graphically. Most cases are above the bisecting line, and
the few that are not tend to lie close under the line. The greatest challenges to this
solution occur at [1, 0.6] and [0.6, 0.0] since they are further from the diagonal
line.
Most of the results of the analysis for the condition ‘non-use of non-lethal
weapons first’ is consistent with theoretical expectations. Being in a conflict
located in the Global North, that is progressing smoothly, or is in one’s own
territory were predicted to not result in chemical weapons use. The finding that
democracies do not escalate from non-lethal to lethal chemical weapons is
surprising, as democracies were expected to use chemical weapons.

121

Since the stratification of the dataset revealed that there were different
mechanisms at work regarding the use and escalation of chemical weapons use in
the interwar and post-WWII periods, any analysis of the conditions that lead to
the non-use of these weapons would be remiss in not checking to see if this
phenomenon is repeated for non-use. Therefore, fs/QCA analyses were run with
the outcome of interest being the non-use of chemical weapons and the non-use of
-non-lethal weapons first being the outcomes of interest in the 1918-1945 and
1945-2014 datasets.

1918-1945 Dataset: Necessary and Sufficient Conditions for ‘non-use of chemical
weapons’ and ‘non-use of non-lethal weapons first’ Table 25 contains the results
of an analysis of the necessary conditions for the outcome ‘non-use of chemical
weapons’ for the 1918-1945 dataset. As can be seen in Table 25, all consistency
scores are below the 0.90 threshold, indicating that none of the conditions are in
themselves individually necessary for states to refrain from using chemical
weapons. The use of chemical weapons when the opponent is less powerful or
equal to the regime, the presence of security commitments, and high importance
in the international system are close to being necessary, but do not quite meet the
requirements. Table 26 contains the results of an analysis of the necessary
conditions for the outcome ‘non-use of non-lethal weapons first’ for the 1918-

122

1945 dataset. As can be seen in Table 26, all consistency scores are below the
0.90 threshold.
The results of the analysis of the sufficiency conditions for the outcome
‘non-use of chemical weapons’ for the 1918-21945 dataset are displayed in Table
27. The overall consistency score (0.877) is high, while the coverage score
(0.574) indicates that the results cover a good section of the cases. Three
combinations of conditions lead to the non-use of chemical weapons: democracies
that are important to the international system with security commitments fighting
in conflicts in which they are at least as powerful as their opponent, regimes with
security commitments that do not perceive challenges to their survival fighting in
conflicts not in their territory (in the Global North) in which the conflict
progresses steadily in their favor and they are at least as powerful as their
opponent, and non-democracies with security commitments fighting in conflicts
not within their territory (in the Global North) in which the conflict progresses
steadily in their favor and they are at least as powerful as their opponent. Each of
the individual consistency scores are very high, although the coverage scores tend
to be low. This indicates that there are clusters of cases explained by each
condition combination fairly exclusively.

123

The condition (a*c*o*D) describes cases such as the United States (1918-
1944), the United Kingdom (1918-1934), and France (1918-1934).56F
57
Among the
cases represented by (t*a*s*c*e*g), the second condition, are Yugoslavia (1935-
1939), German (1925-1934) and South Africa (1935-1944). The last condition,
(a*s*c*d*e*g), contains Yugoslavia (1920-1939), Spain (1940-1944), and Italy
(1918-1919).
One interesting thing to note is that although (a) and (c) are present in each
component of the solution, they are not considered necessary in separate testing.
However, an examination of these two conditions for joint necessity revealed that
they are indeed jointly necessary. This highlights the need for separate necessity
testing of suspect conditions in fs/QCA. Another observation is that the
directionality of the conditions mostly follows the expectations outlined in the
hypotheses. It was expected that the absence of the proposed conditions would be
related to the absence of the use of chemical weapons. In two out of three
component conditions, only the absence of the conditions is included. In the third,
the absence of some conditions paired with a democratic regime resulted in the
non-use of chemical weapons. Lastly, it appears that in two out of three
component conditions the degree to which the state is included in the international
system is irrelevant. During this time period, both states at the fringes of the

57
As indicated by the consistency score, some of these countries used chemical weapons during
this time period, but refrained for the most part.

124

international system and those in the center did not use chemical weapons. This is
a somewhat reassuring observation, as it indicates that states in both conditions
can exhibit restraint.
The results of the analysis of the sufficiency conditions for the outcome
‘non-use of non-lethal weapons first’ are displayed in Table 28. The consistency
score (0.935) is very high, while the coverage score (0.574) indicates that the
results cover a good section of the cases. The individual consistency scores
indicate that there is either perfect or near-perfect agreement regarding the
outcome for each solution component. Four combinations of conditions lead to
the non-use of chemical weapons: non-democracies central to the international
system with security commitments fighting in the Global North, democracies
central to the international system with security commitments that are at least as
powerful as their opponents, regimes with security commitments that do not
perceive challenges to their survival fighting in conflicts in their territory (in the
Global North) that progress steadily against opponents they are at least as
powerful as, and non-democracies with security commitments fighting in conflicts
in their territory (in the Global North) in which the conflict progresses steadily in
their favor against opponents they are at least equally as powerful as.
Examples of cases that fit the first constituent condition (c*o*d*g) are
Germany (1935-1944) and Italy (1918-1919 and 1940-1944). The second set of

125

conditions (a*c*o*D) contains cases such as the United States (1925-1944), the
United Kingdom (1918-1919 and 1925-1929) and Japan (1918-1934), among
others. Among the cases represented by (t*a*s*c*e*g), the third condition, are
Yugoslavia (1935-1939), German (1925-1934) and South Africa (1935-1944).
Lastly, the fourth set of conditions (a*s*c*d*e*g) contains Yugoslavia (1920-
1939), Spain (1940-1944), and Italy (1918-1919).
Since the conditions (a) and (c) appeared in the majority of the component
conditions, a separate analysis for necessity of these conditions was run. It was
confirmed that these conditions, as for the non-use of chemical weapons, were
jointly necessary for the non-escalation of chemical weapons.
Firstly, it is interesting to note that location (either within the state’s
territory or in the Global North) of the conflict appears in three of the component
solutions. While not necessary for the non-escalation of non-lethal to lethal
chemical weapons, it still suggests that the location of the conflict plays a role.
This is consistent with theoretical expectations; states would have a disincentive
to escalate chemical weapons use within their own territories due to the
deleterious long term effects of those agents and in the Global North due to a
higher likelihood of being held to account by European and North American
powers.

126

A second observation is that the conditions under which democracies and
non-democracies refrain from escalation are different. This leads to two
conclusions. Firstly, that neither system of governance exclusively results in non-
escalation. Both types of states have refrained from switching to lethal chemical
weapons after initiating a non-lethal campaign or sticking with one type of
chemical weapons. And secondly, the conditions under which they do so are
different. Most notably, the difference between those conditions that refer to non-
democracies (first and fourth) differ from the one (second) that includes
democracies only with regards to the absence of a stalemate and location (in the
Global North). This conclusion is useful for those groups interested in preventing
the use of chemical escalation; understanding the differences in the restraint
behavior of different types of regimes can help fine tune their surveillance tactics
recommendations for the prevention of this phenomenon.

1945-2014 Dataset: Necessary and Sufficient Conditions for ‘non-use of chemical
weapons’ and ‘non-use of non-lethal weapons first’ Table 29 contains the results
of an analysis of the necessary conditions for the outcome ‘non-use of chemical
weapons’ for the 1945-2014 dataset. As can be seen in Table 29, most consistency
scores are far below the 0.90 threshold, indicating that those conditions are in
themselves not individually necessary for states to refrain from using chemical

127

weapons. However, the regime possessing security commitments from stronger
states (c) and facing an opponent less than or equally powerful to the regime (a)
are both necessary (consistency scores 0.918 and 0.910 respectively). Both of
these conditions cover almost the entirety of cases in the dataset (coverage scores
of 0.928 and 0.908 respectively). Table 30 contains the results of an analysis of
the necessary conditions for the outcome ‘non-use of non-lethal weapons first’ in
the 1945-2014 dataset. As can be seen in Table 30, most consistency scores are
far below the 0.90 threshold. As with the use of chemical weapons, the escalation
from non-lethal to lethal chemical weapons when the opponent is less powerful or
equal to the regime (a) and the presence of security commitments (c) meet
necessity requirements (consistency scores of 0.913 and 0.919 respectively).
These conditions also cover almost the entirety of the dataset, with coverage
scores of 0.936 and 0.955 respectively.
The results of the analysis of the sufficiency conditions for the outcome
‘non-use of chemical weapons’ for the 1945-2014 dataset are displayed in Table
31. The consistency score (0.947) is very high, while the coverage score (0.737)
indicates that the results cover a large section of the cases. Four combinations of
conditions lead to the non-use of chemical weapons in the post-WWII period:
regimes central to the international system with security commitments fighting in
conflicts in the Global South against opponents they are at least as powerful as,
regime central to the international system with security commitments fighting in

128

conflicts within their own territory that progress steadily in their favor against
opponents they are at least evenly matched with, non-democracies central to the
international system with security commitments fighting in conflicts within their
own territory against opponents they are at least evenly matched with, and
regimes the do not perceive challenges to their survival with security
commitments fighting in conflicts within their own territory in the Global North
that progress steadily in their favor against opponents they are at least evenly
matched with.
Examples of cases that fit the first constituent condition (a*c*o*G) are
France (1955-1984) and Italy (1990-2014). The second set of conditions
(a*s*c*o*e) contains cases such as the China (1990-2014), the Egypt (1975-1989
and 1995-2014) and North Korea (1955-1989), among others. Among the cases
represented by (a*c*o*d*e), the third condition, are Spain (1945-1974), the USSR
(1960-1979) and Yugoslavia (1960-1989). Lastly, the fourth set of conditions
(t*a*s*c*e*g) contains Iran (1990-2014), Libya (1990-2009), and Russia (2005-
2014).
All component conditions contain having at least an evenly matched
opponent (a) and the presence of security commitments (c), verifying the finding
that they are necessary for the non-use of chemical weapons after the Second
World War. This observation is consistent with theoretical expectations. Being as
strong or stronger than one’s opponent does not create an incentive to use

129

chemical weapons to even the playing field; one would only conceivably use such
weapons if one had already been committed to doing so. Having security
commitments accomplishes the same thing by evening the playing field using an
outside player.
The directionality of the component conditions is mostly consistent with
theoretical expectations. The presence of a condition only appears in on
component, with the rest all indicating that the absence of these conditions is
jointly sufficient. One interesting thing to note is that regime type only comes into
play in one component (number three). For all the other sets of conditions, both
democracies and non-democracies behave in the same way. This is an important
observation, as it helps demonstrate that in the post-WWII period are restrained
by the same sets of conditions (for the most part). This is a useful fact for those
interested in preventing chemical weapons use. Any state under these specific sets
of conditions can be expected not to use chemical weapons.
The results of the analysis of the sufficiency conditions for the outcome
‘non-use of non-lethal weapons first’ for the 1945-2014 dataset are displayed in
Table 32. The consistency score (0.960) is very high, while the coverage score
(0.719) indicates that the results cover a large section of the cases. Four
combinations of conditions lead to the non-escalation of chemical weapons use
after 1945: regimes central to the international system with security commitments

130

fighting in conflicts in the Global South against opponents they are at least as
powerful as, regime central to the international system with security commitments
fighting in conflicts within their own territory that progress steadily in their favor
against opponents they are at least evenly matched with, non-democracies central
to the international system with security commitments fighting in conflicts within
their own territory against opponents they are at least evenly matched with, and
regimes the do not perceive challenges to their survival with security
commitments fighting in conflicts within their own territory in the Global North
that progress steadily in their favor against opponents they are at least evenly
matched with.
All component conditions contain having at least an evenly matched
opponent (a) and the presence of security commitments (c), verifying the finding
that they are necessary for the non-escalation of chemical weapons use in the
1945-2014 dataset. It is important to note that the solution for the non-use of non-
lethal weapons first fs/QCA analysis is almost identical to the non-use of
chemical weapons in this dataset. The only differences are small fluctuations in
the consistency and coverage scores for each individual set of conditions. This
means that fs/QCA was not able to distinguish the difference (if there was any)
between those states that simply did not use chemical weapons and those that did
not escalate. If there is indeed no difference, then whether or not a state will
choose not to escalate is essentially random.

131

Differing results in the two datasets: what does this mean? Comparing the fs/QCA
results across the two datasets yields one major observation. It is abundantly clear
that fs/QCA solutions for the same outcome in different time periods are not the
same. There are some conditions that appear in multiple places (i.e. t*a*s*c*e*g
appears in both non-uses of chemical weapons and non-use of non-lethal weapons
first analyses), but most conditions are not consistent across the two datasets. As
with the use of these weapons, this indicates that most of the forces driving the
phenomenon of non-use change after the Second World War.
Even though the same conditions did not drive the same phenomena
across time, it does appear that the same conditions drove both the non-use of
chemical weapons and the non-escalation of chemical weapons use in each
dataset. The two outcomes (non-use and non-use of non-lethal first) differ by only
one condition in the 1918-1945 dataset and are identical in the 1945-2014 dataset.
While it would have been useful to be able to identify when states choose not to
escalate, this finding is very encouraging. It simplifies the job for organizations
that are interesting in preventing the use of chemical weapons. They only need
concern themselves with promoting the conditions that are associate with the non-
use of chemical weapons, as non-escalation will take care of itself.

132

What Cases Drive the Use (And Non-Use) of Chemical Weapons?
In order to assess which cases are driving the above results for the use and
non-use of chemical weapons, attack ratios were calculated using fs/QCA truth
tables. Each fuzzy dataset was altered into a crisp set form- all fuzzy values below
[0.5] were assigned a [0] and all values above were assigned a [1]. This
corresponds to case being completely out or in a given set. Transforming fuzzy
values to crisp values was done in order to keep later calculations as close in
meaning to standard epidemiologic attack ratios. Truth tables for each dataset and
outcome of interest were then constructed.
The truth table consistency scores for crisp sets represent the percentage of
cases in that given row that possess a [1] for the outcome of interest. To calculate
the attack ratio, each consistency score was divided by the lowest present
consistency score. The rows were then ranked greatest to least attack ratio.

Use of chemical weapons The results of the analysis of the relative risk of
chemical weapons use for the 1918-1945 dataset are displayed in Table 33. The
more consistent each row is with regards to resulting in chemical weapons use,
the higher an attack ratio it was assigned. There is a break in the attack ratios
between row six and seven, indicating that those cases responsible for the
phenomenon of chemical weapons use lie above the break. These cases include

133

Japan (1935-1949), the USSR (1920-1924), and Spain (1920-1929). The Japanese
cases all occurred during the Second Sino-Japanese War and World War II, the
USSR case occurred during the Tambov Rebellion during the Russian Civil War,
and the Spanish cases take place during the Rif War in Spanish Morocco.
It is interesting to note that the top two Japanese cases only differ
regarding location (in or outside the state). The third Japanese case only differs
from the other two in that there was a stalemate present and Japan had a security
commitment from an outside power. The Spanish cases differ from each other in a
similar war. The latter case lacks a stalemate and is perpetrated by a regime that is
not considered democratic in the crisp set.
The results of the analysis of the relative risk of chemical weapons use for
the 1945-2014 dataset are displayed in Table 34. There is a small break in the
attack ratios between rows three and four, indicating that the cases above that are
driving the phenomenon. These cases include Iran (1980-1989), Libya (1985-
1989), and Syria (2010-2014). The Iranian cases occurred during the decade-long
Iran-Iraq War, the Libyan case occurred during the Chadian-Libyan Conflict, and
the Syrian case occurred during the recent and ongoing Syrian Civil War.
Compared to the previous table, the Iranian, Libyan, and Syrian cases are
all quite different from each other. However, they all share a threat to the survival
of the regime and the presence of a stalemate. They also are all located in what is

134

considered the Middle East/North Africa (MENA). If the number of rows
considered to be driving the phenomenon is shifted to include rows four and five
(above a big break), it is interesting to note that the vast majority of cases occur in
MENA. The only country not in this region is the United States.

Use of non-lethal weapons first The results of the analysis of the relative risk of
non-lethal chemical weapons first for the 1918-1945 dataset are displayed in
Table 35. There is a break between rows three and four, indicating that the cases
responsible for the phenomenon all belong to Japan.
The results of the risk analyses contained in Tables 33 and 35 combined
provide insight as to why fs/QCA was unable to distinguish the differences
between those states that use chemical weapons and those that escalate from non-
lethal to lethal chemical weapons. In the fs/QCA analysis, the non-lethal chemical
weapons first solution only contained component solutions from the general
chemical weapons use fs/QCA analysis. This can be explained by the fact that the
cases Japa1935, Japa1940, and Japa1945 all appear in both tables with relatively
high attack ratios. In other words, these three cases are driving the results in both
analyses.
The results of the analysis of the relative risk of non-lethal chemical
weapons first for the 1945-2014 dataset are displayed in Table 36. There is a

135

break between the first and second rows, indicating that only the first row is
driving the fs/QCA results. This row includes the cases Egypt (1960-1969), Iraq
(1980-1989), and Syria (1975-1979). The Egyptian cases occurred during
Egyptian involvement in the Yemeni Civil War, the Iraq cases occurred during
the Iran-Iraq War, and the Syrian case did not include the escalation of non-lethal
to lethal chemical weapons.
One observation is that the top row which is driving the escalation
phenomenon in the 1945-2014 dataset is not perfectly consistent. This is a
contrast to the previous risk analyses in which the cases driving the phenomenon
are all perfectly consistent. This suggests the conclusion that state may not chose
to escalate from non-lethal to lethal chemical weapons in a consistent fashion
after WWII. Only further study examining these cases and conditions can clear up
if this is indeed the case.

Comparison of the 1918-1945 and 1945-2014 datasets When the results of the risk
analysis of the two datasets are laid side-by-side, some patterns emerge. Firstly, as
mentioned above, the same cases driving both the overall use of chemical
weapons and the escalation from non-lethal to lethal chemical weapons in the
1918-1945 dataset are the same. This explains why the fs/QCA results are so
similar. Secondly, while the top set of conditions in the non-lethal first 1945-2014

136

database is also included in the chemical weapons use analysis from the same
dataset, the majority of cases do not overlap. This could be an indicator that those
states that choose to use chemical weapons are different from those states that
choose to escalate. Lastly, there is a marked difference in the composition of the
truth table rows driving the phenomenon for a given category in the two datasets
(i.e. 1918-1945 vs. 1945-2014). This supports the conclusion that the drivers of
each phenomenon are different in the two time periods. The conditions associated
with chemical weapons use in the interwar period were not connected to chemical
weapons use after WWII. This validates the decision to split the original dataset in
relation to this cut point.

Non-use of chemical weapons The results of the analysis of the relative risk of
non-use of chemical weapons for the 1918-1945 dataset are displayed in Table 37.
A minimum of two cases per truth table row was required due to a higher number
of cases being analyzed.57F
58
There is a break in attack ratios between the fourth and
fifth row, indicating that the top four rows are driving this phenomenon. These
rows are all perfectly consistent and include Germany, Japan, South Africa, the
United Kingdom, the United States, and Yugoslavia.

58
Two cases were required to eliminate a number of rows which only contained a single case. By
definition, these truth table rows are all perfectly consistent, and obscure the overall picture.

