University of Southern California Dissertations and Theses

Domestic Violence And Indicators Of Social Change Within Nations: A Regional Perspective

(USC Thesis Other)

Domestic Violence And Indicators Of Social Change Within Nations: A Regional Perspective

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 70- 16,866 HARRELL, Roger Herman, 1936- DOMESTIC VIOLENCE AND INDICATORS OF SOCIAL CHANGE SffTHIN NATIONS: A REGIONAL PERSPECTIVE. University of Southern California, Ph.D., 1970 Political Science, general University Microfilms, A X E R O X Company, Ann Arbor, Michigan THIS DISSERTATION HAS BEEN MICROFILMED EXACTLY AS RECEIVED DOMESTIC VIOLENCE AND INDICATORS OF SOCIAL CHANGE WITHIN NATIONS: A REGIONAL PERSPECTIVE by Roger Herman Harrell A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Political Science) January 1970 UNIVERSITY O F SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 9 0 0 0 7 This dissertation, written by Harr ell............... under the direction of h.LLs.. Dissertation Com mittee, and approved by all its members, has been presented to and accepted by The Gradu ate School, in partial fulfillment of require ments of the degree of D O C T O R OF P H I L O S O P H Y l 7 Dean Date. DISSERTATION COMMITTEE Chairman TABLE OF CONTENTS Page LIST OF T A B L E S .............................................. iv LIST OF FIGURES..............................................vi INTRODUCTION .............................................. 1 Statement of the Problem An Overview of This Study Chapter I. VIOLENCE AND C H A N G E ................................ 15 Rapid Structural Change as a Precondition for Violent Political Behavior The Concept of Violence The Concept of Rapid Change II. FRAME OF REFERENCE.................................. 42 Control Structures The Social Role of Government The Structure of Conflict Regulation III. VIOLENCE WITHIN NATIONS: DIMENSIONS AND DATA . 72 The Universe of Cases Selection of Variables Method of Measurement Data Sources ii Chapter Page IV. RAPID RATE OF CHANGE: VARIABLES AND DATA . . . 98 Selection of Variables Data Sources Method of Measurement V. CROSS-NATIONAL ANALYSES ......................... 121 General Approach and Problems Correlation Analysis Stepwise Regression on Turmoil Stepwise Regression on Conspiracy Stepwise Regression on Internal War Stepwise Regression on the Total Magnitude of Violence Summary VI. LONGITUDINAL ANALYSES OF NATIONAL PROFILES . . . 158 Introduction Directions of the Correlations Selected National Profiles Summary VII. CONCLUSIONS........................................ 202 APPENDIXES Appendix A: Coding Instructions and Scales .... 215 Appendix B: Correlation Matrices for Seven N a t i o n s ...........................223 BIBLIOGRAPHY 231 LIST OF TABLES Table Page 1. Studies of Conflict Events ....................... 86 2. Percentages of Missing Data on Each Independent Variable by Nation (124 Cases)....................118 3. Time-Lag Correlation Coefficients................. 128 4. Cross-National Correlation Matrix ............. 133 5. Summary of the Stepwise Regression for the Magnitude of Turmoil ........................... 138 6. Summary of the Stepwise Regression for the Magnitude of Conspiracy ......................... 143 7. Summary of the Stepwise Regression for the Magnitude of Internal W a r ........................ 145 8. Summary of the Stepwise Regression for the Total Magnitude of Violence......................150 9. Summary of Regression Analyses .................. 152 10. Directions of Correlations within Variable-Sets, in Frequencies....................................160 11. Iraq: Correlation Matrix ......................... 165 12. Syria: Correlation Matrix ....................... 173 13. Lebanon: Correlation Matrix .................... 179 iv Table Page 14. Jordan: Correlation Matrix ....................... 183 15. Sudan: Correlation Matrix ....................... 187 16. Algeria: Correlation Matrix ..................... 191 17. Turkey: Correlation Matrix ....................... 223 18. Morocco: Correlation Matrix ..................... 224 19. Israel: Correlation Matrix ....................... 225 20. Iran: Correlation Matrix ......................... 226 21. Egypt: Correlation Matrix ....................... 227 22. Libya: Correlation Matrix ....................... 228 23. Tunisia: Correlation Matrix ..................... 229 v LIST OF FIGURES Figures Page 1. The Structure of Conflict Regulation ........... 70 2. Iraq: The Magnitude of Turmoil and the Rate of Change in Urbanization, over Time........... 166 3. Iraq: The Magnitude of Conspiracy and the Rate of Change in Urbanization, over T i m e........... 169 4. Iraq: The Magnitude of Internal War and the Rate of Change in Number of Radios, over T i m e ........................................ 170 5. Iraq: The Total Magnitude of Violence and the Rate of Change in Number of Radios, over T i m e ........................................ 172 6. Syria: The Magnitude of Conspiracy and the Rate of Change in Population, over Time . . .175 7. Syria: The Magnitude of Conspiracy and the Rate of Change in Newspaper Circulation, over T i m e ........................................ 176 8. Syria: The Total Magnitude of Violence and the Rate of Change in Newspaper Circulation, over T i m e ........................................ 177 vi Figure Page 9. Lebanon: The Magnitude of Internal War and the Rate of Change in Electrical Production, over T i m e ........................................ 180 10. Lebanon: The Total Magnitude of Violence and the Rate of Change in Electrical Production, over T i m e ........................................ 181 11. Jordan: The Magnitude of Turmoil and the Rate of Change in Number of Radios, over Time . . . 185 12. Jordan: The Total Magnitude of Violence and the Rate of Change in Number of Radios, over T i m e ........................................ 186 13. Sudan: The Magnitude of Internal War and the Rate of Change in Urbanization, over T i m e ........................................ 188 14. Sudan: The Total Magnitude of Violence and the Rate of Change in Urbanization, over T i m e ........................................ 190 15. Algeria: The Magnitude of Turmoil and the Rate of Change in Urbanization, over Time...........193 16. Algeria: The Magnitude of Internal War and the Rate of Change in Urbanization, over T i m e ........................................ 194 17. Algeria: The Total Magnitude of Violence and the Rate of Change in Urbanization, over T i m e ........................................ 195 18. Algeria: The Magnitude of Turmoil and the Rate of Change in Newspaper Circulation, over T i m e ........................................ 197 vii Figure Page 19. Algeria: The Magnitude of Internal War and the Rate of Change in Newspaper Circulation, over T i m e ........................................ 198 viii Underneath all of the norms of legal and institutional behavior in national societies lies the great beast, the people's capacity for outraged, uncontrolled, and bloody violence. Nieburg ix INTRODUCTION In recent years, along with the notable increase in occasions and intensity of internal violence, both in developing and highly developed societies, we have witnessed a growing body of systematic studies investigating the causes and patterns of such conflict. In general, these largely interdisciplinary efforts have presupposed the existence— and the measurability— of universally valid indicators of the possibility of internal strife. But given the complex subject area, we obviously cannot expect to find adequate consideration of all the most relevant variables in any one investigation. Statement of the Problem The purpose of the present study is to contribute to our understanding of the preconditions of violence within nations. Two specific problems are dealt with: one sub stantive, one methodological. The substantive question concerns the relationship between rapid rate of social 1 change and high magnitudes of violence in the Middle East and North Africa. The methodological problem will be to determine whether we can indeed claim cross-national validity for these change and violence variables. Several studies which illustrate the growing concern with the problem of internal political violence have contributed directly or indirectly to the formulation of this study. I deal with only one aspect of the overall problem of analyzing the preconditions of such violence; earlier projects have dealt with different aspects of the general problem, and thus serve as a context for the present research: these distinct aspects include, for example, possible preconditions such as relative deprivation or levels of economic development within nations. The cumu lative findings of these several large scale projects, along with the specialized findings of this dissertation, should suggest more meaningful approaches to dealing with a socially and politically relevant problem that should properly engage current social scientific inquiry. Arnold Rose defined social problems "as conditions which affect sizeable proportions of the population, which are out of harmony with the values of a significant segment of the population, and which people feel can be improved or eliminated.The serious recognition of the problem of domestic violence, and the desire to eliminate it, has led to the formation of our National Commission on the Causes and Prevention of Violence. This Commission is composed of several Task Forces, each of which is supported by the efforts of social science research institutions and the 2 individual efforts of social scientists. Other organized efforts to analyze violence within nations are the symposium at Princeton University to investigate internal war, and the study group on violent politics and modernization at the University of Chicago. The first group, under the direction of Harry Eckstein, was brought together by the Center of International Studies in 1961. According to Eckstein, their specific task was to write "on the application of contemporary social theories to the study of internal wars, including, for example, ^Arnold Rose, "Theory for the Study of Social Problems," Social Problems. IV (January, 1957), 190. 