137

In contrast to previous analyses, the cases driving the non-use of chemical
weapons in the 1918-1945 dataset are not restricted to any geographical region.
There are cases in Europe, Asia, Africa, North America, and Eastern Europe.58F
59

This would support the conclusion that the non-use of chemical weapons is not
“owned” by a particular region. All states can choose not to use their chemical
arsenals, given the proper set of conditions.
The results of the analysis of the relative risk of non-use of chemical
weapons for the 1945-2014 dataset are displayed in Table 38. Due to the sheer
number of cases, a minimum of five cases per truth table row was required in
order to identify those conditions that were responsible for truly driving the
phenomenon. There is a small break in attack ratios between the sixth and seventh
row, indicating that the top six rows are driving this phenomenon.
While the large number of cases included in these rows precludes the
countries from being listed individually, it should be noted that this is a very good
thing. Not only are the top five rows perfectly consistent regarding the outcome,
the top row includes a total of 42 cases. This indicates that the fs/QCA findings
based on these cases should be fairly robust. The introduction/subtraction of a

59
South America is excluded as there are no countries on that continent that are considered
chemical weapons possessor/user states.

138

case should not change the results in any dramatic sort of way because over 100
cases are responsible for driving the results.59F
60

Non-use of non-lethal weapons first The results of the analysis of the relative risk
of non-use of non-lethal weapons first for the 1918-1945 dataset are displayed in
Table 39. A minimum of two cases per truth table row was required due to a
higher number of cases being analyzed. There is a small break in attack ratios
between the eighth and ninth row, indicating that the top eight rows are driving
this phenomenon. These rows are all perfectly consistent and contain Germany,
Japan, the United Kingdom, the United States, France, Australia, Yugoslavia,
Italy, and South Africa.
As with the non-use of chemical weapons, the cases driving the chemical
weapons non-escalation in the 1918-1945 dataset are not restricted to any
geographical region. There are cases in Europe, Asia, Africa, North America, and

60
While a relatively new field of research, assessments of the reliability of QCA yield mixed
results. Some studies have found that QCA is sensitive to minor parametric and model
specification changes (Krogslund and Michel 2014b; Krogslund, Choi, and Poertner 2015, 21-41).
However, these analyses focus on small-to-medium n crisp set QCA studies. Less research has
been done to assess the validity of large-n QCA results (which this project is). There is, however,
an indication that larger-n studies maybe produce more stable results. Compared to QCA studies
in which there are less than 50 cases, large n-studies tend to produce more reliable results when
the number of conditions is kept low (Krogslund and Michel 2014a). This same study also raised
questions regarding the methods used to evaluate the reliability of QCA. QCA is paradoxically
more stable in drop-one sensitivity tests when the n/k ratio is low (Krogslund and Michel 2014a).

139

Eastern Europe. All states can choose not to escalate their use of chemical
weapons, given the proper set of conditions.
The results of the analysis of the relative risk of non-use of non-lethal
weapons first for the 1945-2014 dataset are displayed in Table 40. A minimum of
five cases per truth table row was required due to a higher number of cases being
analyzed. There is a small break in attack ratios between the seventh and eighth
row, indicating that the top seven rows are driving this phenomenon.
Even though the larger number of cases included in these rows prevents
the countries from being listed individually, this is again a good thing. All the
rows driving the phenomenon are perfectly consistent, and the top row includes a
total of 42 cases. As with Table 38, this is an indicator that the fs/QCA findings
based on these cases should be fairly robust. The introduction/subtraction of a
case should not change the results in any dramatic sort of way because over 120
cases are driving the results.

Comparison of the 1918-1945 and 1945-2014 datasets The results of the truth
table analyses for the non-use and non-escalation of chemical weapons reveal a
very different picture from that of chemical weapons use and escalation. Contrary
to those analyses, it appears that the drivers of both the non-use and non-
escalation of the use of chemical weapons are both similar to each other and in

140

both time periods. The drivers of the non-use of chemical weapons in the 1918-
1945 dataset are also drivers in the non-use of non-lethal first datasets. The
conditions that were not associated with the use of chemical weapons in the
interwar period were also associated with the non-escalation of use overall.
Additionally, some of the drivers of the non-use of chemical weapons in the 1945-
2014 dataset (Table 38 lines 1, 4, and 5) were also found to be drivers of non-
escalation in the same period. The same pattern holds for the non-escalation of
chemical weapons use. Some of the drivers of non-escalation in the 1918-1945
dataset (Table 39 lines 1, 2, 3, and 6) are considered drivers in the 1945-2014
dataset.
Two other observations can be pulled from the truth table analysis. First,
democracies, in the absence of any other condition (threat to regime, stalemates,
etc.) refrained from using chemical weapons (1918-1945 dataset) and escalating
their use of chemical weapons with perfect consistency.60F
61
Secondly, non-
democracies, in the absence of any other condition, have refrained from using and
escalating chemical weapons use since 1945 with perfect consistency. These two
observations, although expected based on theory, are never-the-less reassuring.

61
While this set of conditions also appears in the 1945-2014 dataset, it is not considered a driver
of the phenomenon and is not perfectly consistent. Some democracies did use chemical weapons
in this period.

141

Without a good reason (or a reason that can be construed as one) both systems of
governance refrain from using their chemical arsenals.

Are Supersets Useful for Identifying Potential Sufficient Conditions?
This project was developed to have two investigative purposes: firstly, to
identify under what conditions states will both use and refrain from using their
chemical weapons, and secondly, to test a new method designed to reduce some
methodological problems associated with fs/QCA in the social sciences. The
above pages have accomplished the former: fs/QCA revealed that there was a
major shift in which states used/did not use chemical weapons after WWII and
identified what those conditions were. This next segment addresses the second
purpose of this project: assessing a new method designed to help identify
sufficient conditions in fs/QCA while further minimizing threats to the analyses’
validity.
As mentioned in chapter 3, the problem of limited diversity is significant
for all variants of QCA. The higher the number of logical remainder truth table
rows, the greater the problem grows for QCA regarding logical inference. The
standard method of dealing with this problem is to keep the total number of
conditions being examined low and apply Ragin’s standard analysis.

142

While these two strategies greatly reduce the problem of limited diversity,
it can pose a problem for those studies, like this one, in which the proposed
mechanisms undergirding various social phenomena are not well understood or
hypothesized. How, then, is one able to evaluate a relatively large (for QCA)
number of conditions for their impact on the outcome of interest and minimize the
threat of limited diversity? In this project, I proposed using a third joint strategy in
order to assess larger numbers of potential conditions: researcher constructed
supersets.

What was predicted? According to set theory, a superset is a set that contains all
the elements of a smaller set. By using this property, I proposed that an evaluation
superset could be constructed by using the Boolean AND function. By defining
the superset as the degree to which a given case belongs to all three sets
simultaneously, the superset should function as a means to identify if there is a
condition contained within that is jointly sufficient with other conditions under
consideration.
The logic of supersets leads to the following two predictions: 1) a superset
containing a sufficient condition (or combination) will be sufficient in
combination with the other conditions and 2) the absence or presence of a

143

superset as a jointly sufficient condition will coincide with the joint sufficiency of
the absence or presence of at least one subset.

How was the superset analysis carried out? In order to test whether or not
supersets can predict which conditions will prove jointly sufficient, three
supersets were constructed. The superset Military Vulnerabilities (TAS) was
constructed from the Boolean function TAS = regime survival (T) AND power
asymmetry (A) AND stalemate (S). The superset representing Political
Vulnerabilities (COD) was constructed from the function COD = credible
commitment (C) AND ostracization (O) AND democratic regime (D). Lastly, a
superset on behalf of Conflict Location (EG) was constructed using the function
EG = foreign territory (E) AND Global South (G).
Four outcomes of interest were chosen to test whether supersets were
predictive of subset sufficiency: use of chemical weapons in the 1918-1945
database, non-use of chemical weapons in the 1918-2014 database, non-use of
chemical weapons in the 1945-2014 database, and non-use of non-lethal weapons
first in the 1945-2014 database.61F
62
An analysis for each outcome was conducted in

62
An analysis of supersets using the non-lethal first outcome for the 1918-1945 database would be
of little use, as the outcomes are identical to the overall use of chemical weapons. The use of
chemical weapons and non-lethal weapons first in the 1945-2014 database were rejected as the
intermediate solution did not meet a minimum sufficiency threshold (0.70).

144

the following manner. First, the intermediate fs/QCA solution for these outcomes
were recorded from Tables 15, 27, 31, and 32.62F
63
Next, an fs/QCA analysis for the
first outcome was conducted substituting TAS for the conditions T, A, and S, and
the presence/absence of the superset was recorded.63F
64
Similar analyses using
fs/QCA were then conducted for both COD and EG. This procedure was then
repeated for the remaining three outcomes of interest. The results for all the
fs/QCA analyses were then compiled into Table 41.

Is this a useful strategy to identify potential sufficient conditions? Regarding the
first expectation, that supersets containing a sufficient condition will themselves
be sufficient, proved to be true. In each of the four cases, supersets that were
included in the intermediate solution of fs/QCA also had components in the
fs/QCA analysis using subset conditions. There was no instance in which a
superset was labelled as sufficient and subset testing indicated that it was not
sufficient. The third analysis is a perfect example of this expectation holding true.
The second expectation did not hold up as well as the first. It was expected
that the absence/presence of a sufficient superset condition would correlate with

63
The intermediate solution was selected as it provides a more parsimonious solution compared to
the complex solution, but does not include contradictory assumptions like the parsimonious
solution.
64
The conditions for the fs/QCA analyses in the above tables were replicated in the superset
analysis in order to ensure the comparability of results.

145

the absence/presence of a sufficient subset. The above results demonstrate that
this expectation had mixed results.
Firstly, if the subset condition was considered sufficient at all, either
absent or present, then it showed up in the superset analysis. If, for example,
subset condition S was considered sufficient, the superset analysis returned either
the absence or presence of TAS as sufficient, and vice versa. The exception to this
finding is the use of chemical weapons (1918-1945) fs/QCA analysis. In this
analysis, the superset method failed to indicated that C, D, c, and d were sufficient
conditions.
Secondly, this method of analysis had difficulty when there were
circumstances in which the presence and absence of conditions were considered
sufficient in different solution components. For example, the second analysis in
the table failed to indicate that D was considered sufficient. However, the absence
of this condition was indicated as sufficient in both the superset and subset
analyses.
While there are some flaws to this method, as a way to test if any subset
will be jointly sufficient, this approach passed a minimal test of usefulness. While
a more fine-grained analysis using the subsets themselves is needed to parse out
the relationship between these conditions in an fs/QCA analysis, the supersets
provided a useful tool to indicate if there were potential sufficient condition

146

present. Because the supersets were considered sufficient if any of their
components (either absent or present) are sufficient, the superset could serve as a
good indicator of future sufficiency when testing rafts of different conditions.
Some cautions in using supersets to predict subset sufficiency arise from
this analysis. Firstly, the number of subsets used to create a superset should be
kept to a minimum (i.e. closer to three than 10). This would allow the researcher
to preserve the ability to identify potential sufficient condition while not
worsening the problem of limited diversity later on. Secondly, the results of the
superset analysis should not be considered the final valid solution. As evidenced
in the above table, this method has difficulty if both the presence and absence of a
condition are considered sufficient. While it can tell you if the condition is
sufficient in some way, it cannot tell you more than that. Further fs/QCA analyses
are needed using the subset components.

What Have We Learned?
This project was designed with dual purpose: first to assess what the
conditions were that led states to both use and abstain from using their chemical
weapons, and second to test a new method designed to reduce some
methodological problems associated with fs/QCA in the social sciences. Both
goals were accomplished, some leading to interesting results. Table 42 contains

147

the results of all preceding fs/QCA analyses, as well as the attack ratio testing
conducted to identify key cases/conditions. Table 42 also incorporates the results
of superset testing from Table 41.

Use of chemical weapons There were three major highlights from the fs/QCA
investigation into the conditions surrounding chemical weapons use. First,
fs/QCA could not identify the conditions under which chemical weapons were
used in the post-WWII period. The only solution that fs/QCA was able to find
only applied to two cases from Iran, and even then there was some question as to
whether Iran even used chemical weapons. Since the same circumstances result in
both chemical weapons use and abstention, it appears that there may be two
different, although non-exclusive explanations for the lack of fs/QCA results.
Firstly, it is possible that the conditions drawn from the nuclear
proliferation literature and hypothesized in chapter 3 do not capture the reasons
why a state in the post-WWII world would choose to use chemical weapons. If
this were true, a null finding would be expected from fs/QCA, which is essentially
what the analysis returned. One reason that a state may use chemical weapons that
is not captured by the hypotheses outlined in this study is that the state is
conducting field tests of their munitions. This was the charge levelled at the
Egyptians when they intervened in the Yemeni civil war between 1963 and 1967.

148

The Egyptian military initially launched attacks with non-lethal weapons, but
quickly escalated to using both phosgene and mustard gas, and eventually nerve
gas (Center for Nonproliferation Studies. Monterey Institute of International
Studies 2015a; Cordesman 2004; Hilmas, Smart, and Hill 2008, 9-76; Spiers
2010, 86; Stockholm International Peace Research Institute 1971; Tucker 2006,
191-195). It appears that South Africa may also have engaged in chemical
weapons field tests on opposing forces in 1992 (Bale 2006, 42; de Jonge
November 1997, 6-12; Fauvet 1998; Republic of South Africa. National
Intelligence Service December 1992, 14 ii; RSA December 1992, 14 ii; Truth and
Reconciliation Commission; United Nations 1993; United States of America.
Department of State March 9, 1992; John Yeld. 1998). More research is needed to
determine if this is a widespread phenomenon.
The second reason why fs/QCA was unable to identify the reasons why a
state would use their chemical weapons in the 1945-2014 dataset is that every
decision to use chemical weapons may be highly individualized and context
dependent, even within a given state. Instead of common triggers for this
behavior, the use of chemical weapons could be a response tailored to specific
military and political needs. If this were true, it is likely that fs/QCA would not
return consistent results, as the decision by one state to use chemical weapons
would not be replicated in another. To identify whether this is the case, several
small-n case studies need to be conducted to see if a state (particularly a repeat

149

offender) chose to use their weapons under consistent sets of circumstances or if
they were allowed due to specific needs of the political-military structure.
The second highlight from this study is that fs/QCA was able to identify
the necessary and sufficient conditions under which chemical weapons were used
in the Interwar Period. A conflict fought in the Global South against a
significantly more powerful foe in which there is substantial threat to the survival
of the state’s regime were jointly necessary for states to use chemical weapons.
This is consistent with observations of cases from the time period: every instance
of use occurred outside of Europe and North America and away from the center of
international power. Another observation from the 1918-1945 dataset is that the
degree of inclusion in the international system appears to have played little role in
allowing states to use their chemical weapons. Both influential states (such as the
United Kingdom and France) and relatively marginalized states (such as Japan
and Italy) used their chemical weapons in the interwar period.
Thirdly, fs/QCA was unable to distinguish between those who chose to
escalate their use of chemical weapons and those that stuck with a single type of
weapon. This was due to the fact that the Japanese cases that drove the escalation
phenomenon were also driving cases in the overall use of chemical weapons.
Between 1935 and 1945, Japan engaged in chemical weapons use and
experimentation in China during both the Second Sino-Japanese War and the

150

Second World War. While many Japanese officers were guilty of war crimes at
least on par with the Nazis, most of the Japanese chemical weapons personnel
were given immunity from prosecution in exchange for working in the US
chemical corps.
Superset testing for the use of chemical weapons in the 1918-1945 dataset
yielded mixed results. While it was able to correctly identify that some of the
subsets would later prove to be sufficient, it failed to identify that both the
presence and absence of conditions C and D were also sufficient subset
conditions. This superset analysis yielded the biggest error of the four conducted.

Non-use of chemical weapons There are two major insights that can be drawn
from the fs/QCA analysis of the non-use of chemical weapons. Firstly, the reasons
states refrained from chemical weapons use in the interwar period are different
from reasons for restraint after WWII. What these time periods share, however,
are the same necessary conditions for non-use. In both time periods, being at least
as powerful as one’s opponent in a conflict (a) and the state possessing security
commitments (c) were either individually or jointly necessary conditions for the
non-use and non-escalation of chemical weapons usage.
Secondly, as with chemical weapons use, fs/QCA was not able to
distinguish between those states that chose not to escalate and those that chose to

151

refrain from chemical weapons use all together. The fs/QCA solutions in both
time periods were either identical or varied only slightly. The reason fs/QCA
could not tell the difference is due to the fact that the sets of conditions that drove
the different phenomena overlap.
Superset analysis for the non-use and non-escalation of chemical weapons
use in the 1945-2014 dataset worked perfectly. It was able to correctly identify
which subsets would prove to be significant in later testing as well as bot the
correct presence/absence of that subset. Superset testing fared less well in the
non-use of chemical weapons analysis of the 1918-1945 dataset. While it
correctly identified that the absence of the subsets would be sufficient, it did not
detect that the presence of one subset was also sufficient.

Conclusion
The results of the fs/QCA analyses conducted in this chapter advance our
understanding of the conditions under which states will choose to both use and
abstain from using their chemical weapons arsenals. It also revealed an
unexpected rift in the data, separating those states who possessed chemical
weapons in the interwar period from those that possessed them after the Second
World War.

152

fs/QCA analysis identified the consistent conditions under which states in
the interwar period chose to both use their chemical weapons arsenals and
escalate from using non-lethal chemicals to lethal substances. As the conditions
resulting in both outcomes were nearly identical, it was concluded that fs/QCA
could not differentiate between this special case of chemical weapons escalation
and the overall phenomena. The reason for this lack of differentiation was
revealed through epidemiologic investigation: the cases driving two phenomena
were identical.
fs/QCA was unable to identify consistent conditions under which states
use chemical weapons after WWII. It is hypothesized that this study may not
adequately capture the conditions underling chemical weapons use in the latter
half of the 20
th
and the beginning of the 21
st
centuries and/or that decisions to use
chemical weapons may be highly contextually dependent in the nuclear era.
Additional cases studies are needed in order to assess if these conjectures are
indeed to case.
In contrast to the state use of chemical weapons, fs/QCA was able to
identify highly consistent conditions under which states refrain from using their
chemical arsenals. Most importantly, the conditions under which states refrained
from using their weapons between 1918 and 1945 are not the same as those that
existed after 1945. As with the escalation of chemical weapons use, fs/QCA was

153

unable to differentiate between those states that chose to refrain from use and
those that chose to refrain from escalation. This was due in large part to the
sharing of important cases between the two categories.
Lastly, the use of supersets as indicators of later subset sufficiency passed
its proof of concept test. While this method had difficulty indicating whether it
was the absence and/or presence of a superset of conditions that would prove
sufficient, it did indicate that the superset in general contained a sufficient
condition. Further testing on the limits of this method for exploratory research is
needed to flesh out exactly why these problems occur and how they can be
addressed.

154

Chapter 5
What Have We Learned & Where Do We Go from
Here?

155

This dissertation contributes to our understanding of three distinct areas of
research in international relations: understanding the phenomenon of the state
disinclination to use chemical weapons (see Cole 1998, 119-132; Jefferson 2009;
Mandelbaum 1981; Price and Tannenwald 1996; Price 1997; Van Courtland
Moon 1984, 3-35; Van Courtland Moon 2008, 55-77), the conditions under which
states will use their chemical arsenals, and fs/QCA methodology in the social
sciences (De Meur and Rihoux 2002; De Meur, Rihoux, and Yamasaki 2009;
Ragin 2000; Ragin, Drass, and Davey 2006; Rihoux and Ragin 2009; Schneider
and Wagemann 2012).
The accepted narrative in international relations is that states rarely use
their chemical weapons due to a taboo placed on their use that truly emerged after
WWII. The major debate in the literature is over where this phenomenon came
from. Essentialists claim this taboo arose through an evolutionarily derived
mechanism (Cole 1998, 119-132; Jefferson 2009; Mandelbaum 1981; Van
Courtland Moon 2008, 55-77) while military strategists argue that strategic
deterrence played a central role in preventing chemical weapons use in the WWII
European theatre (Van Courtland Moon 1984, 3-35). Constructivists disagree with
both camps, positing instead that the emergence of the chemical weapons taboo is
an instance of a special and contingent social construction centered in the debate
in the early years of the 20
th
century about what defines a civilized state (Price
and Tannenwald 1996; Price 1997).