2 A recent product of this Commission was the report on the 1968 Chicago disorders by the Chicago Study Team directed by Daniel Walker, Rights in Conflict: Report to the National Commission on the Causes and Prevention of Violence (New York: Bantam Books, 1968). Particular note should also be made of the forthcoming studies of the Com mission's Task Force on History and Comparative Analyses, under the directorship of Ted Gurr, which will parallel the present research. functional theory, action theory, communications theory, theories of group dynamics, theories of attitude formation, 3 and the like." The second group was established in 1967 by the Adlai Stevenson Institute of International Affairs, and organized by Manfred Halpern in order to investigate 4 the relationship between violence and rapid social change. Attempts have also been made to provide a more empirical base of knowledge for understanding violence and its correlates. Examples of efforts to approach the problem from a quantitative base are Ted Gurr's Civil Strife Data Bank, and Ivo and Rosalind Feierabend's Data Bank of Political Instability Events. Several of the products of both of these data banks are used throughout this study. The nature of the phenomenon of violence transcends many of the usual analytic categories that represent the division of labor in the social sciences— i.e., violence is at the outset considered to be a psychological and socio- 3 One product of this group is Harry Eckstein, ed., Internal War: Problems and Approaches (New York: The Free Press, 1964) . 4 Two studies related to this group are Manfred Halpern, "Conflict, Violence, and the Dialectics of Modernization" (paper presented at the 64th Annual Meeting of the American Political Science Association, Washington, D.C., September 3-7, 1968); and Henry Bienen, Violence and Social Change (Chicago; University of Chicago Press, 1968). logical as well as a political phenomenon. Thus the present study embodies an interdisciplinary approach, and this approach exacts a certain price, just as it does in all interdisciplinary studies of social problems. The price is the lack of available disciplinary boundaries to serve as cut-off points in any inquiry. In order to delimit this investigation of the causes of violence, I have taken the "problem approach," which has been useful in exploring other complex areas from an inter disciplinary perspective— as, for example, in studies of political development and urbanization. By the problem approach, I mean simply that the problem definition itself determines the research boundaries. The variables are in dicated by the structure of the problem, rather than disci plinary competence or specialized concerns. The perimeters of this study are directly related to the theoretical defi nition of violence as a correlate of rapid social change. Redefining a phenomenon, however, does not preclude further definition. As Harry Eckstein has pointed out, there are still great difficulties in defining the boundaries of violence as a theoretical problem.^ The 5 Harry Eckstein, "Toward the Theoretical Study of Internal War," in Eckstein, ed., Internal War; Problems and Approaches, p. 12. difficulty primarily lies in the fact that while defining adds precision, it also excludes information. Karl Deutsch has emphasized the point that when we try to define pre cisely an aspect of a phenomenon we define away important things "in order to gain some really hard, reproductable, 0 verifiable knowledge about a limited subject." The present research is circumscribed in this sense. Domestic violence is defined in terms of conflict regulation, which in turn I define below as the political process. As pointed out above, the present study is not an isolated effort, and the relationship between rapid rate of change and violence dealt with here should be considered in the context of the work done by the Halpern group in researching other aspects of the relationship. This project, with its specific inquiries concerning social preconditions of violence, should be considered an indepth extension of the broad studies conducted by Gurr, Eckstein, and others. For example, I take fewer nations over longer periods of time and spend more time testing for cross national validity of variables. The data used here are 0 Karl Deutsch, "The Theoretical Basis of Data Programs," Comparing Nations: The Use of Quantitative Data in Cross-National Research, ed. by Richard Merritt and Stein Rokkan (New Haven: Yale University Press, 1966), p. 30. drawn from the Feierabend and Gurr violence data, and are used in exploring preconditions for violence that have not yet been investigated by others who have used these data. In brief, while this study's findings refer specifically to the relationship between rapid rate of change and high magnitudes of violence, it is recognized that rapid rate of change is only one of many possible preconditions, and that any present findings may be properly valued only in relation to parallel investigations. An Overview of This Study As indicated above, the general task undertaken in this study is to investigate the association between rapid rate of social change and high magnitudes of violence within selected nations, and to evaluate the indicators (variables) of social change and violence. I begin by selecting as a frame of reference a tentative definition of the structure of conflict regulation. Domestic violence and rapid social change must be theoretically defined in accordance with this framework. Diverse analyses of the structure of conflict regulation are brought together as a composite frame of reference in Figure 1. In this figure I have attempted to illustrate the more generally accepted conceptions of the patterns and mechanisms of conflict regulation, and to place the role of violence in the political process, defined as the process of conflict regulation. In addition to providing a conceptual basis for hypothesis generation, the theoretical definitions of violence and social change should enable us to determine when violence does or does not occur, and when there is an unusually high rate of change or when there is not, within a given society. Without working definitions of violence and social change, we can neither measure the two phenomena precisely, nor arrive at meaningful abstract comparisons and generalizations based on findings. In the present case these definitions should reflect the role of violence and social change in the process of conflict regulations, and should be nearly universal in terms of cross-national validity. The selection of cross-nationally valid variables with which to measure these phenomena remains an empirical question and is, to a substantial degree, the major concern of the latter parts of this study. The question of cross-national validity of variables becomes the paramount methodological problem. Other studies have, to a large extent, assumed that most variables 7 are valid across large numbers of nations (for example, see the studies noted in Table 1). It has recently been suggested, however, that the validity of variables may vary from one sociocultural region of the world to another. In view of doubts cast on the above assumption, and in view of the recent preference for regional groupings of nations, I have chosen to investigate selected variables for both violence and social change in nations from a common sociocultural region, a group of nations in which actors attach common meanings to certain types of behavior, such as violence, and for which we may choose certain common major variables affecting change— rapid rates of urbanization, for example, could be assumed to have similar disruptive effects in societies of a common sociocultural region. Throughout this study I refer to three different levels of analysis. Unless otherwise specified, these levels correspond to the size of the universe of nations 7 In order to facilitate reference to studies that deal with large numbers of nations from all regions of the world, I will use the term universalistic. The universal- istic approach or the universalistic level of analysis will hereafter refer to studies that have set aside the question of cross-national validity of variables in order to search for universal relationships between variables. used. Table 1 lists several studies that use the uni versalistic approach, based on assumed cross-cultural validity of variables. In an effort to ensure validity, the next level of analysis is based on a universe of selected nations from a common sociocultural region of the world. I take the position that at the regional level there is a better basis for assuming cross-national validity. The third level involves taking several nations individually and profiling the relationships between vio lence and rate of change variables on a nation by nation basis. The central question is whether the selected vari ables are comparably significant for all the nations in the region. If they do prove consistently relevant, then there will be some basis for assuming regional validity of the variables. If, on the other hand, there does not appear to be cross-national validity at the regional level, then we must also question the comparability of variables at the universalistic level, which includes nations from all cultural regions of the world. In the chapter on national profiles I take up, in more depth, the question of cross national validity of variables within the selected region. As far as possible, I attempt to use some of the variables for both violence and social change that have 11 been used by others at the universalistic level. These variables have