156

This dissertation argues that the literature has misidentified the
phenomenon at hand; the state disinclination to use chemical weapons does not
meet the criteria of tabooness put forth by the above authors and by the
sociological/anthropological literature. Instead, it most closely resembles what
could be described as a prohibitory norm.
A taboo is described as the normative belief that the use of chemical (or
nuclear) weapons is prohibited (Tannenwald 2007). Usage results in a
fundamentally altered reality, from which there is no return. It was argued in this
dissertation that even according to this definition the disinclination of states to use
chemical weapons fails to qualify as a taboo. Firstly, the phenomenon is not as
rare as is commonly perceived. While the use of chemical weapons could be
considered rare when compared to the use of tanks or firearms in a conflict,
compared to other classes of WMD these weapons are used far more frequently.
Usage of these weapons are not just isolated incidents; they occur in a significant
percentage of possessor state conflicts.
Secondly, the use of chemical weapons only results in a lackluster
response from the international system. Taboos inspire a strong retaliatory
response by observers; the use of chemical weapons has typically only inspired
protestations of disgust by international actors. Little is done to actually halt or
punish the behavior. The international disinclination to use chemical weapons

157

better fits a prohibitory norm, which forbids certain socially important actions that
it enforces through mechanisms such as the legal system.
Relabeling the phenomenon of the state disinclination to use chemical
weapons as a prohibitive norm has real world significance. By describing this
phenomenon as a taboo, decision makers perpetuate the message that the risk of
experiencing this behavior is nearly non-existent. This perception leads to certain
policy prescriptions. If the likelihood of chemical warfare is extremely low, then
few resources need to be dedicated to preparing for this eventuality. Instead of
developing and producing protective gear for frontline troops, these monies will
instead go towards other defense or civilian projects. In the event that the
perception that states don’t use chemical weapons is wrong, the consequences can
be devastating. Many thousands of people could die and the survival of the regime
itself could be threatened. Accurately identifying what the phenomenon of the
state disinclination to use chemical weapons is is therefore extremely important in
order to give it an appropriate national security priority.
In order to allow decision makers to understand under which conditions
chemical weapons may be used and not used, this dissertation conducted a
systematic analysis using fs/QCA. As introduced in Chapter 3, QCA is a method
developed by Charles Ragin in the 1980s. This method looks for sets of
conditions that are associated with a particular outcome, positing that these

158

conditions work together to result in the outcome of interest. Traditional statistical
methods, by contrast, look at the impact individual variables have on an outcome
and assume that the effects of all variables are additive. fs/QCA also relaxes the
statistical assumption of equifinality; multiple dissimilar combinations of
conditions can and do result in the same outcome. Statistical methods cannot
account for this phenomenon. Lastly, fs/QCA embraces the idea of asymmetric
causation. In statistics, the non-occurrence of a phenomenon is assumed to be due
to the absence of causal variables. In fs/QCA, the non-occurrence of a
phenomenon is tested for separately and often has little to do with the non-
occurrence of conditions that are associated with the presence of the outcome. In
international relations, this idea is encapsulated in the notion that the causes of
peace are not simply the absence of the causes of war.
In addition to relaxing key assumptions in statistical methodology, fs/QCA
can be used under conditions of great uncertainty. Most of the accounts regarding
the historical usage of chemical weapons have great degrees of uncertainty
attached to them. People lie in their reports, misidentify natural phenomena as a
chemical attack, and devote few media resources to the incident, for example. The
scoring system for fs/QCA allows that uncertainty to be built into the analysis,
with more analytical weight being allocated to those cases where chemical
weapons were definitely(not) used than to those where the accounts are unclear.

159

Chapter 4 contains the results obtained from the fs/QCA analysis of the
conditions associated with chemical weapons use and abstention. For the interwar
period, fighting in a conflict in the Global South in which you are weaker than
your enemy and face a threat to regime survival was a set of conditions
considered necessary for a state to use their chemical weapons. For the non-use of
chemical weapons, being at least as powerful as one’s opponent and possessing
security commitments were deemed jointly necessary.
fs/QCA was also able to identify conditions sufficient for states to use
their chemical weapons. While the overall solution met the consistency
requirements to be considered sufficient, its coverage score indicates that it only
covers a small segment of the cases. The two components of the solution that
individually met sufficiency requirements were a) non-democratic states with
security commitments from a more powerful state that perceive severe challenges
to the regime by a more powerful opponent in the Global South and b) democratic
states without security commitments that perceive severe challenges to the regime
by a more powerful opponent fighting outside their territorial borders in the
Global South are sufficient for the state to use chemical weapons.
In the interwar period, fs/QCA also identified a set of conditions deemed
sufficient for states to refrain from using their chemical weapons. The overall
solution was highly consistent, and the coverage score indicated that the solution

160

covered a decent portion of the cases. The four components of the solution
sufficient to result in the non-use of chemical weapons are a) non-democracies
central to the international system with security commitments fighting in the
Global North, b) democracies central to the international system with security
commitments that are at least as powerful as their opponents, c) regimes with
security commitments that do not perceive challenges to their survival fighting in
conflicts in their territory (in the Global North) that progress steadily against
opponents they are at least as powerful as, and d) non-democracies with security
commitments fighting in conflicts in their territory (in the Global North) in which
the conflict progresses steadily in their favor against opponents they are at least
equally as powerful as.
Identifying the conditions under which states used and abstained from
chemical weapons use after WWII proved more challenging. fs/QCA was unable
to identify a solution identifying the conditions under which states used their
chemical weapons.64F
65
This indicates that there is a severe underlying problem with
the data. Possible reasons for this lack of results include omitted variable(s) and
highly individualized chemical weapons decision making processes.
fs/QCA was able to identify the conditions under which states refrained
from using their chemical weapons, however. Similar to the interwar period, both

65
One set of conditions was deemed sufficient, but only covered two cases.

161

being at least evenly matched with an opponent and possessing security
commitments were separately necessary conditions for chemical weapons
restraint. A very highly consistent solution with very good coverage sufficient for
states to refrain from using their chemical weapons was also identified. In the
post-1945 period, four sets of conditions were considered sufficient for chemical
weapons non-use: a) regime central to the international system with security
commitments fighting in conflicts in the Global South against opponents they are
at least as powerful as, b) regime central to the international system with security
commitments fighting in conflicts within their own territory that progress steadily
in their favor against opponents they are at least evenly matched with, c) non-
democracies central to the international system with security commitments
fighting in conflicts within their own territory against opponents they are at least
evenly matched with, and d) regimes the do not perceive challenges to their
survival with security commitments fighting in conflicts within their own territory
in the Global North that progress steadily in their favor against opponents they are
at least evenly matched with.
Separate analyses evaluating the conditions associated with the escalation
of chemical weapons use from non-lethal to lethal agents revealed problems for
the hypotheses. fs/QCA could not distinguish between the conditions under which
states will use chemical weapons in general and between those that choose to
escalate in both time periods. Attack ratio analysis revealed that fs/QCA could not

162

distinguish between these different outcomes because the cases that drive the
fs/QCA analysis in both general chemical weapons use and chemical escalation
are the same.
In addition to analyzing the conditions associated with chemical weapons
use and restraint in a systematic manner for the first time, this project also
introduced two new methods to be used in conjunction with fs/QCA. In order to
identify the cases that were driving the phenomena under investigation, the use of
attack ratios was imported from epidemiology. Attack ratios allow investigators to
determine the source of an outbreak from a multitude of possible sources. By
applying and comparing attack ratios to the different rows on the fs/QCA truth
tables, the cases considered to be driving the different fs/QCA analyses were
identified. In addition, attack ratio analysis was able to determine why the
conditions specifically associated with chemical weapons escalation could not be
distinguished from general chemical weapons use.
The second innovation added to fs/QCA in this project was designed to
allow this method to examine a larger number of conditions simultaneously while
minimizing the problem of limited diversity. As mentioned in Chapter 3, fs/QCA
is a method particularly susceptible to the problem of limited diversity. This
problem is exacerbated when large numbers of conditions are examined
simultaneously to identify sufficient conditions in fs/QCA. To help ameliorate

163

this problem when confronted with larger number of possibly sufficient
conditions, the construction of artificial supersets was proposed. By combining
conditions that share features (i.e. types of war or political factors) using the
Boolean AND function, a logical superset can be created that describe a case’s
membership in all three individual subsets. These supersets can then be used in
fs/QCA analysis to identify potentially sufficient conditions, allowing insufficient
supersets to be dropped from further analysis without loss of analytical power.
The construction of supersets as a screening tool passed a proof of concept test by
identifying which sets of conditions would later prove sufficient in fs/QCA
analysis. It correctly identified, for the most part, when a superset component was
deemed sufficient in later fs/QCA analysis. This method experienced difficulty,
however, when both the absence and presence of subsets were both considered
sufficient in different components of an fs/QCA solution. More testing is needed
to determine why this is the case and under which conditions this screening
method fails, but the construction of artificial supersets could prove useful for
pilot studies in which the conditions associated with the outcome under
investigation are generally unknown.

164

Future Research
This project has revealed, if nothing else, that the decision for a state to
use or refrain from using their chemical weapons arsenal is a complicated one that
is not fully understood. More research is desperately needed. Greater attention
needs to be payed as to why states used their chemical weapons after WWII. As
mentioned above, systematic analysis using fs/QCA was unable to identify a set
of conditions consistently sufficient to result in this outcome. This is likely due to
at least one of two reasons. First, it is possible that the hypotheses proposed in this
study do not explain the modern use of chemical weapons, leading to at least one
omitted “variable.” It is expected that fs/QCA would return insufficient results if
the conditions most strongly associated with chemical weapons use were not
included in the fs/QCA analysis, as the above did. A second, non-exclusive,
possibility explaining the lack of fs/QCA results could be that each decision to
use chemical weapons in the post-WWII era was highly contextually dependent.
Instead of common triggers for this behavior, the use of chemical weapons could
be a response tailored to specific military and political needs. Determining if
either of these explanations are the cause of the fs/QCA null finding will require
several small-n case studies, specifically looking at the decision making process
employed by the leadership in each state.
A second avenue for future research involves determining the difference
between those states that choose to escalate their use of chemical weapons and

165

those that choose to stick with either lethal or non-lethal weapons. As mentioned
above, fs/QCA could not distinguish between those that escalated and those that
chose not to. Determining the turning point at which a state switches the chemical
agent it uses has an important practical salience, particularly with policy makers
that are faced with an actor that may choose to use any of a number of chemical
weapons at their disposal. If the conditions at the tipping point can be identified, it
creates a window of opportunity for those in the international community to
intervene and prevent mass atrocities. If policy makers identify that a set of
conditions associated with a state switching from non-lethal chemicals to lethal
chemical agents during a conflict, they have the option of intervening in that
conflict to stave off flagrant violations of the CWC. Instead of merely reacting to
chemical weapons use, leaders can be given the opportunity to actively prevent
them through proactive diplomacy.
A number of possible conditions could be associated with the escalation of
chemical weapons use. One reason a state might choose to escalate from non-
lethal to lethal chemical agents is that the state is involved in an active research
and design phase of their chemical weapons program. States with programs in this
level of development might be more susceptible to wanting to assess their use in
real combat. While laboratory analogues yield important information about how
chemical agents may act in the field, nothing beats testing these weapons in actual

166

combat for data acquisition. There is some evidence that testing was occurring in
several conflicts that saw the escalation of non-lethal to lethal chemical weapons.
Another reason why states may choose to escalate may hinge on the
response the state receives by the international community to their use of non-
lethal agents. Several states escalated their chemical weapons use only after a
lackluster response from the international system, in particular the United Nations
and League of Nations. The perception that the international community would
not become involved in the conflict even after violation of international treaties
may have been a permissive or necessary factor for regimes to switch from a non-
lethal agent to more lethal chemicals. Case studies need to be conducted on those
states that chose to escalate to understand if this was a factor in their choice,
paying careful attention to agent supply and the history of that country’s chemical
program.
More research into the effect different treaties had on chemical weapons
use is also needed. While it is common knowledge that the Chemical Weapons
Convention, similar to the Ottawa Treaty on landmines, has been a resounding
success at preventing chemical weapons use since its entry into force in 1997, a
reexamination of this assumption is needed in the light of empirical reality. Did
the average number of chemical weapons accusations decrease after 1997? Was a
potential decrease part of a larger decline in the popularity of the use of chemical

167

agents in warfare? Could nuclear weapons have played a role, relegating chemical
weapons as the weapon of second choice, the “poor man’s bomb?” While the
CWC was significantly more successful than the Geneva Protocol at preventing
chemical weapons use, it has experienced one recent failure. Within months of
signing the CWC in 2013, Syria violated its obligations under the treaty by
bombing civilians with industrial chemicals. Why did the treaty not serve as a
brake on Assad’s actions, especially if the treaty is considered such a strong one?
Instead of constructivist notions of appropriateness, could states be adhering to
the treaty for more realist notions of practicality?
From a public health perspective, the identification of countries which
have experienced chemical weapons usage opens up greater avenues of study into
the long-term burden of these agents. Several agents, including sulfur mustard,
have long-term and multigenerational negative health impacts, including an
increased risk of cancer, birth defects, and lung ailments (see Abolghasemi et al.
2010, 1-6; Ghanei et al. 2004, 495; Karalliedde et al. 2010, 238-248; Sasser,
Cushing, and Dacre 1996, 311-319; Zarchi, Akbar, and Naieni 2004, 579-581).
By increasing the population available for study, a greater understanding of the
society-wide impact chemical weapons use has on public health could be
achieved through ecological studies. The duration of these effects generation-wise
could also be determined. Does the effect of chemical weapons exposure decline
over time, or does it have some sort of fixed effect as many agents persist for

168

significant periods of time? Care would need to be taken to control for the effects
of migration, data acquisition problems, and approximate percentage of
population exposed, but this type of study could be useful in determining the
overall healthcare burden resulting from chemical weapons exposure.65F
66

Understanding how chemical exposure affects populations in the long run is
important information for domestic policy makers. Once the size of the additional
healthcare burden is estimated, these decision makers can increase the amount of
resources dedicated to the healthcare infrastructure. This will allow the state to
better meet the needs of its citizens in the long term, and potentially help create a
more stable society.
The last avenue of research stemming from the project would be to further
investigate the use of supersets as a screening tool for fs/QCA. Although it has
passed a proof of concept test, more work needs to be done to understand the
limits of this method. In particular, the following questions need to be answered.
Mathematically, why do supersets seem to experience difficulty in accurately
identifying whether the absence or presence of a set of conditions will be
sufficient when both are sufficient in different parts of the solution. Can the
construction of supersets be modified to avoid this eventuality? Should the

66
This line of research could be further extended to examine the effect of chemical production,
research, and design facilities on their surrounding populations, particularly those that experience
some sort of accident or exposure event.

169

algorithm used to conduct fs/QCA analysis be modified in some way to allow
higher accuracy? While supersets appear to have worked as indicators for this
particular dataset, will it perform as well in others? In addition, this method needs
to be tested against multiple datasets to ensure reproducibility as well as identify
where the method experiences difficulties.

Chemical Weapons: A Historical Evil and Continuing Problem?
Chemical weapons have had a consistent presence in the international
system since they were introduced to the battlefields of WWI. Since then, three
additional classes of lethal chemical weapons have been developed; the most
dangerous of these weapons are capable of killing a 200 lb. human adult with
approximately 0.6 mg. of agent within seconds. Because of their unique properties
that render them useful in specific tactical situations, it is unlikely that chemical
weapons will ever be eliminated from the earth. While both the US and Russia
will eventually finish destroying the thousands of metric tons of agent left from
their Cold War arsenal, there is little current incentive for those states that are
operating clandestine programs to surrender their arsenals. While there is
significant uncertainty on exactly which governments have clandestine programs
in the open literature, it is almost certain that both Israel and North Korea do.
Neither of these countries would be interested in surrendering their arsenals,

170

unless significant geopolitics-changing incentives were given in exchange. Those
states that have not yet developed chemical weapons are unlikely to do so, as the
vast majority of states are signatories of the CWC.
There are only three states that have not signed the CWC: Egypt, North
Korea, and South Sudan. It is unlikely that any of these states will sign the treaty
in the near future. Egypt is known to have possessed a chemical weapons program
in the past; they famously used chemical weapons in the Yemeni Civil War.
Extremely little has been known in open literature about the state of this program
for several decades. It is likely that given the huge investment the government put
into a chemical arsenal they kept some weapons, even if they mothballed the rest
of the program. Since Egypt is currently ruled by a conservative former military
commander, it is unlikely that Egypt will sign a treaty limiting its military
options. North Korea is a chemical and nuclear weapons possessor state. It is
currently an international pariah, and is perpetually concerned with its own
security vis-à-vis the United States and South Korea. Unless relations with South
Korea and the US were to improve drastically, it is unlikely that North Korea will
surrender either its chemical or nuclear stockpile. This is unlikely to happen, and
would probably require at least a partial dismantling of the North Korean regime.
South Sudan is the final non-signatory of the CWC. South Sudan is a relatively
new state, and only gained its independence in 2011. Since then, South Sudan has
been considered a very fragile state (topped the Fragile State Index in 2015 and

171

2014), with ongoing conflict, human rights abuse, and large populations of
displaced people making it difficult for the country to stabilize. Since this state is
so unstable, it is unlikely that the government will sign a largely optional
international agreement such as the CWC.
While most of the states in the world has surrendered their right to possess
chemical weaponry, it is likely that an increasing number of non-state actors will
seek to acquire and use chemical weapons. Chemical terrorism has occurred
sporadically since the 1990s, when Aum Shinrikyo famously used sarin in the
Tokyo subway system. While previous attempts at chemical and biological
terrorism by this group were unsuccessful and the Tokyo attack was relatively
crude, this Aum Shinrikyo demonstrated that with a minimal level of expertise
small groups could succeed at exposing large numbers of people to toxic chemical
agents. Since then, suicide bombers in Iraq have used chlorine bombs against both
military and civilian targets (2006- 2007). The Islamic State in Iraq and Syria
borrowed and improved on previous tactics in the ongoing Syrian civil war, and
launched several attacks against civilians in 2015 using chlorine filled shells. The
trend of using chlorine is particularly worrisome, as this agent is a readily
available industrial chemical and can be obtained in large quantities. The
threshold of expertise for chemical weapons use is lowering, making these
weapons more obtainable for small groups intent on wreaking havoc or fighting
against existing governments.

172

With the retention of chemical agents by select militaries worldwide and
an increasing attractiveness to non-state actors, it is likely that chemical weapons
will be used again in the future. Since WWI, there has been at least one
accusation of chemical weapons use by a state in each decade, and it is likely that
this trend will continue. It is slightly more likely that the next chemical weapons
attack will be by a non-state actor, rather than a state. This is due to the increase
in successful terrorist attacks perpetrated in Europe, the United States, and the
Middle East. It is now feasible for small terrorist cells to manufacture at least
small quantities of chemical agents and deliver them to soft targets with high
concentrations of vulnerable civilians. This is the current nightmare for security
officials the world over, and is not an eventuality that can be easily guarded
against. If the next chemical weapons user is not a non-state actor, it is likely than
an authoritarian regime, such as that of Assad in Syria, will be the next state to
use chemical weapons. The disorganized and weak international response to the
Syrian regime’s use of chemical weapons twice in a civil war sent a signal to
other autocratic governments the world over. One can only hope that these
governments do not emulate the Assad regime; if they do, only a coherent, severe
international response can hope to stave off future chemical weapons use.