been selected from various studies cited below. Primary attention is given to the variables for social change, because I feel that we are far less certain what is actually being measured by these variables from one nation to another, than we are in the case of variables relating to violence. For the purpose of substantive investigation, I hypothesize a positive relationship between high magnitudes of violence and rapid rates of social change. Several studies cited below have proposed that rapid social change is a correlate of domestic violence, inferring this either on the basis of observations of one nation, or on the basis of a universalistic approach. I test this proposition at the regional level, anticipating a positive association. No attempt will be made to explain why particular pairs of dependent and independent variables are associated in one way in one nation and in other ways in other nations. For example, though we might find a high positive correla tion between a rapid rate of change on the urbanization variable and a high magnitude of violence on the turmoil variable in Iraq, and a high negative association between these same variables in Algeria, no attempt would be made to explain the different implications of turmoil and urbanization in the two countries. Such analyses would require much more indepth review of particular conditions in each nation then can be provided here. This study deals with the question whether a rapid rate of change as measured on the independent variables is a precondition for high magnitudes of violence cross-nationally. Specifically, this study attempts to establish the degree (strength) and the direction (positive or negative) of the relationships between pairs of variables, in order to build a foundation for future studies concerned with why a relationship does or does not exist. The methodological question concerns the cross-national validity of dependent and independent variables in a common cultural region. It is initially assumed that all variables are comparable in direction and significance for all nations in the region. Thus we an ticipate that most relationships between high magnitudes of violence and rapid rates of change variables will be con sistently positive or negative for all nations in the area. If these variables do not appear valid cross-nationally then a series of questions arise: we must ask whether these variables measure the same things from one nation to another; how much more invalid they might be if used across 13 different sociocultural units, as in the above-mentioned universalistic studies; and so on. In Chapters I and II, I deal with theoretical definitions of violence and social change; operational definitions are set forth in Chapters III and IV, which explain the specific variables and the procedures for measurement and scaling. The statistical analyses are discussed in Chapters V and VI. In Chapter I I attempt to conceptualize the somewhat elusive phenomena of violence and social change, in an effort to supply a theoretical basis for deriving the central hypothesis. Chapter II deals with definitions of control structures and the social role of government, and proposes a frame of reference in terms of the process of conflict regulation and the role of violence in this process. The next two chapters primarily provide operational definitions which form the basis for measuring and analyzing the phenomena defined in Chapters I and II. Chapter III explains the basis for selecting the nations (thirteen Middle Eastern and North African coun tries) , and the time periods (yearly time segments from 1948 to 1965) ; the same chapter presents the variables relating to violence, and a discussion of the procedures for scaling the data and determining the dimensions of violence. The variables used to measure rate of social change are set forth in Chapter IV, along with the pro cedures for collecting and processing the rate of change data. Chapters III and IV, then, define the empirical perimeters which form the basis for the analyses carried out in Chapters V and VI. The statistical techniques used in Chapters V and VI include the product moment correlation and stepwise regression analyses. Chapter V employs cross-national and longitudinal comparisons to analyze the relationship between rapid rates of social change and high magnitudes of domestic violence. In this chapter the regression analyses provide the basis for exploring the relative and cumulative importance of the rate of change variables as preconditions for high magnitudes of violence. In Chapter VI I consider national profiles over the eighteen one-year time periods with particular reference to the direction and degree of correlations of rate of change with magni tudes of violence. CHAPTER I VIOLENCE AND CHANGE Rapid Structural Change as a Precondition for Violent. Political Behavior As is generally the case, one can go back at least as far as the Greeks to find earlier attempts to explain phenomena which are still in need of explanation today. For example, Aristotle was concerned about the relationship between changing patterns of behavior and political sta bility. In his words, "the habit of lightly changing the laws is an evil, and, when the advantage is small, . . . the citizen will not gain so much by making the changes as he will lose by the habit of disobedience. . . . For the law has no power to command obedience except that of habit, which can only be given by time, . . .1 1 ^ His statement highlights an important aspect of the problem under ^Aristotle, Aristotle's Politics and Poetics, trans. by Benjamin Jowett and Thomas Twining (New York: The Viking Press, 1957), p. 45. 15 16 investigation in this paper, and also exemplifies the tendency to assume a direct one-to-one relationship between change and violence. The obvious danger is the trend toward a conservative bias against change itself, the implicit feeling that change is synonymous with violence. The intent of this study is not to suggest that change is undesirable. On the contrary, this research contributes to the examination of social conditions that permit or cause violence to break out while the necessary process of change is taking place. Awareness of the preconditions of violence may enable us to try to prevent or at least pre pare for the outbreak of violence as we increasingly 2 promote rapid rates of change. Discussing the background of the French Revolution, Alexis de Tocqueville stated one of his findings as follows: A study of comparative statistics makes it clear that in none of the decades immediately following the Revolution did our national prosperity make such rapid foreward strides as in the two preced ing it. 3 Olson deals with this same question in a more-or- less conservative but reasonable manner as he discusses the desirability of promoting rapid economic development: Mancur Olson, Jr., "Rapid Growth as a Destabilizing Force," The Journal of Economic History, XXIII (December, 1963), 552, 529-552. 3 Alexis de Tocqueville, The Old Regime and The 17 He indicates a thirty- or forty-year period during which the needs and expectations of the people rose at a rapid pace. Tocqueville1s argument parallels the central hypothesis of this study. In effect, Tocqueville1s argument is that rapid social change disrupts the structure of society to a point where it is unable to provide adequate controls over the behavior of its citizens. He moves closer to a consider ation of the control features in the basic social structure when he compares the French and English revolutions. He points out that in England sweeping changes were made in 4 the manifest control structures, while very little change took place at the level of secondary laws or the customs and habits of behavior in the society. In France, on the other hand, it was in the latent control structures that French Revolution (Garden City: Doubleday Anchor, 1955), pp. 171, 174. 4 . . . I am making an analytical distinction between "latent control structures," and "manifest control struc tures." Manifest control structures refer to those struc tures that are designated as most overtly responsible for preventing deviance. The latent control structures are pervasively present in the structure of society and have consequences for the control of deviance, but are not specifically assigned the task of social control. For a further discussion of this distinction see Albert Cohen, Deviance and Control (New Jersey: Prentice-Hall, 1956), p. 40. 18 changes were first taking place and resulting in confusion and lack of orientation, while at the same time no signif icant changes affected political institutions (manifest control structures). Thus in England the upheaval was contained because the elemental, nongovernmental control structures were intact— or, as he put it, "the apex was 5 shattered, but the substructure stood firm"; m France the changes in the substructure account for the violent nature of the French experience. The variables which are ultimately used to measure rates of change in this study are selected to tap what Tocqueville has called the sub structure, and what I am calling the latent control structures. ^ This approach to the relationship between rate of change and violence has more recently been stated in 5 Tocqueville, The Old Regime and The French Revo lution. pp. 201-202. ^Lasswell conceptualizes the problem in much the same way, but he tends to stress the importance of erratic rates of change. His discussion of this phenomenon is directed at the international system, but it is so basically a sociological approach that it is easily incorporated into this discussion. I am primarily focusing on rapid rather than erratic rates of change, but both types, it is assumed, have similar disorienting effects on behavior. See Harold Lasswell, World Politics Faces Economics (New York: McGraw-Hill, 1945), pp. 36-37. 