173

References

174

Federal Meat Inspection Act, (1906): United States Statutes at Large Vol. 34, sec.
678.
Pure Food and Drug Act (1906), (1906): United States Statutes at Large Vol. 34,
sec. 3915. 768.
"Congressional Record, 69th Congress, 2nd Session." 1927 68 (1):150.
"Appeal of the Chinese Government in Virtue of the Covenent (Continuation)."
1932. Official Journal of the League of Nations 3 (1) (March):361-72.
Anonymous. 1935, "Abyssinian Casualties from Poison Gas," The Times, October
16, sec News 14.
Anonymous. 1937a, "Alleged use of Gas by Japanese," The Times, October 8, sec
News 13.
Anonymous. 1937b, "Chinese Bombed in Boats," The Times, October 9, sec
News 12.
Anonymous. 1937c, "Epidemics in China," The Times, October 15, sec News 15.
Anonymous. 1937d, "The Japanese Advance," The Times, October 12, sec News
14.

175

Anonymous. 1937e, "Japan's Next Move in the North," The Times, August 30, sec
News 10.
Anonymous. 1938a, ""Two Battalions Killed by Gas"," The Times, August 26, sec
News 10.
Anonymous. 1938b, "Alleged use of Gas by Japan," The Times, August 10, sec
News 12.
"Appeal by the Chinese Government." 1938a. Official Journal of the League of
Nations (November):863-5.
"Appeal of the Chinese Government." 1938b. Official Journal of the League of
Nations 19 (November):863-5.
"Appeal of the Chinese Government (Continuation)." 1938. Official Journal of
the League of Nations 19 (May-June):378-80.
"Appeal of the Chinese Government (Continuation)." 1938. Official Journal of
the League of Nations 19 (November):878-81.
Anonymous. 1942a, "Chinese Retire from Kinhwa," The Times, June 1, sec News
4.

176

Anonymous. 1942b, "Counter-Blow in Burma," The Times, March 31, sec News
4.
Anonymous. 1942c, "Kinhwa Under Siege," The Times, May 29, sec News 4.
Anonymous. 1943a, "Costly Attack on Changteh," The Times, November 27, sec
News 3.
Anonymous. 1943b, "Japanese Accused of using Gas," The Times, November 6,
sec News in Brief 3.
"Chemical Weapons in the Middle East." 1991a. Arms Control Today (May).
"Fact File: Chemical Weapons in the Middle East." 1991b. Arms Control Today
(May).
"News Chronology: November 1994-February 1995." 1995. Chemical Weapons
Convention Bulletin 28 (June):12-34.
Anonymous. 2005, "Prisoner Sues Over Fingertip in Food," The Los Angeles
Times, July 23. http://articles.latimes.com/2005/jul/23/local/me-finger23.
Anonymous. 2012, "Obama Warns Syria Over Chemical Weapons use," Al
Jazeera, December 4.

177

Anonymous. 2013a, "Prime Minister of the United Kingdom David Cameron
Holds A Discussion on Syria," CQ Transcriptions, August 29 (accessed
February 7, 2014).
Anonymous. 2013b, "S. Korea Slams Syria for Apparent use of Chemical
Weapons," Yonhap News Agency, September 1.
United Nations Security Council. Resolution 2209 (2015). S/RES/2209. 7401
sess. (March 6, 2015).
5250th Technical Intelligence Company. 1946. The use of poison gas by imperial
japanese army in china, 1937-1945. Tokyo, Japan: 5250th Technical
Intelligence Company.
Abbott, A. 1995. "Sequence Analysis: New Methods for Old Ideas." Annual
Review of Sociology 21 :93-113.
———2001. Time matters: On theory and method. Chicago: University of
Chicago Press.
Abolghasemi, Hassan, Mohammad H. Radfar, Mehdi Rambod, Parvin Salehi,
Hossein Ghofrani, Mohammad R. Soroush, Farahnaz Falahaty, et al. 2010.
"Childhood Physical Abnormalities Following Paternal Exposure to Sulfur
Mustard Gas in Iran: A Case-Control Study." Conflict and Health 4 (13):1-6.

178

Ali, Javed. 2001. "Chemical Weapons and the Iran-Iraq War: A Case Study in
Noncompliance." The Nonproliferation Review (Spring):43-58.
Altman, David, and Anibal Perez-Linan. 2002. "Assessing the Quality of
Democracy: Freedom, Competativeness and Participation in Eighteen Latin
American Countries." Democratization 9 :85-100.
Badoglio, Pietro. 1937. The war in abyssinia. London: Methuen.
Bahgat, Gawdat. 2005. "Nuclear Proliferation and the Middle East." The Journal
of Social, Political, and Economic Studies 30 (4) (Winter):401-24.
Peter Baker. 2012, "Obama Says ‘World Set a Red Line’ on Chemical Arms," The
New York Times, September 4, sec Europe.
Peter Baker, and Michael R. Gordon. 2012, "U.S. Warns Syria on Chemical
Weapons," The New York Times, December 3.
Bale, Jeffrey. 2006. "South Africa's Project Coast: 'Death Squads,' Covert State-
Sponsored Poisonings, and the Dangers of CBW Proliferation." Democracy
and Security 2 (1) (July):27-59.
Baskerville, Beatrice. 1937. What next, O duce?. London: Longmans.

179

Baum, Seth D. 2015. "Winter-Safe Deterrence: The Risk of Nuclear Winter and
its Challenge to Deterrence." Contemporary Security Policy 36 (1):123-48.
Beaton, Leonard, and John Maddox. 1962. The spread of nuclear weapons. New
York: Praeger.
Bennett, Andrew. 2010. "Process Tracing and Causal Inference." In Rethinking
social inquiry: Diverse tools, shared standards, eds. David Collier, Henry E.
Brady. Lanham, MD: Rowman and Littlefield.
Berg-Schlosser, Dirk, Gisele De Meur, Benoit Rihoux, and Charles C. Ragin.
2009. "Qualitative Comparative Analysis (QCA) as an Approach." In
Configurational comparative methods: Qualitative comparative analysis
(QCA) and related techniques, eds. Benoit Rihoux, Charles Ragin. Los
Angeles: Sage Publications, Inc.
Betts, Richard K. 1977. "Paranoids, Pygmies, Pariahs and Non-Proliferation."
Foreign Policy 26 :157-83.
Black, Robin M., Raymond J. Clarke, Robert W. Read, and Michael T. J. Reid.
1994. "Application of Gas Chromatography Mass Spectrometry and Gas
Chromatography Tandem Mass Spectrometry to the Analysis of Chemical
Warfare Samples, found to Contain Residues of the Nerve Agent Sarin,

180

Sulphur Mustard and their Degradation Products." Journal of
Chromatography 662 :301-21.
Black, Robin M., and Graham S. Pearson. 1993. "Unequivocal Evidence."
Chemistry in Britain (July).
Bolinger, Bwight. 1980. Language the loaded weapon: The use & abuse of
language today. New York: Longman.
Bollen, K. A., B. Entwisle, and A. S. Alderson. 1993. "Macrocomparative
Research Methods." Annual Review of Sociology 19 :321-51.
Brambor, Thomas, William Clark, and Matt Golder. 2006. "Understanding
Interaction Models: Improving Empirical Analyses." Political Analysis 14
:63-82.
Brecher, Michael, and Jonathan Wilkenfeld. 2007. International crisis behavior
project, 1918-2004. Ann Arbor, MI: Inter-university Consortium for Political
and Social Research (ICPSR).
Brown, Frederic J. 1968. Chemical warfare: A study in constraints. Princeton, N.
J.: Princeton University Press.

181

Burck, Gordon M., and Charles C. Flowerree. 1991. International handbook on
chemical weapons proliferation. New York: Greenwood Press.
Center for Nonproliferation Studies. Monterey Institute of International Studies.
2015a. "Egypt: Chemical Overview." Nuclear Threat Initiative
http://www.nti.org/country-profiles/egypt/chemical/ August.
———2015b. "Iran: Chemical Overview." Nuclear Threat Initiative
http://www.nti.org/country-profiles/iran/chemical/ August.
Chi-tsai, Hoo. 1938. "Letter Dated August 5th, 1938, from the Chinese
Government to the Secretary General." Official Journal of the League of
Nations (August-September):665-6.
———. 1938. "Letter, Dated August 24th, 1938, from the Chinese Government
to the Secretary-General." Official Journal of the League of Nations (August-
September):666.
———. 1938. "Letter, Dated September 5th, 1938, from the Chinese Government
to the Secretary General." Official Journal of the League of Nations (August-
September):667-8.

182

Cialdini, R. B., R. R. Reno, and C. A. Kallgren. 1990. "A Focus Theory of
Normative Conduct: Recycling the Concept of Norms to Reduce Littering in
Public Places." Journal of Personality and Social Psychology 58 :1015-26.
Clark, William R., Michael J. Gilligan, and Matt Golder. 2006. "A Simple
Multivariate Test for Asymmetric Hypotheses." Political Analysis 14 :311-
31.
Lesley Clark, and McClatchy Washington Bureau. 2013, "Tough Sell: Obama
Heads to Russia Summit Looking for Support of Syria Strike," McClatchy
Washington Bureau, September 3 (accessed February 7, 2014).
Cole, Leonard A. 1998. "The Poison Weapons Taboo: Biology, Culture, and
Policy." Politics and the Life Sciences 17 (2) (September):119-32.
Coleman, Kim. 2005. A history of chemical warfare. London: Palgrave
Macmillan.
Cordesman, Anthony H. 2004. The proliferation of weapons of mass destruction
in the middle east. Washington, DC: Center for Strategic and International
Studies.
Crocker, Jennifer, and Brenda Major. 1989. "Social Stigma and Self-Esteem: The
Self-Protective Properties of Stigma." Psychological Review 96 (4):608-30.

183

Day, Jon. August 27, 2013. Syria: Reported chemical weapons use. London: Joint
Intelligence Committee of the United Kingdom, Jp 115.
de Jonge, Klaas. November 1997. "The Chemical Warfare Case." In The (secret)
truth commission files, Anonymous.
De Meur, G., and Benoit Rihoux. 2002. L'analyse quali-quantitative comparée
(AQQC-QCA): Approache, techniques et applications en sciences humaines.
Louvain-la-Neuve, Belgium: Academia-Bruylant.
De Meur, Gisele, Benoit Rihoux, and Sakura Yamasaki. 2009. "Addressing the
Critiques of QCA." In Configurational comparative methods: Qualitative
comparative analysis (QCA) and related techniques, eds. Benoit Rihoux,
Charles Ragin. Los Angeles: Sage Publications, Inc.
Del Boca, Angelo. 1969. The ethiopian war 1935-1941. Chicago: The University
of Chicago Press.
———April 5, 1965. Personal interview with tullio pastori, Asmara.
Dempsey, Judy. 2013. "Merkel and intervention in syria: Damned either way."
EurActiv.Com. August 28.

184

Leonard Doyle, Donald Macintyre, and Tom Wilkie. 1991, "Saddam's Nerve Gas
Secrets," The Independent (London), August 4 1.
Dunn, Lewis A., and Herman Kahn. 1976. Trends in nuclear proliferation, 1975-
1995. Croton-on Hudson, NY: Hudson Institute.
Eisenstadt, Michael. 1998. Chemical and biological weapons and deterrence case
study 4: Iran. Alexandria, VA: Chemical and Biological Arms Control
Institute.
Eliason, Scott R., and Robin Stryker. 2009. "Goodness-of-Fit Tests and
Descriptive Measures in Fuzzy-Set Analysis." Sociological Methods and
Research 38 :102-46.
Fauvet, Paul. 1998. "Mozambican Claims on 1992 Chemical Attack Now Appear
Correct." The Star (June 17).
Fedorov, Lev A. 1994. Chemical weapons in russia: History, ecology, politics.
Moscow: Center of Ecological Policy of Russia.
Finnemore, Martha. 1996. National interests in international society. Ithaca:
Cornell University Press.

185

Finnemore, Martha, and Kathryn Sikkink. 1998. "International Norm Dynamics
and Political Change." International Organization 52 (4) (Autumn):887-917.
Fuhrmann, M. 2009. "Spreading Temptation: Proliferation and Peaceful Nuclear
Cooperation Agreements." International Security 34 (1):7-41.
George, A. L., and A. Bennett. 2005. Case studies and theory development in the
social sciences. Cambridge: MIT Press.
Gerring, J. 2005. "Causation: A Unified Framework for the Social Sciences."
Journal of Theoretical Politics 17 (2):163-98.
Ghanei, Mostafa, Hosnieh Fathi, Mohammad M. Mohammad, Jafar Aslani, and
Fariborz Nematizadeh. 2004. "Long-term respiratory disorders of
claimers with subclinical exposure to chemical warfare agents." Inhalation
Toxicology 16 (491):495.
Goldthorpe, J. H. 1997. "Current Issues in Comparative Macrosociology: A
Debate on Methodological Issues." Comparative Social Research 16 :1-26.
Hennekens, Charles H., and Julie E. Buring. 1987. Epidemiology in medicine.
Boston: Little, Brown and Company.

186

Herby, Peter. 1992. The chemical weapons convention and arms control in the
middle east. Oslo: International Peace Research Institute.
Hilmas, Corey J., Jefferey Smart K., and Benjamin A. Hill. 2008. "History of
Chemical Warfare." In Medical aspects of chemical warfare, ed. Martha K.
Lenhart. Washington, D.C.: Office of the Surgeon General of the United
States Army, 9-76.
Hogendoorn, E. J. 1997. "A Chemical Weapons Atlas." Bulletin of the Atomic
Scientists 53 (5).
Hume, David. 1758. An enquiry concerning human understanding. Chicago:
Open Court Publishing, Co.
Jefferson, Catherine. 2009. "The Taboo of Chemical and Biological Weapons:
Nature, Norms and International Law." DPhil Science & Technology Policy.
University of Sussex.
Jepperson, Ronald, Alexander Wendt, and Peter J. Katzenstein. 1996. "Norms,
Identity, and Culture in National Security." In The culture of national
security, ed. Peter J. Katzenstein. New York: Columbia University Press.
Jo, Dong-Joon, and Erik Gartzke. 2007. "Determinants of Nuclear Weapons
Proliferation." The Journal of Conflict Resolution 51 (1):167-94.

187

Johnson, Nathan H., Joseph C. Larsen, and Edward Meek. 2015. "Historical
Perspective of Chemical Warfare Agents." In Handbook of toxicology of
chemical warfare agents, ed. Ramesh C. Gupta. London: Academic Press, 7-
16.
Joy, R. J. T. 1997. "Historical Aspects of Medical Defense Against Chemical
Warfare." In Medical aspect of chemical and biological warfare,
Anonymous. US Government Printing Office, Washington, D.C.: Office of
the Surgeon General, Department of the Army, 87-109.
Kam, Cindy, and Robert Franzese. 2007. Modeling and interpreting interactive
hypotheses in regression analysis. Ann Arbor, MI: University of Michigan
Press.
Karalliedde, L., H. Wheeler, R. Maclehose, and V. Murray. 2010. "Possible
Immediate and Long-Term Health Effects Following Exposure to Chemical
Warfare Agents." Public Health 114 (4) (July):238-48.
Katzenstein, Peter J. 1996a. Cultural norms and national security: Police and
military in postwar japan. Ithaca: Cornell University Press.
———1996b. Introduction: Alternative perspectives on national security. New
York: Columbia University Press.

188

Kerr, Norbert L., and John M. Levine. 2008. "The Detection of Social Exclusion:
Evolution and Beyond." Group Dynamics: Theory, Research, and Practice
12 (1):39-52.
King, G., Robert O. Keohane, and S. Verba. 1994. Designing social inquiry:
Scientific inference in qualitative research. Princeton, NJ: Princeton
University Press.
Klotz, Audie. 1995. Norms in international relations: The struggle against
apartheid. Ithaca: Cornell University Press.
Koo, V. K. W. 1937. "Communication, Dated September 12th, 1937, from the
Chinese Government to the Secretary-General." Official Journal of the
League of Nations 18 (August-September):655-7.
———. 1938. "Letter, Dated may 9th, 1938, from the Chinese Government to the
Secretary-General." Official Journal of the League of Nations 19 (May-
June):381.
———. 1938. "Letter, Dated September 16th, 1938, from the Chinese
Government to the Secretary-General." Official Journal of the League of
Nations (December):1133-5.

189

Kroenig, M. 2009. "Exporting the Bomb: Sensitive Nuclear Assistance and
Proliferation." American Political Science Review 103 (1):113-33.
Krogslund, Chris, Donghyun D. Choi, and Mathias Poertner. 2015. "Fuzzy Sets
on Shaky Ground: Parameter Sensitivity and Confirmation Bias in fsQCA."
Political Analysis 23 :21-41.
Krogslund, Chris, and Katherine Michel. 2014a. "A Larger-N, Fewer Variables
Problem? the Counterintuitive Sensitivity of QCA." Qualitative & Multi-
Method Research 12 (1) (Spring).
——— 2014b. "Testing the Ability of Set-Theoretic Methods to Recover Data
Generating Processes." Presented at the Midwest Political Science
Association 72nd Annual Conference, Chicago, Illinois.
Kurzban, Robert, and Mark R. Leary. 2001. "Evolutionary Origins of
Stigmatization: The Functions of Social Exclusion." Psychological Bulletin
127 (2):187-208.
Lakoff, Robin. 1973. "Language and Women's Place." Language in Society 2 (1)
(April):45-80.
Lapinski, Maria K., and Rajiv N. Rimal. 2005. "An Explication of Social Norms."
Communications Theory 15 (2) (May):127-47.

190

Legewie, Nicolas. 2013. "An Introduction to Applied Data Analysis with
Qualitative Comparative Analysis (QCA)." Forum: Qualitative Social
Research 14 (3) (September).
I. Lundberg. 1969, "Rapport Om Ett Gaskrig," Aftonbladet, February 5 4.
Mahoney, J. 2003. "Strategies of Causal Assessment in Comparative Historical
Analysis." In Comparative historical research, eds. J. Mahoney, D.
Rueschemeyer. Cambridge, UK: Cambridge University Press, 337-372.
———. 2004. "Revisiting General Theory in Historical Sociology." Social
Forces 83 (2):459-89.
———. 2010. "What is a Concept? Two Definitions and their Research
Implications." In Mimeo, Anonymous. Evanston, IL: Northwestern
University.
Mandelbaum, M. 1981. The nuclear revolution: International politics before and
after hiroshima. New York: Cambridge University Press.
Maryam, Wolde. 1936. "Letter, Dated may 2nd, 1936, from the Ethiopian
Representative to the Secretary-General." The Official Journal of the League
of Nations Annex 1597 (June):605-43.