19 hypothetical form by Harry Eckstein, under the heading of "characteristics of social process." He considers this relationship to be one of the most interesting, and has stated the general hypothesis as follows: political vio- 7 lence is generated by rapid social change. This same general hypothesis is used in this study, and is stated as follows: rapid social change is a precondition for high magnitudes of violence. The use of this hypothesis assumes the existence of composite measures for both rate of change and violence. It will become evident as we proceed, however, that composite measures are unfortunately neither reliable nor valid at this time. Before dealing with this question in more depth the two major concepts need to be assigned more specific theoretical definitions. These further definitions are developed in the remaining sections of this chapter, and the problem of validity becomes increasingly important as we proceed. The Concept of Violence Any theoretical definition of violence will retain 7 See Eckstein's discussion of the major hypotheses in "On the Etiology of Internal War," History and Theory. IV (1965), 152. some of the imprecision and confusion that now obscures its meaning. It would be premature, and perhaps presump tuous at this point, to attempt an exact definition of violence at the theoretical level. Thus I am only attempt ing to expose and clarify some of the ambiguities and complexities that one encounters when trying to deal with the idea of violence. The purpose is to suggest that violent events provide certain indications about social system. For example, I assume that we do not study riots just to become specialists on riots. The theoretical definition developed below presupposes that a riot, for example, has an impact on the system, and is furthermore a product of the system. This is an attempt to provide a theoretical framework for observing, measuring and interpreting certain types of violent events. The structure of a conflict regulation process usually defines what is acceptable behavior in conflict situations. It is assumed that the political structure provides a set of acceptable alternative channels for processing a conflict situation— channels designed to direct the behavior of all involved actors. If an actor behaves in an unexpected way, then his behavior will either command acceptance as a new. alternative, or be condemned and resisted by other actors. If the extraordinary behavior is accepted, then the structure of the political system undergoes some degree of change as a new alternative channel is included in the system. This is a case where violence brings about change and is not an area that I propose to go into at this time. If the extraordinary behavior is considered unacceptable, then it will be opposed by other actors; the behavior would then qualify as viola tion of acceptable patterns, and would constitute what I propose to call violence. A judgment is made, for one reason or another, of some extraordinary type of behavior, and it is accepted or rejected by other actors. Violence is, then, defined as some observable behavior that is perceived as unacceptable and opposed by other actors. The degree to which these latter actors are able to influence or curtail the violence will generally depend directly upon their political capacity, regardless of whether they are individuals, groups, or governments. When Johan Galtung defines nonviolence as "any activity that aims at enlarging action-spaces by including actions or making them more attractive," thus assigning a positive rather than merely a passive role to nonviolence, he defines violence as "any activity that aims at reducing 22 g human action-spaces by excluding actions ..." In effect he defines violence from the point of view of the actor attempting to curtail the extraordinary behavior of another actor. My definition is simply broader. Galtung's conception stresses the idea of one actor influencing the behavior of others and, as such, essentially accords with my definition of violence; indeed, whenever one engages in violent behavior (i.e., "activity that aims at reducing human action-spaces") he is, by definition, meeting with resistance. This becomes clear when we consider the actor's evaluation of possible benefits and perils of undertaking violence. According to Galtung, he must take into account (1) the intrinsic value of the behavior; (2) the internal ized personal sanctions; and (3) the institutionalized 9 social sanctions. In this study I am only dealing with social sanctions; any behavior that violates these sanctions is likely to be considered as unacceptable by actors who support the existing system. Galtung has also pointed out that the concept of nonviolence is overloaded with emotional and ideological 8 Johan Galtung, "On the Meaning of Nonviolence," Journal of Peace Research. II (1965), 235. 9Ibid.. pp. 230-231. 23 connotations."^ This observation, I feel, applies at least equally to the concept of violence. Therefore, in order further to define perimeters of the concept of violence, it is necessary at this point to isolate certain emotionally charged connotations associated with it. I am primarily concerned with the ideas of aggression and force. Peter Lupsha1s discussion of aggression suggests two basic ways of approaching the concept: either as an innate, instinctual behavior, or as a learned behavior. I find that when I consider aggression a significant element in violence, the latter approach is more useful. What I say about aggression, then, only relates to that portion of it that can be determined as learned behavior, and I will argue that the acceptance or rejection of aggression is also a learned response— i.e., the level of aggression considered acceptable is determined by the given cultural context. Galtung defines aggression "as drives towards change, even against the will of others," "somewhat in the ~^Ibid., p. 228. "^Peter Lupsha, "On Theories of Urban Violence" (paper presented at the 64th Annual Meeting of the American Political Science Association, Washington, D.C., September 3-7, 1968), p. 12. 24 sense of self-assertion, but only insofar as this self- assertion implies an effort to change social relations, 12 i.e., no longer to comply with existing conditions." This closely approaches my definition of violence, but I would repeat that only when the drive toward change (extra- ^ordinary behavior) is opposed by others do we have violence. In societies like the United States we have something called "the aggressive young businessman, " and many actors place a high value on aggressive competition and innovation in pursuit of certain goals. This kind of aggression is generally admired and is seldom considered violence in our cultural context— while the same behavior in another society might be considered a violation of custom. The acceptance of aggression exists in every society to some degree, and there is no evidence that it has a general level at which it universally becomes unac ceptable. In other words, aggressive behavior is not always violence. Likewise, the use of force is not always violence. Extraordinary behavior may be undertaken without the use of any physical force; force, moreover, is only one of several 12 Johan Galtung, "A Structural Theory of Aggres sion, " Journal of Peace Research. I (1964) 95. types of coercion or influence that might be used to discourage or promote this behavior. Unacceptable behavior* for example* may be "bought-off*" or its acceptance may be 13 purchased. We know very well that violence and force frequently co-appear. Nevertheless* violence and force are different kinds of things* and do not necessarily appear in the same conflict situations. Take* for example* a situation in which a group of students occupies a univer sity administration building without forced-entry* and when police arrive the students are arrested without resistance. Other types of demonstrations can also take place without the use of force* and in each case if the government or 14 other regime-oriented actors define the demonstration as unacceptable and move against it, then it may be seen as a violation of or detriment to the established social order— an event that would be herein defined as violent. All events are considered acceptable or unacceptable in terms 13 Amitai Etzioni* The Active Society; A Theory of Societal and Political Processes (New York: The Free Press* 1968)* pp. 364-375 (see particularly his discussion of the general methods of gaining compliance with structures). 14 By regime-oriented actors I mean individuals or groups who generally desire the maintenance of the estab lished structure of society— actors who might otherwise be referred to as elites, dominants* central or top-dog actors. 