191

McCarthy, Claudine. 2005. "Japan and WMD." In Weapons of mass destruction:
An encyclopedia of worldwide policy, technology, and history, eds. Eric A.
Croddy, James J. Wirtz and Jeffrey A. Larsen. Santa Barbara: ABC-CLIO.
McLean, N. 1965. "The War in Yemen." Presented at the Lecture given at the
Royal United Services Institute, October 20, 1965, London.
McNeill, Daniel, and Paul Freiberger. 1993. Fuzzy logic. New York: Touchstone.
Mill, J. S. 1967 [1843]. A system of logic: Ratiocinative and inductive. Toronto:
University of Toronto Press.
Ministero dell'Africa Italiana. 1935. 263/97, de bono to graziani, october 16,
1935, telegram no. 14323/6748. October 16. Archivio Segreto di Gabinetto.
Ministero delle Colonie. 1935. Ministero delle colonie to graziani, telegram no.
11889. October 17. Archivio Segreto di Gabinetto.
Mueller, John. 1989. Retreat from doomsday: The obsolescence of major war.
New York: Basic Books, Inc.
Muller, Harald. 2013. "Where it all Began." In Studies in security and
international affairs: Norm dynamics in multilateral arms control: Interests,

192

conflicts, and justice, eds. Harald Muller, Carmen Wunderlich. Athens, GA:
University of Georgia Press.
North, D. 1990. Institutions, institutional change and economic performance.
Cambridge: Cambridge University Press.
Organization for the Prohibition of Chemical Weapons. Office of the Director-
General. December 18, 2014. Note by the technical secretariat: Third report
of the OPCW fact-finding mission in syria. The Hague: Organization for the
Prohibition of Chemical Weapons.
Owen, Wilfred. 1963. "Dulce Et Decorum Est." In The collected poems of wilfred
owen, Anonymous. New York: Chatto & Windus, Ltd.
Pierson, P. 2003. "Big, Slow Moving, and...Invisible: Macrosocial Processes in
the Study of Comparative Politics." In Comparative historical research, eds.
J. Mahoney, D. Rueschemeyer. Cambridge, UK: Cambridge University
Press, 177-207.
Potter, William C. 1982. Nuclear power and nonproliferation. Cambridge, MA:
Oeleschlager, Gunn, and Hain.
Pozdnyakov, V. V. 1956 In The red army, ed. B. H. L. Hart. London.

193

Price, Richard M. 1995. "A Genealogy of the Chemical Weapons Taboo."
International Organization 49 (1) (Winter):73-103.
———1997. The chemical weapons taboo. Ithaca: Cornell University Press.
Price, Richard M., and Nina Tannenwald. 1996. "Norms and Deterrence: The
Nuclear and Chemical Weapons Taboos." In The Culture of national
security: Norms and identity in world politics, ed. Peter J. Katzenstein. New
York: Columbia University Press.
Quester, George H. 1973. The politics of nuclear proliferation. Baltimore: Johns
Hopkins University Press.
———2005. Nuclear first strike: Consequences of a broken taboo. Baltimore:
Johns Hopkins University Press.
Radcliffe-Brown, A. R. 1939. Taboo. Cambridge: Cambridge University Press.
Ragin, Charles C. 2000. Fuzzy-set social science. Chicago, IL: University of
Chicago Press.
———. 2006. "Set Relations in Social Research: Evaluating their Consistency
and Coverage." Political Analysis 14 (3):291-310.

194

Ragin, Charles C., Kriss A. Drass, and Sean Davey. 2006. "Fuzzy-Set/Qualitative
Comparative Analysis." 2.0.
———2009. Fuzzy-set/qualitative comparative analysis 2.5. Tucson, AZ:
Department of Sociology, University of Arizona.
Ragin, Charles C., and Claude Rubinson. "Fuzzy Set/Qualitative Comparative
Analysis: What is QCA?"
http://www.u.arizona.edu/~cragin/fsQCA/index.shtml March 16.
Republic of South Africa. National Intelligence Service. December 1992. Staff
paper prepared for the steyn commission on alleged dangerous activities of
SADF components. Annexure B.
Rihoux, Benoit, and Charles C. Ragin, eds. 2009. Configurational comparative
methods: Qualitative comparative analysis (QCA) and related techniques.
Los Angeles: Sage Publications, Inc.
Robinson, Julian P., and Jozef Goldblat. 1984. "Chemical Warfare in the Iraq-Iran
War." Stockholm International Peace Research Institute Fact Sheet (May).
Rochat, Giorgio. 2005. "The Italian Air Force in the Ethiopian War (1935-1936)."
In Italian colonialism, eds. Ruth Ben-Ghiat, Mia Fuller. New York: Palgrave
Macmillan, 37-46.

195

Romaine, Suzanne. 2000. Language in society: An introduction to
sociolinguistics. Cary, NC: Oxford University Press.
Rosen, Steven J. 1975. "Nuclearization and Stability in the Middle East." In
Nuclear proliferation and near-nuclear countries, eds. Onkar Marwah, Ann
Schulz. Cambridge, MA: Ballinger, 157-184.
RSA, National Intelligence Service. December 1992. Staff paper prepared for the
steyn commission on alleged dangerous activities of SADF components.
Annexure B.
Rueschemeyer, D., and J. D. Stephens. 1997. "Comparing Historical Sequences- a
Powerful Tool for Causal Analysis. A Reply to John Goldthorpe's "Current
Issues in Comparative Macrosociology. “.” Comparative Social Research 16
:55-72.
Sagan, Scott D. 2011. "The Causes of Nuclear Proliferation." The Annual Review
of Political Science 14 :225-244.
Sarkees, Meredith R., and Frank Wayman. 2010. Resort to war: 1816 - 2007.
Washington D. C.: CQ Press.
Sartori, Giovanni. 1970. "Concept Misinformation in Comparative Politics."
American Political Science Review 64 :1033-53.

196

Sasser, L. B., J. A. Cushing, and J. C. Dacre. 1996. "Two-Generation
Reproduction Study of Sulfur Mustard in Rats." Reproductive Toxicology 10
(4) (July-August):311-9.
Sbacchi, Alberto. 2005. "Poison Gas and Atrocities in the Italo-Ethiopian War
(1935-1936)." In Italian colonialism, eds. Ruth Ben-Ghiat, Mia Fuller. New
York: Palgrave Macmillan, 47-56.
Eric Schmitt, and David E. Sanger. 2013, "Hints of Syrian Chemical Push Set Off
Global Effort to Stop it," The New York Times, January 7.
Schneider, Carsten Q., and Claudius Wagemann. 2007. Qualitative comparative
analysis (QCA) und fuzzy sets. ein lehrbuch für anwender und jene, die es
werden wollen. Opladen/ Farmington Hills, MI: Barbara Budrich.
———2012. Set-theoretic methods for the social sciences: A guide to qualitative
comparative analysis. Cambridge: Cambridge University Press.
Selassie, Haile. 1949. Documents on italian war crimes submitted to the united
nations war crimes commission by the imperial ethiopian government. Addis
Ababa: Ministry of Justice.

197

Singer, J. D., Stuart Bremer, and John Stuckey. 1972. "Capability Distribution,
Uncertainty, and Major Power War, 1820-1965." In Peace, war, and
numbers, ed. Bruce Russett. Sage: Beverly Hills, 19-48.
Singh, Sonali, and Christopher R. Way. 2004. "The Correlates of Nuclear
Proliferation: A Quantitative Test." The Journal of Conflict Resolution 48
(6):859-885.
Hannah Lucinda Smith. 2015, "Assad 'Dropping Chemical Barrel Bombs'," The
Times, February 25, sec Middle East.
Spiers, Edward M. 2010. A history of chemical and biological weapons. London:
Reaktion Books, Ltd.
Steiner, Franz. 1956. Taboo. New York: Philosophical Library.
Stepanov, A. A., and J. N. Popov. 1962. Chemical weapons and principles of
anti-chemical defence. Moscow.
Stephens, J. D. 1998. "Historical Analysis and Causal Assessment in Comparative
Research." APSA-CP. Newsletter of the Organized Section in Comparative
Politics 9 (1):22-5.

198

Stockholm International Peace Research Institute. 1971. The rise of CB weapons.
New York: Humanities Press.
Sunstein, Cass. 1997. Free markets and social justice. New York: Oxford
University Press.
Tannenwald, Nina. 2007. The nuclear taboo: The united states and the non-use of
nuclear weapons since 1945. Cambridge: Cambridge University Press.
Thiem, Alrik. 2010. "Set-Relational Fit and the Formulation of Transformational
Rules in fsQCA." COMPASSS Working Paper Series 61.
Thomas, Ann V. W., and A. J. Thomas Jr. 1970. Legal limits on the use of
chemical and biological weapons. Dallas: Southern Methodist University
Press.
Thomas, William I., and Dorothy S. Thomas. 1928. The child in america:
Behavior problems and programs. New York: A. A. Knopf.
Truth and Reconciliation Commission. "Testimony of Jan Lourens." In Hearings
on south africa's chemical and biological warfare programme. Anonymous.
Tucker, Jonathan B. 2006. War of nerves: Chemical warfare from world war I to
al-qaeda. New York: Pantheon Books.

199

United Nations. 1993. Report of the investigations into the allegations of the use
of chemical weapons in mozambique. New York: United Nations.
United Nations Mission to Investigate Allegations of the Use of Chemical
Weapons in the Syrian Arab Republic. 2013a. Report on the alleged use of
chemical weapons in the ghouta area of damascus on 21 august 2013.
Secretary General of the United Nations.
———2013b. United nations mission to investigate Allegations of the use of
chemical weapons in the syrian arab republic: Final report.
United States of America. 2013. Assessment of the syrian government's use of
chemical weapons on august 21, 2013. Washington, D.C.: U.S. Government.
United States of America. Department of Defense. January 2001. Proliferation:
Threat and response. Office of the Secretary of Defense.
United States of America. Department of State. March 9, 1992. Recent chemical
weapons (CW) use allegations - africa memo (declassified). Department of
State.
Van Courtland Moon, John. 1984. "Chemical Weapons and Deterrence: The
World War II Experience." International Security 8 (4) (Spring):3-35.

200

———. 2008. "The Development of the Norm Against the use of Poison: What
Literature Tells Us." Politics and the Life Sciences 27 (1) (March):55-77.
Van Evera, Stephen. 1997. Guide to methods for students of political science.
Ithaca: Cornell University Press.
D. M. van Rosmalen. 1967, "Yemen- a Testing Ground for Poison Gases,"
Elseviers' Weekblad, November 25.
Waitt, Alden H. 1922. Gas warfare: The chemical weapon, its use, and protection
against it. New York: Duell, Sloan and Pearce.
Webster, Hutton. 1942. Taboo: A sociological study. Stanford: Stanford
University Press.
Werth, Nicolas. 1999. "From Tambov to the Great Famine." In The black book of
communism: Crimes, terror, repression, eds. Stéphane Courtois, Nicolas
Werth, Jean-Louis Panné, Andrzej Paczkowshi, Karel Bartošek and Jean-
Louis Margolin. Cambridge: Harvard University Press, 108-131.
World Bank. 2014. World development indicators. Washington, D. C.: World
Bank.

201

John Yeld. 1998, "SADF Bombed Frelimo in Chemical Weapons Test and
Blamed ANC," The Cape Argus, June 17.
Zadeh, Lofti A. 1995. "Probability Theory and Fuzzy Logic are Complementary
rather than Competitive." Technometrics 37 :271-6.
Zarchi, Karimi, Ali Akbar, and Kourosh H. Naieni. 2004. "Long-Term Pulmonary
Complications in Combatants Exposed to Mustard Gas: A Historical Cohort
Study." International Journal of Epidemiology 33 (3):579-81.

202

Appendix A
List of Figures

203

Figure 1. Definition of a superset

B

A
B

204

Figure 2. Superset S with subsets A, B, C

S
A
B
C

205

Figure 3. Effect of superset creation on explanatory condition scores

0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
Outcome
Explanatory Conditions

206

Figure 4. Number of times a state was accused of using chemical weapons

Note: The total number of states included in the dataset is 46. The total number of
accusations of chemical weapons use is 43.

0
5
10
15
20
25
0 1 2 3 4 5
Number of States
Times State was Accused of Chemical Weapons Use
Frequency

207

Figure 5. Chemical weapons use by region

0
2
4
6
8
10
12
14
16
18
Number of Chemical Weapons Use
Allegations
Region
Chemical Weapons Used In Chemical Weapons Used By

208

Figure 6. Chemical weapons use by and against the state

Note: Times Accused of Chemical Weapons Use statistic is the number of times the state has been accused of using chemical weapons. The
Times Chemical Weapons Used On By An Outside State statistic is the number of times an outside state has used chemical weapons against the
designated state. State territories were not counted as part of the state. They form their own “state” label. This was done in order to preserve the
distinction between the owner of territories and the territories themselves for theoretical purposes.
0
1
2
3
4
5
Afghanistan
Albania
Algeria
Angola
Australia
Belarus
Bosnia
Cambodia
Chad
China
Cuba
Egypt
Ethiopia
France
Germany
Greece
India
Iran
Iraq
Israel
Italy
Japan
Laos
Libya
Moroccan Sahara
North Korea
Poland
Portugal
Portuguese Guinea
Rhodesia
Russia
Serbia
Serbia
South Africa
Spain
Spanish Morocco
Sudan
Syria
Ukraine
United Kingdom
United States
USSR
Uzbekistan
Vietnam
Yemen
Yugoslavia
NUMBER OF ACCUSATIONS
STATE
Times Chemical Weapons Used On By An Outside State Times Accused of Chemical Weapons Use

209

Figure 7. Chemical weapons usage by location in or out of the state

Note: This chart depicts the location in which a state has been accused of using chemical weapons. Black bars denote that the state used
chemical weapons within its borders; grey bars indicate that the state used chemical weapons outside its borders. Mandates and colonies are not
considered inside the colonial overlord’s borders.
0
1
2
3
4
5
Afghanistan
Albania
Algeria
Angola
Australia
Belarus
Bosnia
Cambodia
Chad
China
Cuba
Egypt
Ethiopia
France
Germany
Greece
India
Iran
Iraq
Israel
Italy
Japan
Laos
Libya
Moroccan Sahara
North Korea
Poland
Portugal
Portuguese Guinea
Rhodesia
Russia
Serbia
Serbia
South Africa
Spain
Spanish Morocco
Sudan
Syria
Ukraine
United Kingdom
United States
USSR
Uzbekistan
Vietnam
Yemen
Yugoslavia
NUMBER OF ACCUSATIONS
STATE
Times Accused Use Within Own Borders Times Accused Use Outside Borders

210

Figure 8. Sufficient conditions for the outcome ‘chemical weapons use’

211

Figure 9. Sufficient conditions for the outcome ‘non-lethal weapons first

212

Figure 10. Necessary conditions for the outcome ‘chemical weapons use’ in 1918
Dataset

213

Figure 11. Sufficient conditions for the outcome ‘non-lethal weapons first’ in
1918 dataset

214

Figure 12. Sufficient conditions for the outcome ‘non-use of non-lethal weapons
first’

215

Appendix B
List of Tables

216

Table 1. List of instances of chemical weapons use and their acceptance status
Year Used By Used in
Location
Conflict Acceptance
Status
1918-
1921
USSR USSR
Various
Rebellions
Mixed
Acceptance
1919
Great
Britain
USSR
Russian Civil
War
Accepted
1920
Great
Britain
Iraq Insurgency Contested
1921-
1927
Spain and
France
Spanish
Morocco
Rif War Accepted
1929-
1930
Italy Libya
Rebellion
against Italian
rule
Accepted
Early
1930s
China Manchuria Insurgency Contested
1930 Japan Taiwan Insurgency Accepted
1935-
1936
Italy Ethiopia
Second Italo-
Abyssinian War
Accepted
1936 Spain Spain
Spanish Civil
War
Contested
1937-
1945
Japan
China, Burma,
Singapore
Second Sino-
Japanese War
Accepted
1939
Great
Britain
Germany World War II Disconfirmed
1942 Germany Poland World War II Accepted
1942 Germany Crimea World War II Contested
1945-
1949
China China
Chinese Civil
War
Contested
1947 France Vietnam Insurgency Contested
1948 Israel Israel Insurgency Contested
1949 Greece Greece Insurgency Accepted
1951 US
North Korea
and China
Korean War Contested
1957 Cuba Cuba Insurgency Contested

217

1957 France Algeria Insurgency Contested
1958
France and
Spain
Moroccan
Sahara
Insurgency Contested
1958 China China Insurgency Contested
1963-
1967
Egypt Yemen
Yemen Civil
War
Accepted
1964-
1975
US Vietnam Vietnam War Accepted
1968 Portugal
Portuguese
Guinea
Insurgency Contested
1969 Israel Israel Insurgency Contested
1970 Rhodesia Rhodesia Insurgency Contested
1975
Vietnam
and USSR
Laos Vietnam War Disconfirmed
1978 Vietnam Cambodia Insurgency Contested
1978
South
Africa
Angola
Cross-border
Raid
Contested
1979 USSR Afghanistan Afghan War Disconfirmed
1983-
1988
Iraq Iran Iran-Iraq War Accepted
1983-
1988
Iran Iraq Iran-Iraq War Contested
1984-
1985
Vietnam Cambodia
Dry season
offensive
Disconfirmed
1987 Libya Chad
Chadian-Libyan
Conflict
Accepted
1988 Iraq Iraq Insurgency Accepted
1991 Yugoslavia Croatia
Breakup of
Yugoslavia
Contested
1992
South
Africa
Mozambique*
Cross-border
Raid
Contested
1992-
1999
Serbia Bosnia
Breakup of
Yugoslavia
Contested
1995
Iraq and
Sudan
Sudan Insurgency Contested

218

1999 India
India/Pakistan
Border
Border Dispute Contested
2000 India
India/Pakistan
Border
Border Dispute Contested
2013-
2015
Syria Syria
Syrian Civil
War
Accepted
Note: * indicates uncertainty of location.