26 of other actors' reactions to the given event. Another reason for establishing the analytical distinction between violence and force is that there might well be situations, such as debates, elections, co-options, and coups, where force is considered courageous and looked upon favorably, not as violation of the system. In other words, there are cases where a system perceives the use of force as legitimate; as, for example, when force is ex erted by the dominant actors. Other examples would include situations that call up an "ends justifies the means" type reaction, or occasions when the use of force needs no justification (as when used by the regime). The concept force, it seems, is still too indeterminant to be useful in comparative analysis— unless in a highly structured situa tion where it can be assumed that most of the actors attach a similar significance to a given amount of force. Up to now I have argued that aggression and force are not universally valid indicators of violence and I have suggested another basis for determining the existence of violence. Now I would like to discuss briefly how I con sider aggression and force to be relevant in any analysis of violence. The basic reason for analytically separating the categories aggression and force from the definition of 27 violence is to produce a relatively uncomplicated basis for classifying certain behavior as violent. I would now like to define a certain critical level of aggression or force, a level at which aggressive or forceful behavior may, due to resistance encountered, qualify as violence. Violence may erupt in a conflict situation, even though the situation is in the process of being resolved within acceptable procedural channels. Violence may occur if one, or all, of the actors use methods which are considered too aggressive or employ an unacceptable amount of force. For example, to physically abuse one's opponent in Parliament would cause a reaction that could close this channel to the actors who were too aggressive or too forceful. The r alternatives would then be for the violators to return to using acceptable methods, or to pursue a deviant (violent) course outside of Parliament. Aggression and force are resources that are always to some degree available, but which can only be used in acceptable amounts. Here again, what is acceptable can only be determined within the system in which the behavior takes place. Thus, violence can occur as a result of how the actor chooses to propel himself through the political process. Or it can occur because of where the 28 behavior takes place; standards of acceptable behavior outside of the acceptable political structures will differ from standards within these structures. This approach permits the classification of a nonaggressive, nonforceful demonstration as unacceptable behavior, if regime-oriented actors decide that the street is not the place where conflict should be expressed. The utility of this approach is again manifest if we apply it to a case in which it is acceptable for the police to bring order in the streets— they are where they should be. If, however, they employ too much force in the situation then they might encounter opposition to how they brought order to the streets. In this case the police action would come to be classified as violence, even though the system generally permits police action in the streets. Aggression and force are then intervening vari ables that can be employed at any time and anywhere within the political process, and can at any point push a conflict situation out of the procedural realm. Also, unacceptable amounts of aggression and force can be considered as indi cators of violence when they appear in the procedural channels. In such a case, if aggression and force do not return to an acceptable level, or the system refuses to 29 lower its level of acceptance, the situation will be pushed to the violence category as the actors find procedural channels closed to them. When a conflict situation is initially regulated in procedural channels but unacceptable amounts of aggression and force are employed, then aggres sion and force have an either-or influence which can cause the situation to be reclassified from the procedural to a violent approach. When a situation is initially violent or becomes violent, then increased amounts of aggression and force have a more-or-less influence on the situation, affecting the magnitude of the violence rather than precipitating it. In any case, violence is not indicated merely by the presence of either aggression or force. In both types of cases, it is the perception of and the reaction to an unacceptable condition that indicates the presence of violence. I have pointed out that what is unacceptable varies from system to system and that there is no such thing as a universally acceptable amount of aggression or force. Conditions within each system determine the acceptable levels of aggression or force, the levels at which violence may erupt or increase. Acceptable behavior for some determines how they must play the game, but for others it determines how they will win the game— and most actors 15 want to be in a winning position. If we accept the proposition that, for the most part, dominant actors define what is acceptable, while the subordinant actors seek to improve their position in the hierarchy of the system, then when the dominants permit unusual amounts of upward mobility we have a case of acceptable structural change; if they oppose it, then pressure for upward mobility is defined as unacceptable behavior. The degree of flexibil ity in the social hierarchy, and the perceptions and attitudes of both dominant and subordinant actors, interact to make for the differences in acceptable levels of aggres sion or force from one society to another. The regime and regime-oriented actors also have acceptable or procedural alternatives for resisting extra ordinary behavior. When actual behavior does not match ideal behavior (as defined by the regime and regime-oriented 15 See, for example, Nieburg's discussion of the role of law in defining the elite structure of a system: H. L. Nieburg, "Uses of Violence," Journal of Conflict Resolution. VII (1963), 43-54. Also see Scott's discussion of the preconditions of aggression present in the social organization: John Scott, Aggression (Chicago: The Uni versity of Chicago Press, 1958), pp. 90-105. 31 actors), then the procedures for punishing those who deviate too much are used. The procedures utilized to punish violators are, for the most part, observable in the structure of the system."^ When dominant actors do not have established procedures for punishing certain types of deviant behavior, then they will find themselves in an unstructured situation and the dominants themselves may behave in a violent manner. Thus, the dominant actors are also subject to the guidance and control structures which form the "maze," and present all actors with cues as to 17 what is acceptable behavior. Structures usually impose restrictions on all 18 actors, even though seldom equally. Inequallity results from the differing status of different actors in the system. There are, of course, psychological factors conditioning actors' perception of violence, but as explained above, 16 See Scott's discussion of this point m Scott, Aggression, p. 98. 17 See Anthony Wallace, "Mazeway Disintegration: The Individual1s Perception of Socio-Cultural Disorganiza tion, " Human Organization. XVI (Summer, 1957), 23-27. 18 There is considerable flexibility in these structures, as discussed to some extent in Cohen, Deviance and Control, pp. 7, 6-11. 32 they will not be considered. We are primarily concerned with relating the division of labor and authority, as defined by the structure of society, to ways of perceiving and responding to given kinds of behavior at various levels of society. Peter Lupsha has pointed out that urban riots involve the questioning of the authority patterns estab- 19 lished by the dominant actors. Not only are dominant actors not accepting the challenge to their authority; such challenges are particularly rejected when they can be defined as unacceptable by the dominant actors in order to justify the occasional use of more force against deviants. The magnitude of the reaction against the challenges to authority will vary from system to system depending on a variety of conditions; the cross-system variance in these conditions, reflecting the structure of each system, will determine the validity of selected variables. The vari ables of violence, then, may differ from one system to another, though the theoretical definition of violence will remain the same: that is, violence will by definition be indicated by the presence of unacceptable and opposed 19 Lupsha, "On Theories of Urban Violence," p. 16. 33 behavior in the respective systems. In conclusion, at the theoretical level of analysis, violence is some observable behavior perceived as unacceptable and opposed by other actors. The Concept of Rapid Change The relationship between change and violence can and does go in two directions— at this point, either of the two concepts could be justifiably considered the dependent variable. I have selected change as the independent variable because I am interested in measuring the kinds of violence that correlate with rapid social change— and, as pointed out in the opening section of this chapter, I am exploring the possibility that rapid social change is a precondition for violence. Having defined violence, I can now limit my observation of rapid change to disruptive aspects. This operational focus should not be taken as 20 implied definition of the concept of social change. 20 For a recent example of a work that selects vio lence as an independent variable, and then limits the oper ational definition of violence to those aspects that relate to change, see Bienen, Violence and Social Change, pp. 102, 106. Also, for a more politically motivated attempt to establish violence as an independent variable see Georges Sorel, Reflections on Violence (New York: Collier Books, 1961), pp. 26-56. 34 When reflecting on the social role of government, Karl Deutsch and others have pointed out that political commitments are not bounded by formal political institutions but rather must tend to reflect the needs and expectations of all politically relevant segments of any population. They make this point in relation to effects of changes in social conditions on social needs and expectations. Their final concern is with the effects of change on the role and 21 condition of government. The present study adopts a similar concern, relating rapid social change to violence in terms of the disorienting effects of such change as it blurs the expectations of actors in the political process— i.e., as it affects their expectations of goals and pro cedures and of the behavior of other actors in the political process. Any investigation of disruption of expectations or cues in the structure of the political process must go beyond the political system, and include, for example, areas as vast as those sometimes dealt with by structural 21 Karl Deutsch, Harold Lasswell, Richard Merritt, and Bruce Russett, "The Yale Political Data Program," Comparing Nations, ed. by Richard Merritt and Stein Rokkan, p. 82. 