219

Table 2. Average chemical weapons use across cases
Number
of
Cases
Alleged
State Use
of
Chemical
Weapons
Alleged
Number of
Times
Chemical
Weapons
Used
Against the
State by a
Second
Party
Alleged
State Use of
Chemical
Weapons
Within the
State by the
State
State Use of
Chemical
Weapons
Outside the
State
Against a
Second
Party
Average of
those
states
accused of
chemical
weapons
use
43 1.8 0.4 0.5 1.4
Note: These averages are of states that were accused of using chemical weapons
at least once. States possessing chemical weapons, but never accused of using
them, were not included in these calculations

220

Table 3. Statistical testing for use of chemical weapons by and against the state
Variable Alleged State
Use of
Chemical
Weapons
Alleged State Use
of Chemical
Weapons Within
the State
Alleged State Use
of Chemical
Weapons Outside
the State
Constant 1.05
(0.23)
0.22
(0.12)
0.83
(0.21)
Chemical
Weapons Use
Against the
State
0.03
(0.28)
0.14
(0.15)
-0.11
(0.26)

Adj. R Squared -0.02 -0.00 -0.02
N 45 45 45
*p<0.05; **p<0.01; ***p<0.001
Note: Entries in this table are unstandardized regression coefficients with standard
errors in parentheses

221

Table 4. Military vulnerability set membership scores
Condition
Name
Hypothesis Membership Scale Coding
Instructions
Regime
Survival (T)
A perceived
threat to regime
survival is
sufficient for the
state to use
chemical
weapons to
neutralize the
threat.
1.0- A severe threat to
regime survival is
perceived to exist
A Major Response
to Crisis Trigger
(MAJRES) ranking
of 8 or 9
0.8- A major threat to
regime survival is
perceived to exist
A MAJRES
ranking of 6 or 7
0.6- A moderate threat
to regime survival is
perceived to exist
A MAJRES
ranking of 4 or 5
0.4- A minor threat to
regime survival is
perceived to exist
A MAJRES
ranking or 2 or 3
0.2- An extremely
minor threat to regime
survival is perceived to
exist
A MAJRES
ranking of 1
0.0- State does not
perceive a threat to the
regime
State has no entry
in the ICB database
Power
Asymmetry
(A)
A large power
asymmetry
during a conflict
is sufficient for a
state to decide to
use chemical
weapons in order
to correct that
imbalance.
1.0- State’s most
powerful opponent is a
nuclear power and the
state is not
State’s opponent is
believed to possess
nuclear weapons
and the state does
not
0.8- State’s most
powerful opponent is
not a nuclear power,
but has an
overwhelming military
superiority; the state is
not a nuclear power
The ratio of the
state’s Composite
Index of National
Capability (CINC)
ranking to that of
its closest rival is
roughly 1:2 or
smaller
0.6- State’s most
powerful opponent is
not a nuclear power,
but is clearly militarily
The ratio of the
state’s CINC to
that of its closest
rival is between
1:1.7 and 1:1.9

222

superior; the state is
not a nuclear power
0.4- State’s most
powerful opponent is
not a nuclear power,
but is slightly more
powerful; the state is
not a nuclear power
The ratio of the
state’s CINC to
that of its closest
rival is between
1:1.4 and 1:1.6
0.2- The state is
slightly less powerful
than its most powerful
opponent; state is a
nuclear power and its
rival is as well
The ratio of the
state’s CINC to
that of its closest
rival is between
1:1.1 and 1:1.3
0.0- The state and its
most powerful
opponent are roughly
even in power or the
state is the more
powerful of the two;
state is a nuclear power
and its rival is or is not;
state not in an armed
conflict
The ratio of the
state’s CINC to
that of its closest
rival is roughly 1:1
or larger; state is a
nuclear power and
its rival state is not;
state has no entry
in the COW
database
Stalemate
(S)
The existence of
a stalemate
during an armed
conflict is
sufficient to for a
state to use
chemical
weapons.
1.0- State is in an
armed conflict and has
had periods where
forward progress has
been stalled for several
months to years
State is in an armed
conflict, and
forward progress in
winning the
conflict has been
stalled for several
months to years
0.8- State is in an
armed conflict, and
forward progress in
winning stalls for up to
a month at a time
State is in an armed
conflict, and
forward progress in
winning the
conflict stalls for
up to a month at a
time
0.6- State is in an
armed conflict, and
forward progress in
State is in an armed
conflict, and
forward progress in

223

winning stalls for up to
two weeks at a time
winning the
conflict stalls for
up to two weeks at
a time
0.4- State is in an
armed conflict, and
while the state is
mostly winning, there
are short periods of
stalling
State is in an armed
conflict, and
forward progress in
winning the
conflict is good
overall, but there
are occasional
short periods of
stalling
0.2- State is in an
armed conflict in
which it has made
constant progress
towards winning
State is in an armed
conflict, and there
is constant forward
progress in
winning the
conflict from
beginning to
conclusion
0.0- State is not in an
armed conflict
State has no entry
in the COW war
databases
Note: COW stands for the Correlates of War (Sarkees and Wayman 2010). The
CINC ranking is taken from the National Material Capabilities Database (Singer,
Bremer, and Stuckey 1972, 19-48). The MAJRES ranking is taken from the
International Crisis Behavior Database (Brecher and Wilkenfeld 2007). The
response to the perceived threat is assumed to be proportional to the perceived
degree of threat.
Regime Survival Note: The highest MAJRES score for a country within a given
date range was recorded as the overall score for the country year. The average of
the scores was not used, as the state would be the most likely to use chemical
weapons under the conditions of highest threat to the state. Reporting the
maximum MAJRES score preserves the level of threat to the state.
Power Asymmetry Note: The CINC score for each state in a conflict was
calculated by averaging the CINC scores across the dates of the conflict. If there
is no CINC score in the conflict period, the CINC value immediately preceding
the conflict dates was used. When more than one conflict is present in any

224

stateyear, the lowest ratio between the state and its most powerful opponent was
recorded. This value was chosen to preserve the conflict information in which the
state is most likely to use their chemical weapons.
If there are only conflicts in a given stateyear that involve non-state actors or
states that do not have a CINC recorded, the CINC ratio was recorded in the
following manner. If the state is in a conflict with a non-state actor or a state with
missing CINC data and eventually wins against that actor, reaches a compromise,
or is locked in a stalemate, the conflict will be coded as [0.0] according to the
assumption that opponent was equal to or less powerful than the state. If the state
eventually loses to the opponent, the conflict will receive a score of [0.4] if it is
recorded that the state fought them for less than 5 years and a score of [0.2] if it is
recorded that the state fought them for longer than 5 years.
Stalemate Note: For each stateyear, a conflict will be scored according to the
forward momentum of that state. If there is not enough detailed information on
the progress of the conflict in the literature and the conflict lasts over one year,
that conflict will receive a score of [1.0]. If a war ends in a stalemate or
compromise, that conflict will be scored as [1.0]. In this case, the two sides were
too evenly matched to declare a victor, a similar situation to a stall in the progress
of a war. Additionally, if it is recorded that the state eventually lost the war or was
in retreat, they will also receive a score of [1.0]. If the state is involved in either a
civil war or a war with a non-state actor for longer than one month, the conflict
will be coded as [0.8]. If the state then loses that conflict, the score changes to
[1.0].

225

Table 5. Political vulnerability set membership scores
Condition
Name
Hypothesis Membership Scale Coding
Instructions
Credible
Commitment
(C)
The lack of a
credible
conventional or
nuclear
security
commitment
from a more
powerful state
is sufficient for
a state to use
chemical
weapons in a
conflict.
1.0- State lacks an
alliance with any
country; state lacks a
nuclear program
No Entry in the
COW Formal
Alliance
Database; state
lacks a nuclear
program
0.8- State has a neutrality
agreement with states not
in the conflict; state has
no nuclear program
Entry in COW
Formal Alliance
Dataset, Class 2
type, neutrality;
state lacks a
nuclear program
Alternate: Entry
in COW Formal
Alliance Dataset,
Class 2, 3 type,
neutrality, entente
0.6- State has an entente
understanding with states
not in the conflict; state
has no nuclear program
Entry in COW
Formal Alliance
Dataset, Class 3
entente; state
lacks a nuclear
program
Alternate: Entry
in COW Formal
Alliance Dataset,
Nonaggression
0.4- State has a non-
aggression treaty with
states not in the conflict;
state has a small nuclear
program
Entry in COW
Formal Alliance
Dataset, non-
aggression; state
has a small
nuclear program
Alternate: Entry
in COW Formal
Alliance Dataset,
Class 1, defense,
minor power

226

0.2- State has a defense
pact in place with a
major power; state is a
near nuclear power
Entry in COW
Formal Alliance
Dataset, Class 1,
defense; state is a
near nuclear
power
Alternate: Entry
in COW Formal
Alliance Dataset,
Class 1, defense,
major power
0.0- State has a nuclear
defense pact with a
stronger state; state is a
nuclear power or a
regional hegemon; state
is not in an armed
conflict
Presence of a
nuclear defense
pact; state is a
nuclear power
and/or regional
hegemon; no
entry in COW
database
Ostracization
(O)
Ostracization
or international
disregard is
sufficient for a
state to use
chemical
weapons in a
conflict.
1.0- State is ostracized
by the international
community, has little
trade or diplomatic
contact
State has more
than one major
sanction from the
UN, state has
more than one
major sanction
signed by at least
two major powers
and average GDP
is ranked lower
than “lower
middle income”
by the World
bank for a given
period
0.8- State is at the fringes
of the international
economy and is
generally considered
problematic or engaged
in bad behavior
State has one
major sanction
from the UN,
state has one
major sanction
signed by at least
two major powers

227

and average GDP
is ranked lower
than “lower
middle income”
by the World
bank for a given
period
0.6- State is on the
fringes of the
international economic
system and is generally
considered insolvent or
unimportant
Average GDP is
ranked lower than
“low income” by
the World bank
for a given period;
no known
oil/natural gas
reserves
0.4- State is a small
player in the
international system and
is generally considered
unimportant
Average GDP is
ranked lower than
“low income” by
the World bank
for a given period;
there are some
known gas
reserves
0.2- State is a moderate
player in the
international system and
is generally considered a
contributing/productive
member of the system
Average GDP is
ranked between
“lower middle
income” and
“upper middle
income” by the
World Bank for a
given period;
there are
significant known
gas reserves
0.0- State is a major
player in the
international system
State a member of
OPEC, permanent
member of the
UN Security
Council; average
GDP is ranked
“high income

228

OECD/non-
OECD” or above
for a given period
by the World
Bank
Democratic
Regime (D)
The state being
a democracy is
sufficient for it
to use chemical
weapons
1.0- State is considered
to be a consolidated
democracy
Average Polity IV
score of 8 to 10
0.8- State considered to
be a democracy
Average Polity IV
score of 4 to 7
0.6- State considered to
be a more democratic
anocracy
Average Polity IV
score of 1 to 3
0.4- State considered to
be a more authoritarian
anocracy
Average Polity IV
score of -2 to 0
0.2- State considered to
be an autocracy
Average Polity IV
score of -6 to -3
0.0- State is considered
to be a hereditary
monarchy
Average Polity IV
score of -10 to -7
Note: COW stands for the Correlates of War (Sarkees and Wayman 2010). GDP
rankings can be found in the World Development Indicators report (World Bank
2014).
Credible Commitment Note: For each stateyear there are typically multiple
alliances recorded in the COW Database. The score for any stateyear was taken
from the alliance rating that yielded the lowest possible score (i.e. neutrality
agreement vs defense pact from a major power  0.2). The stateyear was given a
score less than [1.0] if there was an alliance in any given year, with the exception
of alliances that were in place during WWI and ended in 1918.
Ostracization: The score for any given stateyear is awarded based on inclusion in
the state system. If a state is subject to more than one sanction by the UN that is
military or economic in nature (embargo, military intervention, asset freezes of
the state, etc.) or by more than one major economic sanction signed by at least
one major power, the state will receive a score of [1.0]. If the state is subject to
only on military or economic sanction by the UN or an economic sanction signed
by at least one major power, the state will receive a score of [0.8]. Reparations
required by major treaties qualify as one set of sanctions.

229

If the state does not have economic data recorded by the World Bank for a given
stateyear, the state will be scored in the following manner. If the state is a major
power, it will receive a score of [0.0]. If the state is considered a member of the 1
st

or 2
nd
world but is not a major power, the state will receive a score of [0.2]. If the
state is a 3
rd
world country and possesses oil reserves known at the time, it will
receive a score of [0.4]. If the state is a 3
rd
world country and is not known to
possess oil reserves, it will receive a score of [0.6]. If a state is a major power or a
permanent member of the UN Security Council, it will be awarded a score of [0.0]
exclusively. All other countries will receive the highest score possible.
Democratic Regime: The score for any given country year was calculated using a
weighted average from the Polity IV dataset. For countries that entered/exited the
dataset in a year ending not in a 0 or 5, the first/last entry was included in the
country/year bin in which it fell (i.e. 1963 in the country year country1960).
However, scores for that country year were only used from the years it was
included in the dataset (i.e. 1963 and 1964). For years that were coded as in
transition (i.e. -66, -77, -88), the democracy score was taken from the other years
in the date range. For periods of transition that fell in the middle of the date range
with scores both before and after, the average of the beginning and ending scores
was used in the weighted average, assuming a linear path between beginning and
end.

230

Table 6. Conflict location set membership scores
Condition
Name
Hypothesis Membership Scale Coding Instructions
Foreign
Territory
(E)
State
participation in a
conflict not in its
home territory is
sufficient to
allow one state
to use its
chemical
weapons.
1.0- State is
involved in an
armed conflict in
which all the
fighting is in
another state
Entry in COW Extra-
State war database; entry
in COW Inter-State
Database in which all the
fighting is in another
state
0.8- State is
involved in an
armed conflict in
which the fighting
is mostly
conducted in an
outside state
Entry in COW Inter-State
Database, region fought
in not same as state or
more than half the battles
fought outside the state
0.6- State is
involved in an
armed conflict in
which the fighting
is conducted half in
another state and
half within the
state’s prewar
boundaries
Entry in COW Inter-State
Database, region fought
in same as the state or
half the fighting is
conducted within the
state
0.4- State is
involved in an
armed conflict in
which a small
number of battles
are fought outside
the pre-war
boundaries of the
state
Entry in COW Intra-State
Database, entry as a
regional internal war;
entry in COW Inter-State
Database, only a small
number of battles are
fought outside the state
0.2- State is
involved in an
armed conflict in
which the
occasional battle is
fought outside the
Entry in COW Intra-State
Database, entry as a civil
or intercommunal war;
entry in COW Inter-State
Database, only a few
battles fought outside the
state

231

pre-war boundaries
of the state
0.0- State not in an
armed conflict
No entry in COW
database
Global
South (G)
State
involvement in a
conflict being
conducted in the
Global South is
sufficient for it
to use chemical
weapons
1.0- State involved
in an armed
conflict in a
country considered
to be in the Global
South or 3
rd
World
and is a relatively
unimportant state
State involved in an
armed conflict in a
country considered to be
in the Global South or 3
rd

World and has a World
Bank GDP ranking of
“low income” or below
for a given period
0.8- State involved
in an armed
conflict in a
country considered
to be in the Global
South or 3
rd
World
and is not a
member of OPEC
State involved in an
armed conflict in a
country considered to be
in the Global South or 3
rd

World, is not a member
of OPEC, and is ranked
by the World bank as
having a GDP of “lower
middle income” for a
given period
0.6- State involved
in an armed
conflict in a
country considered
to be in the Global
South or 3
rd
World
and is a member of
OPEC
State involved in an
armed conflict in a
country considered to be
in the Global South or 3
rd

World and is a member
of OPEC
0.4- State involved
in an armed
conflict in a non-
aligned country, a
non-permanent
member of the
UNSC, or
relatively
moderately
wealthy state
State involved in an
armed conflict in a non-
aligned country, a non-
permanent member of the
UNSC, or a state ranked
“upper middle income”
in GDP for a given
period by the World
Bank

232

0.2- State involved
in an armed
conflict in a
country considered
to be in the 1
st
or
2
nd
World, Global
North, a permanent
member of the
UNSC, or in a
wealthy state
State involved in an
armed conflict in a
country considered to be
in the 1
st
or 2
nd
World,
Global North, a
permanent member of the
UNSC, or in a country
ranked as “high income
OECD/non-OECD” or
above for a given period
by the World Bank
0.0- State not
involved in an
armed conflict
State has no entry in
COW database
Note: COW stands for the Correlates of War (Sarkees and Wayman 2010). GDP
rankings can be found in the World Development Indicators report (World Bank
2014).
Foreign Territory Note: The score for where a conflict was fought was determined
by which COW database a state falls into in any given stateyear. If the state has
entries in more than one database in any given stateyear, the database that yields
the highest overall score will be the one recorded. This preserves the higher
theoretical likelihood that a state will use their chemical weapons. If the fighting
is along the border between the state and an adjoining state, the fighting is
considered half in the adjoining state.
Global South Note: The score for Global South was calculated from the location
of the conflict selected in the Foreign Territory condition to maintain consistency.
If there are multiple conflicts in that condition that score the same, the one that
yields the highest overall score for the Global South condition is recorded. A state
is considered First World if it is a signatory of the North Atlantic Treaty
Organization (NATO) or is in the US sphere of influence after WWII, Second
World if it is a signatory of the Warsaw Pact or is in the Soviet sphere of
influence after WWII, and Third World if it is neither. A non-aligned state is one
that is a member of the Non-Aligned Movement (NAM).
If the fighting is carried out at least half in the other state, the opponent state is the
one that is assessed for the Global South condition. If the fighting is carried out
less than half in the other state, the state is the one assessed for the Global south
condition.

233

The Global North is considered to be North America, Western Europe, and the
developed parts of East Asia. The Global South is considered to be Africa, Latin
America, the developing countries in Asia, and the Middle East.
For the pre-WWI and interwar period, economic data from the Maddison Project
will be used. If a state’s GDP is on par or greater than the average of the United
States and Western Europe, the state will be coded as [0.2]. The rest of the scale is
pegged to the Western Europe and US average. If the average ratio of state’s
economy to that of Western Europe and the US during the conflict is between
0.99-0.80 it will receive a code of [0.4], between 0.79-0.60 a score of [0.6], and
between 0.59-0.40 a score of [0.8]. A score of [1.0] is given to those countries
whose GDP ratio to Western Europe and the US is less than or equal 0.39. If a
GDP is unavailable for a given time period, the GDP nearest the missing value
time wise will be substituted.
If there is no GDP data available and the state a conflict is fought in is a colony or
part of a mandate, the conflict will receive a score of [1.0]. If the fighting is in a
state that would later be considered a Third World country, but is not a colony or
mandate, it will receive a score of [0.8].
If the state does not have an entry in the World Bank Database, the Maddison
Project data and scoring system will be substituted. For the special case of WWII,
all fighting in Europe is coded as in a major power. Fighting between the US, its
allies in the Pacific, and the Japanese is also coded as in a major power. All
fighting between the Chinese and the Japanese is coded as if the fighting were in
China.