35 22 functionalists. In that functions are considered to be universal, however, many structural functionalists would argue that the place to measure differences or change is 23 in the structure of society. By rapid rate of social change, then, I am primarily concerned with structural change, and I am not only referring to something akin to social mobility. I also mean the development of new roles and structures. The problem of integrating new actors into old roles can be disruptive, but the socialization process can usually be effective here, and new actors can, to a degree, have expectations about the established roles and structures. While the problem of recruitment into established struc tures will be considered, I am primarily dealing with the changes that produce the need to develop new structures. The basic assumption here is that there is little initial awareness of the cues and expectations needed to relate the actor to the new structures. More precisely, I am referring 22 See, for example Talcott Parsons, The Social System (New York: The Free Press, 1951), pp. 24-68. 23 For example see Fred Riggs, "Structure and Function: A Dialectical Approach" (paper presented at the 63rd Annual Meeting of the American Political Science Asso ciation, Chicago, September 5-9, 1967), p. 5. 36 to what Eckstein has called changes in the control struc- 24 tures of society. Along this same line Galtung has also argued that 25 social change is primarily structural. For example* he focuses on the structures that rank social actors both horizontally and vertically— ranks that result in dominant and subordinant relationships. The structural changes that I am referring to* then* displace actors and frequently asign them new roles* for whatever reason* and the result is disequilibrium in authority patterns. This can* for example* make for social conflict between dominant and subordinant actors by stimulating awareness of their condi tion among the subordinants* and by eroding the justifica- 2 6 tions for an existing distribution of social roles. These structural changes are often closely related to modernization* which entails moving from a more tradi tional type social arrangement to a more structurally 24 Eckstein* "On the Etiology of Internal Wars*" p. 13 5. 25 Galtung* "A Structural Theory of Aggression*" p. 112. 26 For a discussion of this problem see Lewis Coser, The Functions of Social Conflict (New York: The Free Press* 1964), p. 37. 37 differentiated and functionally specific arrangement. In this regard, Neil Smelser points out that structural dif ferentiation brings into conflict new and old activities, 27 norms, rewards, and sanctions. There is often a period of time where the new system of behavior is only partially internalized and there are areas where the effectiveness of the old system has eroded. More specifically, Collomb and Valantin have termed this sort of situation as a rupture 2 8 of the human continuum. They are concerned, for the most part, with the disruption of the socialization process. The outcome of these transitional periods (which may take decades) is often confusion and lack of orientation on the part of social actors. Ross Stagner discusses this problem in terms of the need for a predictable environment 29 which is strongly related to a stable milieu. Actors need not behave in the same way all the time, but they 27 Neil Smelser, "Toward a Theory of Modernization, " Social Change: Sources. Patterns and Consequences, ed. by Amitai Etzioni and Eva Etzioni (New York: Basic Books, 1964), p. 270. 28 H. Collomb and S. Valantin, "Patterns of Mother ing, Organization of the Personality and Rapid Social Change," International Journal of Social Science. XX (1968), 445. 29 Ross Stagner, "Personality Dynamics and Social Conflict," Journal of Social Issues, XVII (1961), 29. 38 should have relatively stable expectations of the conse quences of their behavior and that of others. Learned behavior is frequently useless in periods of rapid struc tural change, and often leads to increasing frustration which usually lasts until the actor learns to relate to 30 the new structures. Such situations are conducive to periods of relatively unstructured behavior (approaching anomie) which has a short-run multiplier effect because of the fact that other actors, regime-oriented as well as non regime-oriented, are unable to develop expectations about the behavior of others. These types of conditions frequent ly make it difficult to apply sanctions, and can make social orientation difficult for both dominant and subordinant actors. Etzioni suggests, in fact, that one of the social roles of the government is to have the capacity to respond to rapid social change with meaningful and adequate reforms. If government fails to respond effectively, then alienation and/or anomie tend to accumulate as preconditions for 30 Alexander relates this trend to situations with and without hope, attacking the problem I only touch upon in Chapter II of this paper. Franz Alexander, "A World Without Psychic Frustration," American Journal of Sociol ogy. IL (1944), 468. violence. In addition to the social changes taking place in the drive toward modernization, there is usually demand for new structural arrangements in the political system itself, arrangements which might add to the burden already over taxing the capacity of the political structures to provide orientation. The assumption is that during periods of rapid social change the latent control structures are ineffective, and the burden of control falls increasingly on the governmental structures. When, however, in such periods of rapid social change, there are also changes taking place in the political structures, then disorienta tion is compounded. Political changes are particularly disorienting when attempted in a relatively short period of time. Eric Nordlinger identifies certain familiar phases of rapid political development: efforts to forge national identity; expansion of central government; formation of mass polit ical parties; development of mass electoral 31 Etzioni and Etzioni, "Introduction, " in Social Change: Sources. Patterns and Consequences, p. 405. Also see Deutsch1s discussion of rate of change and governmental capacity, using the analytical tools of load, lag, gain, and lead. Karl Deutsch, The Nerves of Government (New York The Free Press, 1966), pp. 189-191. 40 32 participation. He suggests that, if change proceeds too rapidly, either the preconditions of repression or of violence itself will result. Thus, here as well as in the social system, a rapid rate of change may be disruptive, however desirable the specific changes might be. Thus I am not concerned with all of the aspects and implications of social change, but rather with only one aspect of it— the rates at which it takes place. The underlying assumption is that rapid change produces periods with large numbers of alienated actors who lack orientation, cues, and expectations in terms of the established social 33 and political structures. The focus then is on rate of social change, regardless of whether it results from rapid economic or social decline or growth— change in either direction will alter the relative positions of the actors, and the relative importance of one structure over 32 Eric Nordlinger, "Political Development: Time Sequences and Rates of Change," World Politics. XX (April, 1968), 500. 33 Along with other sources cited in this section, Eckstein also supports this assumption in reference to authority structures. Harry Eckstein, A Theory of Stable Democracy. Research Monograph No. 10 (Princeton: Center of International Studies, Princeton University, 1961), pp. 22-23, 40-41. 41 34 another. More specifically the focus is on the condi tions created by the rapid rates of change, rather than the extent or stage of decline or progress. In Eckstein's words, "there is overwhelming evidence to show that 1 anomie. 1 the feeling that one lacks guidelines to behav ior, is increased by rapidity of change in any direc- „35 tion. ..." For a discussion that directly expands this point, see Olson, "Rapid Growth as a Destablizing Force," p. 543. 35 , Eckstein, "On the Etiology of Internal Wars," p. 152. CHAPTER II FRAME OF REFERENCE Control Structures The notion of control suggests such related terms as legitimacy, power of force. This is because the posses sion of these attributes by certain actors generally coincides with their ability to influence the establishment, change or maintenance of structured behavior in a society. I am not, however, focusing on actors who determine structure, or on their capacity or motivations, but rather on the subtle and pervasive control features that reveal the existence of the structured behavior itself, and on the controlling influence of informal, acceptable ways of doing things. Structured behavior is considered something separate from the actual relationship between actors. It is a recognizable pattern of behavior.'*' The significant "^Riggs, "Structure and Function: A Dialectical Approach," p. 7. 43 implication is that structure derives from repeated occur rence of observable consistent behavior. The repetition of observable behavior is in this sense structural development (this applies to the development of political structure as well as social structure). When I use the term devel opment, I am referring to the idea of development of structures— regardless of what the structures come to be 2 called. Even instability involves a structure: an unstable situation may persist and begin to supply actors 3 with cues facilitating orientation. An important aspect of all structures is the control exerted over the behavior of actors. The authority of structures is ubiquitous, but is frequently latent. In some societies the major source of control is centered in the government, while in others social controls may be to a large extent diffuse in the structure of society. 1 am not concerned here with the degree to which 2 I use the term development in much the same sense as Herbert Blumer, "The Idea of Social Development," Studies in Comparative International Development. II (1966), 7. 3 Joseph Firestone has illustrated this point with his example of post-war Italy, using perceptual data: "Political Conflict: A Theory and Comparative Analysis in Five Cultures" (unpublished Ph.D. dissertation, Michigan State University, 1965), p. 225. 44 authority is centered in governments, but rather with the breakup of nongovernmental structures and consequently of latent controls during periods of rapid social change. For example, a breakdown in the structure of the extended family will place a greater burden on the manifest control 4 structures such as the government. It is through one or both of these types of structures that actors learn how to respond to cues and to have mutual expectations of behavior. Certainly some patterns of behavior will have been institutionalized in laws, political parties or bureaucracies. Governmental structures like all other structures, however, provide only limited controls and if the less obvious or latent control structures of society become temporarily ineffective then the government will soon be overburdened and unable to control behavior in broad areas of society (except in any case where an 4 For a good analysis of the control provided by a family structure, in the above sense, see Collomb and Valantin, "Patterns of Mothering, Organization of the Per sonality and Rapid Social Change," pp. 431-447, passim. 5 Johnson, for example, argues that these expecta tions permit actors to orient their behavior and that this is essential for maintaining the structure of any society. Chalmers Johnson, Revolutionary Change (Boston: Little, Brown and Co., 1966), p. 8. 45 extensive totalitarian regime exists; I would, however, consider this a temporary condition, too expensive socially and economically to maintain for long periods of time). In this study I deal with social meanings as re flections of values implicit in social control structures. I assume that actors derive their social cues for appropri ate or inappropriate behavior from the social structures, and further assume that when structures are disrupted so are the cues. The final assumption is that disrupted cues result in confused meanings. It is further assumed that confused meanings precipitate and prolong conflict situations. Although we cannot consider all of the possible specific causes of social conflict, one may want to recall the several major types of factors leading to conflict situations.^ Maurice Duverger discusses biological, psychological, demographic, geographical, socioeconomic and cultural sources of con flict. But regardless of which of these sources may contribute to the conflict situation, I am primarily con cerned with disrupted or confused meanings as causal factors in conflict. 6 Maurice Duverger, The Idea of Politics, trans. by Robert North and Ruth Murphy (Chicago: Gateway, 1966), pp. 3-86. 46 A situation is considered political when govern ment structures are used to resolve or regulate a conflict between two or more nongovernmental actors^ or when the government itself is an actor in the conflict situation. The Social Role of Government The social role of government implicitly relates to the question of the major purposes of government— in other words, why a government? This question, while basic, is only remotely relevant to the problem of this study, and if dealt with too extensively adds little to precise definition. I will therefore merely note two major opposed perspectives adopted by political scientists. My purpose is to clarify the basic assumptions underlying the study of conflict regulation. Duverger contrasts these opposed political philos ophies as follows: Behind all the systems of values and all individual judgments there are generally to be found two basic attitudes. Ever since men have been reflecting on politics they have oscillated between two dramatically opposed interpretations. According to one, politics is conflict, a struggle in which power allows those who possess it to ensure their hold on society and to profit by it. According to the other view, politics is an effort to bring about the rule of order and justice, in which power guarantees the general interest and the 47 common good against the pressures of private interests. In the first case, politics serves to maintain the privileges of a minority against the majority. In the second, it is a means of realizing the integration of all citizens into the community and of creating the just state of which Aristotle spoke so long ago.^ I suggest that while the second interpretation indicates the goals of the government, the first interpretation describes the process by which these goals must be achieved. These two views of politics have developed into polar positions— one of which sees politics as conflict, a struggle in which the role of government is to establish a process for conflict regulation and nonviolent struggle, while the other concentrates on the ideal goals of govern- g ment, such as community building and cultural integration. Basic assumptions about man and the role of government in society are major determinants of the aca demic's approach to the study of politics, and indeed of real directions in politics itself. I would like to clarify 7 Duverger, The Idea of Politics, p. xii. 8 For the best overall effort to bring together some of the important works of both the integration and conflict approaches, across all of the major levels of analysis, see J. K. Zawodny, Man and International Relations: Contribu tions of the Social Sciences to the Study of Conflict and Integration, (2 vols.; San Francisco: Chandler Publishing Co., 1966). 48 further the position of this study by making explicit some of its basic assumptions. The Nieburg quote at the beginning of this study does to a degree reflect the Hobbesian view of man and the role of government in soci ety. As Etzioni points out, however, the reason for starting from such a view is to investigate how we might 9 escape from the anarchy of violence, not simply to under score unpleasant realities. The regulation of conflict and the elimination of social violence was viewed by Hobbes as the primary and most important social role of government. The premise is that there is sufficient evidence from man's history of violence to suggest that if man is to be con tained in a social system then controls external to man himself must be provided. This view is implicit in the present study. The focus of this study is based on the position that conflict regulation is one of the most important roles of the government. In fact, as defined in this study, political systems are simply the manifest control structures necessary for conflict regulation in a society. The frame of reference for this study is based on 9 For a further discussion of this point see Etzioni, The Active Society, pp. 354-355. 49 the general structure of conflict regulation as observed and analyzed in various studies. Too few of the studies used to construct the frame of reference have an adequate empirical base, however, and thus the proposed structure can only be considered tentative at this time. It does, nevertheless, provide the conceptual context for consider ation of the problem of violence in politics. The Structure of Conflict Regulation Conflicts, as well as structures for regulating conflict, have been observed from many angles, and have produced a body of intuitive, empirical and mixed research findings. In this regard the composite frame of reference, as an illustration of the structure of conflict regulation, has a two-fold purpose: first, to provide a means of systematically classifying the variety of generalizations that have been made about the structure of conflict regula tion; and second, to provide a system for classifying types of observable political acts (such as voting, judicial decisions, and riots) in the overall process. The frame of reference also provides a theoretical definition of what has, in various places, been termed the structure of the conflict regulating process. Etzioni has 50 referred to the process as "encapsulation," in which "conflicts are modified in such a way that they become limited by the rules (the 'capsule'). The rules exclude some modes of conflict, while they legitimate other modes." One of the most important social roles of govern ment is to develop and maintain an effective capsule, or structure for regulating social conflict. Kenneth Boulding refers to this as one of the greatest organizational prob lems of mankind— in any social unit. In general, he is referring to procedural institutions for the control of conflict, and more specifically for the control of vio lence . In an attempt to isolate types of actors in con flict situations Ted Gurr has pointed out three basic roles for actors: regime-oriented, dissident-controlled, and neutral. The dissident, or opposition, actors present the greatest continual challenge to any procedural structures provided by the government and supported by the regime- oriented actors. The sharpest challenge arises when the 10 Etzioni, The Active Society, p. 587. ■^Kenneth Boulding, Conflict and Defense: A Gen eral Theory (New York: Harper, 1963), p. 324. 