234

Table 7. Outcome condition set membership scores
Condition
Name
Hypothesis Membership Scale Coding Instructions
Chemical
Weapons
Use (W)
A state is
believed to
have used
chemical
weapons
1.0- State is
believed, beyond a
doubt, to have used
chemical weapons
State admits that it used
chemical weapons; There
is laboratory proof that
chemical weapons had
been used where it was
claimed to have been used;
International medical aid
organizations with
multiple reports verify that
symptoms of victims
clinically are indicative of
chemical weapons
exposure
0.8- There are
accusations of
chemical weapons
use; it is generally
accepted that
chemical weapons
have been used
International medical aid
organizations with
multiple reports verify that
symptoms of victims
clinically are indicative of
chemical weapons
exposure; UN or League
of Nations investigations
conclude that chemical
weapons have been used,
even if they do not state by
who; Consensus in the
literature that chemical
weapons had been used;
Intelligence assessments
claim that chemical
weapons have been used
(but do not give away their
sources and methods)
0.6- There are
accusations of
chemical weapons
use; international
attempts to verify
There are accusations of
chemical weapons use;
international attempts to
verify use that judge it
was more likely than not;
some consensus in the

235

use that judge it was
more likely than not
literature that the events in
fact occurred, but there is
some doubt; intelligence
agencies believe the
events to have occurred,
but no solid proof is
offered; there is reporting
in the media that the
events have occurred
involving a number of
survivors
0.4- There are
accusations of
chemical weapons
use; accusations
lead to some
international
attempts to verify
use; no consensus in
the literature that the
events in fact
occurred
There are accusations of
chemical weapons use;
accusations lead to some
international attempts to
verify use; no consensus
in the literature that the
events in fact occurred;
some intelligence agencies
believe the events to have
occurred, but others
dispute the claim; some
reporting in the
international media which
quickly goes away
0.2- There are
accusations of
chemical weapons
use; there is little to
no consensus that
the weapons were
used; former
accusations that
were later
scientifically proven
to not have occurred
There are accusations of
chemical weapons use;
accusations lead to little to
no action by international
and state officials
(typically only small
number of newspaper
reporting); former
accusations (even
substantiated by
intelligence agencies) that
were later scientifically
proven to not have
occurred

236

0.0- No accusations
of state use of
chemical weapons
No mention in the
literature that the state has
been accused of chemical
weapons usage
Non-lethal
first (R)
A state is
believed to
have used
non-lethal
chemical
weapons and
then
progresses to
lethal
weapons
1.0- State is
believed, beyond a
doubt, to have used
non-lethal and lethal
chemical weapons
State is believed, beyond a
doubt, to have used non-
lethal and lethal chemical
weapons; there is
laboratory proof that
chemical weapons have
been employed; multiple
histories have stated that
chemical weapons use did
occur; state acknowledges
that it has used chemical
weapons; “Smoking test
tube”
0.8- There are
accusations of non-
lethal and lethal
chemical weapons
use; it is generally
accepted that both
lethal and non-lethal
chemical weapons
have been used
There are accusations of
non-lethal and lethal
chemical weapons use; it
is generally accepted that
both lethal and non-lethal
chemical weapons have
been used; accusations
lead international attempts
to verify use and use is
verified; consensus in the
literature that the events in
fact occurred; intelligence
agencies believe the
events to have occurred;
substantial reporting in the
media of the events and of
survivors’ testimony
0.6- There are
accusations of lethal
and non-lethal
chemical weapons
use; it is generally
accepted that non-
lethal usage has
There are accusations of
lethal and non-lethal
chemical weapons use; it
is generally accepted that
non-lethal usage has
occurred but there is a
lesser consensus on the

237

occurred but there is
a lesser consensus
on the use of lethal
chemical weapons
use of lethal chemical
weapons; accusations lead
to little to no action by
international and state
officials (typically only
small number of
newspaper reporting)
0.4- There are
accusations of non-
lethal chemical
weapons use that are
generally accepted
by the international
community to have
occurred
There are accusations of
non-lethal chemical
weapons use that are
generally accepted by the
international community
to have occurred;
consensus in the literature
that the events in fact
occurred; intelligence
agencies believe the
events to have occurred;
substantial reporting in the
media of the events and of
survivors’ testimony
0.2- There are
accusations of non-
lethal chemical
weapons use; there
is a lesser consensus
on the use of non-
lethal chemical
weapons; former
accusations that
were later
scientifically proven
to not have occurred
There are accusations of
lethal and non-lethal
chemical weapons use;
accusations lead to little to
no action by international
and state officials
(typically only small
number of newspaper
reporting); former
accusations (even
substantiated by
intelligence agencies) that
were later scientifically
proven to not have
occurred
0.0- No accusations
of state use of
chemical weapons
No mention in the
literature that the state has
been accused of chemical
weapons usage

238

Chemical Weapons Use Note: The score for chemical weapons use is calibrated
according to the availability of evidence in the literature that chemical weapons
use occurred. The total number and specificity of reports will be considered, as
well as if the suspect agent is named. In lieu of a formal admission by a state of
chemical weapons use, a state will receive a score of [1.0] if there is archival
evidence unearthed in that state that documents the use of chemical weapons or
the members of that state are convicted in a tribunal of using chemical weapons.
Non-Lethal First Note: The score for non-lethal first is calibrated according to the
availability of evidence in the literature that a) chemical weapons use occurred,
and b) the state progressed from using non-lethal weapons to the lethal variety. In
the event that a state only uses lethal chemical weapons in a conflict, the state will
receive the following scores: [0.2] if there are accusations of lethal chemical
weapons use, but there is a lesser consensus in the literature, or former
accusations of lethal chemical weapons use that were later scientifically proven to
not have occurred; [0.4] if there are accusations of lethal chemical weapons use
that are generally accepted by the international community to have occurred. This
scoring preserves the information that the state is accused of using chemical
weapons, but is insufficient for supporting inclusion in the set of states that used
non-lethal and then lethal chemical weapons.

239

Table 8. Summary table of all fuzzy set scores for states that have been accused of using chemical weapons
State Year
Regime
Survival
Power
Asymmetry
Stalemate
Lack of
Credible
Commitment
Ostracization
Democratic
Regime
Foreign
Territory
Global South
Chemical
Weapons
Use
Non-lethal
First
D+g+e+s
a*c*o*D
T*A*S*D*E*G
Abbreviation T A S C O D E G W R
Ango1985 0.0 0.0 0.0 0.0 0.4 0.0 0.0 0.0 0.2 0.2 1.0 0.0 0.0
Chin1930 1.0 0.0 0.4 1.0 0.6 0.2 0.2 1.0 0.4 0.4 0.8 0.0 0.0
Chin1945 0.4 0.0 0.0 0.2 0.0 0.2 0.2 0.2 0.2 0.2 1.0 0.2 0.0
Chin1955 1.0 0.0 1.0 0.0 0.0 0.0 0.2 0.2 0.2 0.2 0.8 0.0 0.0
Cuba1955 0.0 0.0 0.0 0.0 0.4 0.0 0.0 0.0 0.2 0.2 1.0 0.0 0.0
Egyp1960 1.0 0.0 1.0 0.4 0.4 0.0 1.0 1.0 0.8 0.8 0.0 0.0 0.0
Egyp1965 1.0 0.0 1.0 0.4 0.4 0.0 1.0 1.0 0.8 0.8 0.0 0.0 0.0
Fran1920 0.8 0.0 0.8 0.0 0.0 1.0 1.0 1.0 0.8 0.4 1.0 1.0 0.0
Fran1925 0.0 0.0 0.2 0.0 0.0 1.0 1.0 1.0 0.6 0.4 1.0 1.0 0.0
Fran1945 1.0 0.2 1.0 0.0 0.0 1.0 1.0 1.0 0.2 0.2 1.0 0.8 0.2
Fran1955 1.0 0.4 0.8 0.0 0.0 1.0 1.0 1.0 0.2 0.2 1.0 0.6 0.4
Germ1940 1.0 0.8 0.8 0.0 0.0 0.0 0.6 0.2 1.0 0.4 0.8 0.0 0.0
Gree1945 0.0 0.0 0.0 0.0 0.2 0.8 0.0 0.0 1.0 0.4 1.0 0.8 0.0
Indi1995 1.0 0.0 0.8 0.0 0.4 1.0 0.6 1.0 0.2 0.2 1.0 0.6 0.0
Indi2000 0.8 0.0 0.8 0.0 0.4 1.0 0.2 0.8 0.2 0.2 1.0 0.6 0.0
Iran1980 1.0 0.0 1.0 1.0 0.0 0.2 0.2 0.6 0.6 0.4 0.8 0.0 0.0
Iran1985 1.0 0.0 1.0 1.0 0.0 0.2 0.2 0.6 0.6 0.4 0.8 0.0 0.0

240

Iraq1980 1.0 0.2 1.0 0.4 0.0 0.0 0.8 0.6 1.0 1.0 0.4 0.0 0.0
Iraq1985 1.0 0.0 1.0 0.4 0.0 0.0 0.8 0.6 1.0 1.0 0.4 0.0 0.0
Iraq1995 1.0 0.0 1.0 0.4 0.8 0.0 0.2 0.6 0.2 0.2 0.8 0.0 0.0
Isra1945 1.0 0.8 0.6 1.0 0.2 1.0 0.2 1.0 0.2 0.2 1.0 0.0 0.2
Isra1965 1.0 0.8 1.0 1.0 0.2 1.0 0.4 0.2 0.2 0.2 1.0 0.0 0.2
Ital1925 0.4 0.0 0.8 0.0 0.0 0.2 1.0 1.0 1.0 1.0 0.2 0.2 0.0
Ital1930 0.8 0.0 0.8 0.0 0.0 0.0 1.0 1.0 1.0 1.0 0.2 0.0 0.0
Ital1935 0.4 0.0 0.8 0.0 0.0 0.0 1.0 1.0 1.0 1.0 0.2 0.0 0.0
Japa1930 1.0 0.0 0.0 0.0 0.0 0.6 0.0 0.0 0.8 0.4 1.0 0.6 0.0
Japa1935 1.0 0.8 0.4 0.8 0.0 0.6 0.8 1.0 1.0 1.0 0.6 0.2 0.4
Japa1940 1.0 0.8 0.8 0.4 0.0 0.6 1.0 1.0 1.0 1.0 0.6 0.2 0.6
Japa1945 1.0 1.0 1.0 0.4 0.0 0.6 0.2 1.0 1.0 1.0 0.8 0.0 0.2
Liby1985 1.0 0.0 1.0 0.2 1.0 0.0 1.0 0.4 0.8 0.4 0.6 0.0 0.0
Port1965 0.0 0.0 0.0 0.0 0.6 0.0 0.0 0.0 0.8 0.4 1.0 0.0 0.0
Rhod1970 0.0 0.0 0.0 0.0 0.8 0.8 0.0 0.0 0.2 0.2 1.0 0.2 0.0
SAfr1975 1.0 0.0 0.8 0.0 0.6 0.8 1.0 0.4 0.8 0.4 0.8 0.4 0.0
SAfr1990 0.0 0.0 0.0 0.0 0.6 0.8 0.0 0.0 0.8 0.4 1.0 0.4 0.0
Serb1990 1.0 0.0 1.0 1.0 0.4 0.2 1.0 0.2 0.4 0.6 0.8 0.0 0.0
Serb1995 1.0 0.0 0.0 0.0 0.4 0.0 0.0 0.0 0.4 0.6 1.0 0.0 0.0
Spai1920 0.0 0.0 1.0 1.0 0.2 0.6 1.0 1.0 0.8 0.4 0.6 0.0 0.0
Spai1925 0.0 0.0 0.2 0.8 0.2 0.2 1.0 1.0 0.8 0.4 0.8 0.2 0.0
Spai1935 1.0 0.8 1.0 0.8 0.2 0.8 0.2 1.0 0.4 0.4 0.8 0.2 0.2
Spai1955 1.0 0.0 0.0 0.0 0.2 0.0 0.0 0.0 0.2 0.2 1.0 0.0 0.0
Suda1995 0.0 0.0 0.0 0.0 0.4 0.0 0.0 0.0 0.2 0.2 1.0 0.0 0.0
Syri2010 1.0 0.0 1.0 0.4 0.4 0.0 0.2 1.0 1.0 0.4 0.8 0.0 0.0

241

UKin1918 1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.4 1.0 1.0 0.0
UKin1920 0.8 0.0 0.8 0.0 0.0 1.0 1.0 1.0 0.2 0.2 1.0 1.0 0.0
UKin1935 0.8 0.6 0.2 0.0 0.0 1.0 0.6 0.2 0.2 0.2 1.0 0.4 0.2
USAm1950 1.0 0.0 1.0 0.0 0.0 1.0 1.0 0.2 0.2 0.2 1.0 1.0 0.0
USAm1960 1.0 0.0 1.0 0.0 0.0 1.0 1.0 0.2 1.0 0.4 1.0 1.0 0.0
USAm1965 1.0 0.0 0.8 0.0 0.0 1.0 1.0 0.2 1.0 0.4 1.0 1.0 0.0
USAm1970 1.0 0.0 1.0 0.0 0.0 1.0 1.0 0.2 1.0 0.4 1.0 1.0 0.0
USAm1975 1.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 1.0 0.4 1.0 1.0 0.0
USSR1918 1.0 0.8 1.0 1.0 0.0 0.4 0.2 1.0 0.2 0.2 0.8 0.0 0.2
USSR1920 1.0 0.8 1.0 0.0 0.0 0.0 0.2 1.0 0.8 0.4 0.8 0.0 0.0
USSR1975 1.0 0.0 0.8 0.0 0.0 0.0 0.2 0.4 0.2 0.2 0.8 0.0 0.0
Viet1975 1.0 0.0 0.8 0.8 0.2 0.0 1.0 0.2 0.2 0.2 0.8 0.0 0.0
Yugo1990 1.0 0.0 1.0 1.0 0.4 0.2 1.0 0.4 0.2 0.2 0.6 0.0 0.0
Note: The total number of cases included in the dataset was 270. This table only includes those cases in which states
were accused of using chemical weapons. Fuzzy set membership was scored between completely not belonging to a set
[0.0] and completely belonging to a set [1.0]. The range of scores between the minimum and maximum correlate to the
case’s degree of inclusion in the given set, and more specific meanings can be found in Tables 4-7. The score [0.5]
represents the point of maximum ambiguity (equally in and out of a set) and was therefore excluded from use. All cases
scored above [0.5] are to differing degrees considered belonging to the set, and all scores lower than [0.5] are to
differing degrees considered out of the set. Upper case letters denote the presence of a condition; lower case letters
denote the absence of a condition. The absence of all conditions were calculated with the given formula: a=1-A.

242

Table 9. Analysis of necessary conditions for
the outcome ‘chemical weapons use’
Condition Tested Consistency Coverage
T 0.784 0.194
A 0.185 0.185
S 0.698 0.216
C 0.284 0.282
O 0.253 0.131
D 0.494 0.120
E 0.654 0.200
G 0.636 0.229
Note: The minimum consistency required
for necessity is 0.90.

243

Table 10. Analysis of necessary conditions for
the outcome ‘non-lethal weapons first’
Conditions Tested Consistency Coverage
T 0.807 0.147
A 0.218 0.160
S 0.748 0.170
C 0.345 0.252
O 0.303 0.115
D 0.454 0.081
E 0.723 0.162
G 0.748 0.198
Note: The minimum consistency required
for necessity is 0.90.

244

Table 11. Analysis of sufficient conditions for the outcome ‘chemical
weapons use’
Solution T*a*S*C*G + T*S*C*d*G → W
Consistency: 0.592 0.517
Raw Coverage: 0.180 0.191
Unique Coverage: 0.012 0.025
Solution Consistency: 0.500 Solution Coverage: 0.204
Note: The consistency threshold (yconsist) has been set at 0.70. The next
highest consistency score is 0.73.

245

Table 12. Analysis of sufficiency conditions for the outcome ‘non-lethal
weapons first’
Solution T*a*S*C*G + T*S*C*d*G → R
Consistency: 0.551 0.483
Raw Coverage: 0.227 0.244
Unique Coverage:
Solution Consistency: 0.470 Solution Coverage: 0.261
Note: The consistency threshold (yconsist) has been set at 0.63. The next
highest consistency score is 0.64.

246

Table 13. Analysis of necessary conditions for
the outcome ‘use of chemical weapons” in the
1918-1945 Dataset
Conditions Tested Consistency Coverage
T 0.743 0.306
A 0.357 0.368
S 0.686 0.353
C 0.300 0.350
O 0.071 0.116
D 0.500 0.205
E 0.700 0.340
G 0.814 0.373
T+A+G 1.000 0.290
Note: The minimum consistency required
for necessity is 0.90.

247

Table 14. Analysis of necessary conditions for
the outcome ‘non-lethal first” in the 1918-
1945 Dataset
Conditions Tested Consistency Coverage
T 0.755 0.235
A 0.396 0.309
S 0.755 0.294
C 0.321 0.283
O 0.094 0.116
D 0.509 0.158
E 0.774 0.285
G 0.906 0.314
Note: The minimum consistency required
for necessity is 0.90.

248

Table 15. Analysis of sufficient conditions for the outcome ‘chemical weapons
use’ in the 1918-1945 dataset
Solution T*A*S*c
*d*G
+ T*A*C*D*
E*G
+ T*A*S*c*E
*G
+ T*A*S*D*
E*G
→
W
Consistenc
y:
0.786

0.800

0.643

0.692
Raw
Coverage:
0.157

0.114

0.129

0.129
Unique
Coverage:
0.057

0.014

-0.000

-0.000
Solution
Consistenc
y:
0.727
Solution
Coverage: 0.229

Note: The consistency threshold (yconsist) has been set at 0.70.

249

Table 16. Analysis of sufficient conditions for the outcome ‘non-lethal
weapons first’ in the 1918-1945 dataset
Solution T*A*C*D*E*G + T*A*S*D*E*G → R
Consistency: 0.800 0.692
Raw Coverage: 0.151 0.170
Unique Coverage: 0.019 0.038
Solution Consistency: 0.714 Solution Coverage: 0.189

250

Table 17. Analysis of necessary conditions for
the outcome ‘use of chemical weapons” in the
1945-2014 Dataset
Conditions Tested Consistency Coverage
T 0.815 0.157
A 0.054 0.056
S 0.707 0.171
C 0.272 0.243
O 0.391 0.133
D 0.489 0.091
E 0.620 0.149
G 0.500 0.155
TA 0.815 0.147
Note: The minimum consistency required
for necessity is 0.90.

251

Table 18. Analysis of necessary conditions for
the outcome ‘non-lethal weapons first” in the
1945-2014 Dataset
Conditions Tested Consistency Coverage
T 0.848 0.117
A 0.076 0.056
S 0.742 0.129
C 0.364 0.233
O 0.470 0.115
D 0.409 0.055
E 0.682 0.117
G 0.621 0.139
Note: The minimum consistency required
for necessity is 0.90.

252

Table 19. Analysis of sufficient conditions for the outcome ‘chemical
weapons use’ in the 1945-2014 dataset
Solution T*a*S*C*o*d*e*G → W
Consistency: 0.786
Raw Coverage: 0.120
Unique Coverage: 0.120
Solution Consistency: 0.786 Solution Coverage: 0.120
Note: The consistency threshold (yconsist) has been set at 0.70. The next
highest consistency score is 0.786. A minimum of one case per truth table
row was required

253

Table 20. Analysis of sufficient conditions for the outcome ‘non-lethal
weapons first’ in the 1945-2014 dataset
Solution T*a*S*C*o*d*e*G → R
Consistency: 0.643
Raw Coverage: 0.136
Unique Coverage: 0.136
Solution Consistency: 0.643 Solution Coverage: 0.136
Note: The consistency threshold (yconsist) has been set at 0.60. The next
highest consistency score is 0.643. A minimum of one case per truth table
row was required

254

Table 21. Analysis of necessary conditions for
the outcome ‘non-use of chemical weapons
Conditions Tested Consistency Coverage
t 0.557 0.950
a 0.889 0.889
s 0.656 0.941
c 0.902 0.902
o 0.771 0.883
d 0.508 0.880
e 0.643 0.932
g 0.708 0.934
Note: The minimum consistency required
for necessity is 0.90.

255

Table 22. Analysis of necessary conditions for
the outcome ‘non-use of non-lethal weapons
first’
Conditions Tested Consistency Coverage
t 0.548 0.967
a 0.890 0.922
s 0.648 0.964
c 0.901 0.934
o 0.775 0.920
d 0.504 0.905
e 0.639 0.960
g 0.707 0.967
Note: The minimum consistency required
for necessity is 0.90.

256

Table 23. Analysis of sufficient conditions for the outcome ‘non-use of chemical weapons’
Solution t + O + d*g + s*c + D*e + c*D*G + S*D*G → w
Consistency: 0.950 0.952 0.956 0.946 0.929 0.907 0.910
Raw
Coverage:
0.557 0.251 0.416 0.639 0.265 0.188 0.170
Unique
Coverage:
0.054 0.016 0.034 0.034 0.015 0.010 0.000
Solution
Consistency:
0.924
Solution
Coverage:
0.933

Note: The consistency threshold (yconsist) has been set at 0.74. A minimum of two cases per truth table row was
required.

257

Table 24. Analysis of sufficient conditions for the outcome ‘non-use of non-
lethal weapons first’
Solution D + g + e + s →r
Consistency: 0.944 0.967 0.960 0.964
Raw Coverage: 0.510 0.707 0.639 0.648
Unique Coverage: 0.128 0.033 0.029 0.011
Solution Consistency: 0.945 Solution Coverage: 0.945
Note: The consistency threshold (yconsist) has been set at 0.70. The
minimum number of cases was set to 5.

258

Table 25. Analysis of necessary conditions for
the outcome ‘non-use of chemical weapons’ in
the 1918-1945 dataset
Conditions Tested Consistency Coverage
t 0.498 0.867
a 0.817 0.810
s 0.626 0.870
c 0.834 0.800
o 0.838 0.752
d 0.421 0.734
e 0.596 0.870
g 0.591 0.914
a+c 0.923 0.786
Note: The minimum consistency required
for necessity is 0.90.