51 opposition actors attempt to establish alternative patterns of behavior— the most threatening of which provide the 12 means and cues for violent behavior. Opposition may, however, simply aim at influencing the way in which conflicts are regulated or the way a particular conflict situation is resolved. Opposition- oriented actors need not necessarily be set on the revision 13 or destruction of a given structure. In the following discussion, three basic approaches to conflict regulation are discussed: (1) procedural means, (2) avoidance, and (3) violence. I discuss the ways in which each of these three approaches are used to control or manipulate behavior in a conflict situation. The above discussions of Etzioni and Boulding refer to the capsule, or to procedural provision for regulating conflict, as society's best or primary resource for resolving conflict situations among its members. The use of the term "best" obviously implies that a value judgment is being made. — Gurr also discusses the major role of the struc ture of the political system in this general area: Ted Gurr, "The Genesis of Political Violence: A Theoretical Analysis," chap. ix. 13 See, for example, Simmel's discussion of the func tions of opposition actors. Georg Simmel, Conflict and the Web of Group-Affiliations. trans. by Kurt Wolff and Reinhard Bendix (New York: The Free Press, 1964), p. 19. 52 This judgment, however, stems directly from this paper1s definition of the social role or purpose of government: to move a conflict situation to a termination point in a pro cedural manner. Avoidance is discussed as another means of resolving a conflict situation, or of altering the nature of the situation. In this paper, however, I argue that avoidance seldom resolves a conflict, and in fact, can become a precondition for violence. Mack and Snyder have even suggested that in some cases the use of violence can be conducive to moving a situation to a terminal point— i.e., to a point at which one or all actors leave the conflict situation, whether the issue on which the conflict was based has been resolved or not. This, however, is still an unsettled question, and there is sufficient concern over violence as a social problem to indicate that violent be havior within nations is now considered usually the least 14 desirable activity. At any rate, given the above defini tion of the major function of government in society, we must conclude that violence would be the least desirable 14 Raymond Mack and Richard Snyder, "The Analysis of Social Conflict— Toward an Overview and Synthesis, 1 1 Journal of Conflict Resolution. I (June, 1957), 240; Boulding suggests that violence frequently perpetuates conflicts, see Boulding, Conflict and Defense, pp. 323-324. 53 method of handling a conflict situation. A conflict situation requires at least two actors and can originate anywhere in society. Kenneth Boulding has pointed out that an actual conflict situation can exist only when the actors see value in remaining in the situation when, for example, each actor feels that it pro vides the best possible structure to maximize his individ- 15 ual gains. Actors within nations may be individuals, groups, organizations, governments, or intervening actors from the international system. Conflict situations frequently involve two or more actors, and in most of these situations behavior is controlled by the structures of society. While it is generally recognized that the sources of social conflicts can seldom be eliminated, a particular conflict situation can usually be resolved, at least temporarily. It is assumed that a conflict situation then usually moves toward a termination point. ^ The role of the government, as previously defined, should be to provide the procedural structures which regulate behavior and hopefully move the 15 Boulding, Conflict and Defense, pp. 7, 10. 16 For a discussion of this position see Lewis Coser, "The Termination of Conflict," Journal of Conflict 54 conflict situation to a termination point. The following section discusses the related functions of procedural modes of conflict regulation, as they come together to structure a political process. Procedural The procedural channels for conflict regulation generally center on three basic ideas: socialization, compromise and award. In general these are the available structures that are used as alternatives to violence. In order to gain complicance with its procedural structures, a government must seek to develop, through the process of socialization, common meanings and cues in society. Ideally, all actors in conflict situations will know what makes for political influence, and each actor will be able to assess his relative influence (i.e., will be knowledgeable in regard to the dominant-subordinant 17 hierarchy). There should be some mutually agreed upon limits on conflict, ensuring some advantage to all actors in the situation, and thus inducing them to follow the established political procedures. For example, total Resolution, V (1951), 347-353. 17Ibid., p. 352. 55 victory and elimination of one1s opponent should not be considered as a termination point for the conflict situa tion. There must be some incentive for all actors to remain in the procedural channels. Ralf Dahrendorf discusses three basic conditions that render conflicts susceptible to effective procedural regulation.18 When these conditions are present, then actors will tend to regulate their conflicts within the established procedures. They may be summarized as follows: 1. A community of interests encompassing the conflict situation; recognition of the oppon ent's position as legitimate. 2. Involvement of identifiable organized interest groups and opposition groups, as opposed to diffuse, unfamiliar aggregates of actors. 3. Mutually accepted meanings for the formal structures and rules that form the control capsule. Whenever these basic conditions are not adequately fulfilled in conflict situations, there is some likelihood that the procedural forms of regulation will be ineffective and that violence will occur. Several authors have attempted to isolate and 18 Ralf Dahrendorf, Class and Class Conflict in In dustrial Society (Stanford: Stanford University Press, 1959), pp. 225-227. define types of procedures that generally constitute the procedural channels established and/or supported by governments and regime-oriented actors. The following review of sources concerns the types of procedures included in the illustrated frame of reference in Fig. 1 below. Quincy Wright has described several basic forms of conflict regulation. These forms may all be said to require, to some extent, the above three situational con ditions. Three of Wright's forms are: (1) negotiation; (2) adjudication and decision (under the direction of laws) 19 and (3) dictation of a decision in favor of one actor. Mack and Snyder suggest that there are numerous ways of controlling conflict situations, a few of which are (1) arbitration; (2) mediation; (3) conciliation; (4) ne gotiation; (5) inquiry; (6) legislation; (7) judicial 20 settlement; and (8) socialization. Dahrendorf also recognizes that there are many ways in which conflicts can be controlled, but suggests that three basic types can be identified. He cites 19 Quincy Wright, "The Nature of Conflict," Western Political Quarterly. IV (June, 1951), 207. 20 Raymond Mack and Richard Snyder, "The Analysis of Social Conflict— Toward an Overview and Synthesis," p. 238. 57 (1) conciliation; (2) mediation; and (3) arbitration; all as ways of patterning behavior in order to decrease the 21 likelihood that behavior will become violent. Kenneth Boulding also isolates three types of procedural conflict regulation, which he defines as 22 follows: 1. reconciliation, where the value systems and meanings (images) of the actors so change that they develop common preferences and want a sim ilar state of affairs— the conflict is eliminated; 2. compromise, where each actor has different val ues and meanings, but each actor is willing to settle for something less than his ideal position in order to resolve the conflict situation; 3. award. where a settlement is reached because both parties have agreed to accept the verdict of another actor rather than to continue the conflict situation. The three forms of procedural conflict regulation used in the present study's frame of reference— socializa tion, compromise, and award— are in basic agreement with the observations of the above authors, particularly with Boulding's and incorporate the common elements in their efforts. 21 Dahrendorf, Class and Class Conflict in Indus trial Society, p. 230. 22 Boulding, Conflict and Defense, p. 310. 58 While in practice the measures employed to implement the three types of regulation often overlap, one can identify general procedural modes that characterize the different forms of conflict regulation: for example, socialization results from conversation, argument, debate, or any learning type situation; compromise results from bargaining games in which mediation is an important 23 factor; and award results from arbitration or legal trial. A conflict approach to the study of social phe nomena generally relates to fixed sources of contradictions and conflicts which can be regulated but seldom resolved. Socialization, however, may in the long run supply means 24 of resolving major conflicts. The socialization process is essential for establishing the common values and mean ings which are preconditions for the regulation of conflicts in a procedural manner. The sharing of common interpreta tions of given types of behavior presupposes a relatively stable socialization process. Any rapid change or any 23 For a further discussion of these points see ibid. 24 Arnold Rose, "Theory for the Study of Social Problems," p. 190. The other aspect of socialization is that it is a continual process of conflict: for example, see Boulding, Conflict and Defense, p. 167. 59 defect in the socialization process can, however, reduce the effectiveness of the established communication struc tures. For example, Marion Levy has pointed out that with very rapid change taking place in less than a generation we have a situation where the next generation is always 25 being socialized for an unknown future. Thus rapid change is now resulting in a breakdown or gap in the com munication system, which plays a major role