259

Table 26. Analysis of necessary conditions for
the outcome ‘non-use of non-lethal weapons
first’ in the 1918-1945 dataset
Conditions Tested Consistency Coverage
t 0.484 0.904
a 0.813 0.865
s 0.619 0.923
c 0.829 0.853
o 0.849 0.817
d 0.429 0.806
e 0.591 0.925
g 0.583 0.967
a+c 0.929 0.848
Note: The minimum consistency required
for necessity is 0.90.

260

Table 27. Analysis of sufficient conditions for the outcome ‘non-use of chemical
weapons’ in the 1918-1945 dataset
Solution a*c*o*D + t*a*s*c*e*g + a*s*c*d*e*g →w
Consistency: 0.850 1.000 0.953
Raw Coverage: 0.434 0.260 0.174
Unique Coverage: 0.230 0.051 0.051
Solution
Consistency:
0.877
Solution
Coverage:
0.574

Note: The consistency threshold (yconsist) has been set at 0.70. The next highest
consistency score is 0.767. A minimum of two cases per truth table row was
required

261

Table 28. Analysis of sufficient conditions for the outcome ‘non-use of non-lethal weapons first’ in
the 1918-1945 dataset
Solution c*o*d*g + a*c*o*D + t*a*s*c*e*g + a*s*c*d*e*g →r
Consistency: 0.980 0.917 1.000 1.000
Raw Coverage: 0.194 0.437 0.242 0.171
Unique Coverage: 0.056 0.238 0.048 0.028
Solution Consistency: 0.935 Solution Coverage: 0.627
Note: The consistency threshold (yconsist) has been set at 0.70. The next highest consistency score
is 0.857. A minimum of two cases per truth table row was required

262

Table 29. Analysis of necessary conditions for
the outcome ‘non-use of chemical weapons’ in
the 1945-2014 dataset
Conditions Tested Consistency Coverage
t 0.575 0.970
a 0.910 0.908
s 0.667 0.959
c 0.918 0.928
o 0.753 0.927
d 0.526 0.914
e 0.656 0.947
g 0.736 0.938
Note: The minimum consistency required
for necessity is 0.90.

263

Table 30. Analysis of necessary conditions for
the outcome ‘non-use of non-lethal weapons
first' in the 1945-2014 dataset
Conditions Tested Consistency Coverage
t 0.567 0.982
a 0.913 0.936
s 0.659 0.974
c 0.919 0.955
o 0.755 0.955
d 0.521 0.929
e 0.653 0.968
g 0.738 0.966
Note: The minimum consistency required
for necessity is 0.90.

264

Table 31. Analysis of sufficient conditions for the outcome ‘non-use of chemical weapons’ in the
1945-2014 dataset
Solution a*c*o*G + a*s*c*o*e + a*c*o*d*e + t*a*s*c*e*g →w
Consistency: 0.923 0.972 0.974 0.959
Raw Coverage: 0.214 0.405 0.278 0.420
Unique Coverage: 0.187 0.044 0.13 0.121
Solution Consistency: 0.947 Solution Coverage: 0.737
Note: The consistency threshold (yconsist) has been set at 0.70. The next highest consistency score
is 0.720. A minimum of four cases per truth table row was required.

265

Table 32. Analysis of sufficient conditions for the outcome ‘non-use of non-lethal weapons first’
in the 1945-2014 dataset
Solution a*c*o*G + a*s*c*o*e + a*c*o*d*e + t*a*s*c*e*g →r
Consistency: 0.955 0.990 0.989 0.976
Raw Coverage: 0.216 0.401 0.275 0.416
Unique Coverage: 0.186 0.046 0.012 0.120
Solution Consistency: 0.967 Solution Coverage: 0.719
Note: The consistency threshold (yconsist) has been set at 0.70. The next highest consistency
score is 0.744. A minimum of four cases per truth table row was required.

266

Table 33. Analysis of attack ratios for the outcome ‘chemical weapons use’ for
1918-1945 dataset
T A S C O D E G Attack Ratio Cases
1 1 1 1 0 0 1 1 1 4.0 Japa1940
2 1 1 1 0 0 1 0 1 4.0 Japa1945
3 1 1 1 0 0 0 0 1 4.0 USSR1920
4 1 1 0 1 0 1 1 1 4.0 Japa1935
5 0 0 1 1 0 1 1 1 4.0 Spai1920
6 0 0 0 1 0 0 1 1 4.0 Spai1925
7 0 0 1 0 0 0 1 1 2.7 Ital1920, Ital1925, Ital1935
Note: The minimum number of cases for a given row was set to 1.

267

Table 34. Analysis of attack ratios for the outcome ‘chemical weapons use’
for 1945-2014 dataset
T A S C O D E G Attack Ratio Cases
1 1 0 1 1 0 0 0 1 27.0 Iran1980, Iran1985
2 1 0 1 0 1 0 1 0 27.0 Liby1985
3 1 0 1 0 0 0 0 1 27.0 Syri2010
4 1 0 1 0 0 0 1 1 21.6
Egyp1960, Egyp1965,
Iraq1980, Iraq1985, Syri1975
5 1 0 1 0 0 1 1 0 20.3
USAm1950, USAm1960,
USAm1965, USAm1970
6 1 0 1 0 1 1 1 0 13.5 SAfr1975, SAfr1985
Note: The minimum number of cases for a given row was set to 1.

268

Table 35. Analysis of attack ratios for the outcome ‘non-lethal chemical
weapons first’ for 1918-1945 dataset
T A S C O D E G Attack Ratio Cases
1 1 1 1 0 0 1 1 1 2.0 Japa1940
2 1 1 1 0 0 1 0 1 2.0 Japa1945
3 1 1 0 1 0 1 1 1 2.0 Japa1935
4 0 0 1 0 0 0 1 1 1.3 Ital1920, Ital1925, Ital1935
5 1 0 1 0 0 0 1 1 1.0 Ital1930, USSR1935
Note: The minimum number of cases for a given row was set to 1.

269

Table 36. Analysis of attack ratios for the outcome ‘non-lethal chemical
weapons first’ for 1945-2014 dataset
T A S C O D E G Attack
Ratio
Cases
1 1 0 1 0 0 0 1 1 2.4
Egyp1960, Egyp1965, Iraq1980,
Iraq1985, Syri1975
2 1 0 1 1 0 0 1 0 1.0 Serb1990, Viet1975, Yugo1990
3 1 0 0 0 0 0 0 0 0.2
Chin1980, Chin1995, Egyp1975,
Egyp1980, Egyp1985, NKor1965,
NKor1975, Serb1995, Spai1955,
Syri1985, USSR1960, USSR1965,
USSR1970
Note: The minimum number of cases for a given row was set to 1.

270

Table 37. Analysis of attack ratios for the outcome ‘non-use of chemical weapons’
for 1918-1945 dataset
T A S C O D E G Attack
Ratio
Cases
1 0 0 0 0 0 1 0 0 3.0
Germ1925, Germ1930, Japa1918,
Japa1920, SAfr1935, SAfr1940,
UKin1925, USAm1925,
USAm1930, USAm1935
2 1 0 0 0 1 0 0 0 3.0 Yugo1920, Yugo1925, Yugo1930
3 0 0 1 0 0 1 1 1 3.0 USAm1918, USAm1920
4 0 0 0 0 1 1 0 0 3.0 SAfr1935, SAfr1940
5 1 0 1 0 0 1 1 1 2.3
Fran1918, Fran1920, UKin1920,
UKin1930
6 1 0 0 0 0 1 0 0 1.5
Aust1930, Japa1925, Japa1930,
UKin1918
Note: The minimum number of cases for a given row was set to 2.

271

Table 38. Analysis of attack ratios for the outcome
‘non-use of chemical weapons’ for 1945-2014 dataset
T A S C O D E G Number
of Cases
Attack
Ratio
1 0 0 0 0 0 0 0 0 42 5.0
2 1 0 0 0 0 0 0 0 13 5.0
3 0 0 1 0 0 1 1 1 13 5.0
4 1 0 1 0 0 1 1 1 12 5.0
5 1 0 0 0 0 1 1 1 7 5.0
6 0 0 0 0 0 1 0 0 27 4.8
7 1 0 0 0 0 1 0 0 11 4.5
8 0 0 0 0 1 0 0 0 11 4.5
9 1 0 1 0 0 0 1 1 5 1.0
Note: The minimum number of cases for a given row
was set to 5.

272

Table 39. Analysis of attack ratios for the outcome ‘non-use of non-lethal
weapons first’ for 1918-1945 dataset
T A S C O D E G Attack Ratio Cases
1 0 0 0 0 0 1 0 0 3.0
Germ1925, Germ1930,
Japa1918, Japa1920,
UKin1925, USAm1925,
USAm1930, USAm1935
2 1 0 1 0 0 1 1 1 3.0
Fran1918, Fran1920,
UKin1920, UKin1930
3 1 0 0 0 0 1 0 0 3.0
Aust1930, Japa1925,
Japa1930, UKin1918
4 1 0 0 0 1 0 0 0 3.0
Yugo1920, Yugo1925,
Yugo1930
5 1 1 1 0 0 0 1 0 3.0 Germ1940, Ital1940
6 0 0 1 0 0 1 1 1 3.0 USAm1918, USAm1920
7 0 0 0 0 1 1 0 0 3.0 SAfr1935, SAfr1940
8 0 0 0 0 0 1 1 1 3.0 Fran1925, Fran1930
9 1 0 1 0 0 0 1 1 1.5 Ital1930, USSR1935
10 0 0 1 0 0 0 1 1 1.0 Ital1920, Ital1925, Ital1935
Note: The minimum number of cases for a given row was set to 2.

273

Table 40. Analysis of attack ratios for the outcome
‘non-use of non-lethal weapons first’ for 1945-2014
dataset
T A S C O D E G Number
of Cases
Attack
Ratio
1 0 0 0 0 0 0 0 0 42 5.0
2 0 0 0 0 0 1 0 0 27 5.0
3 0 0 1 0 0 1 1 1 13 5.0
4 1 0 1 0 0 1 1 1 12 5.0
5 1 0 0 0 0 1 0 0 11 5.0
6 0 0 0 0 1 0 0 0 11 5.0
7 1 0 0 0 0 1 1 1 7 5.0
8 1 0 0 0 0 0 0 0 13 4.6
9 1 0 1 0 0 0 1 1 5 1.0
Note: The minimum number of cases for a given row
was set to 5.

274

Table 41. Analysis of the predictive power of supersets in the 1918-1945 and
1945-2014 datasets
1918-1945
Use of
Chemical
Weapons
1918-1945
Non-use of
Chemical
Weapons
1945-2014
Non-use of
Chemical
Weapons
1945-2014
Non-use of
Non-Lethal
Weapons First
Included in Sufficient Solution
TAS X
COD
EG X X X
tas X X X
cod X X X
eg X X X
T X
A X
S X
C X
O
D X X
E X
G X X X
t X X X
a X X X
s X X X
c X X X X
o X X X
d X X X X
e X X X
g X X X
Note: For all analyses, the intermediate solution was used. A minimum of 1 case
and 0.70 sufficiency was required per truth table row. X indicates that the given
condition was included in the solution. Capital letters indicate presence of a
condition and lowercase letters indicate absence.

275

Table 42. Summary of fs/QCA analysis for the use and nonuse of chemical weapons and their escalation from 1918 to 2014
Outcome Necessary
Condition(s)
(Consistency/
Coverage)
Sufficient Conditions
(Consistency/Coverage)
Overall
Consistency/
Coverage
Driving Cases/
Conditions
Superset
Chemical
Weapons
Use
-
T*a*S*C*G + T*S*C*d*G
(0.592/0.180) (0.517/0.191)
0.500/0.204
Non-
Lethal
Weapons
First
-
T*a*S*C*G + T*S*C*d*G
(0.551/0.227) (0.483/0.244)
0.470/0.261
Chemical
Weapons
Use
1918-1945
Dataset
T+A+G
(1.00/0.290)
T*A*S*c*d*G+T*A*C*D*E*G+T*A*S*c*E*G+
T*A*S*D*E*G
(0.786/0.157) (0.800/0.114) (0.643/0.129) (0.692/0.129)
0.727/0.229
Japan (1935-1945),
USSR (1920-1924),
Spain (1920-1929)
+/-
Non-
Lethal
Weapons
First
1918-1945
Dataset
G
(0.906/0.314)
T*A*C*D*E*G + T*A*S*D*E*G
(0.800/0.151) (0.692/0.170)
0.714/0.189 Japan (1935-1945)
Chemical
Weapons
Use
1945-2014
Dataset
-
T*a*S*C*o*d*e*G
(0.786/0.120)
0.786/0.120
Iran (1980-1989),
Libya (1985), Syria
(2010-2014)

Non-
Lethal
Weapons
First
-
T*a*S*C*o*d*e*G
(0.643/0.136)
0.643/0.136
Egypt (1960-1969),
Iraq (1980-1989),
Syria (1975-1979)

276

1945-2014
Dataset
Non-use of
Chemical
Weapons
c
(0.902/0.902)
t + O + d*g + s*c + D*e + c*D*G + S*D*G
(0.950/0.557)(0.952/0.251)(0.956/0.416)(0.946/0.639)
(0.929/0.265)(0.907/0.188)(0.910/0.170)
0.924/0.933
Non-use of
Non-
Lethal
Weapons
First
c
(0.901/0.934)
D + g + e + s
(0.944/0.510) (0.967/0.707) (0.960/0.639) (9.64/0.648)
0.945/0.945
Non-use of
Chemical
Weapons
1918-1945
Dataset
a+c
(0.923/0.786)
a*c*o*D + t*a*s*c*e*g + a*s*c*d*e*g
(0.850/0.434) (1.000/0.260) (0.953/0.174)
0.877/0.574
Germany (1925-
1934), Japan (1918-
1924), South Africa
(1935-1944), United
Kingdom (1918-
1929), United States
(1925-1939),
Yugoslavia (1920-
1934)
+/-
Non-use of
Non-
Lethal
Weapons
First
1918-1945
Dataset
a+c
(0.929/0.848)
c*o*d*g + a*c*o*D + t*a*s*c*e*g + a*s*c*d*e*g
(0.980/0.194) (0.917/0.437) (1.000/0.242) (1.000/0.171)
0.935/0.627
Germany (1925-
1934, 1940-1944),
Japan (1918-1934),
United Kingdom
(1918-1934), United
States (1918-1939),
France (1918-1934),
Australia (1930-
1934), Yugoslavia
(1920-1934), Italy
(1940-1944), South
Africa (1935-1944)

Non-use of
Chemical
Weapons
a, c
(0.910/0.908),
(0.918/0.928)
a*c*o*G + a*s*c*o*e + a*c*o*d*e + t*a*s*c*e*g
(0.923/0.214) (0.972/0.405) (0.974/0.278) (0.959/0.420)
0.947/0.737
t*a*s*c*o*d*e*g,
T*a*s*c*o*d*e*g,
t*a*S*c*o*D*E*G,
+

277

1945-2014
Dataset
T*a*S*c*o*D*E*G,
T*a*s*c*o*D*E*G
Non-use of
Non-
Lethal
Weapons
First
1945-2014
Dataset
a, c
(0.913/0.936),
(0.919/0.955)
a*c*o*G + a*s*c*o*e + a*c*o*d*e + t*a*s*c*e*g
(0.955/0.216) (0.990/0.401) (0.989/0.275) (0.976/0.416)
0.967/0.719
t*a*s*c*o*d*e*g,
t*a*s*c*o*D*e*g,
t*a*S*c*o*D*E*G,
T*a*S*c*o*D*E*G,
T*a*s*c*o*D*e*g,
t*a*s*c*O*d*e*g,
T*a*s*c*o*D*E*G,
+
Note: Coverage scores reported here are the raw coverage scores reported in corresponding tables. – indicates that no condition was identified.
+ indicates that a condition was identified

Abstract (if available)

Abstract The use of chemical weapons in Syria from 2013-2015 raises grave questions about the ability of the international system to forestall the use of these weapons. Despite international pressure and approbation, chemical weapons were used against civilian populations in Syria in flagrant violation of international norms and treaty commitments. Using fuzzy set Qualitative Comparative Analysis (fs/QCA) to systematically analyze cases, this dissertation focuses on identifying the conditions under which state possessors of chemical weapons use and abstain from using their arsenals, both in domestic and international disputes. Conditions drawn from the nuclear proliferation literature are assessed alongside political and geographic conditions for joint sufficiency and necessity. fs/QCA analysis reveals that these variables cannot adequately explain why states choose to use their chemical weapons, necessitating future research. The analysis also uncovered that both possessing security commitments from other states and being more powerful than an opponent are necessary conditions for the non-use of chemical weapons. In addition to the fs/QCA analysis, the construction of artificial supersets was evaluated for efficacy in reducing the limited diversity problem inherent in fs/QCA analysis. This method passed a proof of concept test, and could prove useful for pilot studies and research in which explanatory conditions are not thoroughly documented.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Intelligent knowledge acquisition systems: from descriptive to predictive models

PDF

Lesser evils? WMD pursuit beyond nuclear weapons

PDF

Pathways to nuclear weapon reversal: exploring mechanisms and understanding non-proliferation policy

PDF

Thawing rivalries and fading friendships: a multi-method approach to rapprochement and alienation

PDF

The paradox of personalism: internal political constraints on nuclear posture in authoritarian regimes

PDF

Legitimizing self-determination: advancing the sovereignty of separatist movements

PDF

A two-level analysis of foreign policy decision making: an empirical investigation of the case of China-Taiwan

PDF

Code 'war' theorizing: information and communication technology's impact on international relations theorizing, negotiation, and cyber relations

PDF

Are you as old as you feel? A quantitative and qualitative analysis of leader age and foreign policy decision-making

PDF

Development of metabolic chemical reporters for the investigation of protein glycosylation

PDF

Things fall apart: the unraveling of international institutions through withdrawal

PDF

Changing political attitudes and behavior in a diverse America: incorporating individual and contextual determinants

PDF

Hard feelings: friendship, betrayal, and outrage in international relations

PDF

Supply and demand of economic hierarchy: the Northeast Asia case

PDF

Developing countries in the digital era: state and business interactions for industrial upgrading

PDF

Concurrent monitoring and diagnosis of process and quality faults with canonical correlation analysis

PDF

The allocation and accessibility of aid: the role of recipient government agency in dictating donor decisions

PDF

Creating destruction: the political economy of zombie firms

PDF

Expressing values and group identity through behavior and language

PDF

Goldilocks’ signal for security cooperation in East Asia: China’s rise, hedging, and joint military exercises

Asset Metadata

Creator MacNeal, Kymberly G. (author)

Core Title When bullets are not enough: determinants of chemical weapons usage and abstention

Contributor Electronically uploaded by the author (provenance)

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science and International Relations

Publication Date 09/21/2016

Defense Date 05/06/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag chemical weapons,fuzzy set qualitative comparative analysis,international norm,OAI-PMH Harvest,taboo

Format application/pdf (imt)

Language English

Advisor James, Patrick (committee chair), English, Robert (committee member), Figueiredo, Jane (committee member)

Creator Email kymmacneal@gmail.com,macneal@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-303435

Unique identifier UC11279545

Identifier etd-MacNealKym-4791.pdf (filename),usctheses-c40-303435 (legacy record id)

Legacy Identifier etd-MacNealKym-4791.pdf

Dmrecord 303435

Document Type Dissertation

Format application/pdf (imt)

Rights MacNeal, Kymberly G.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA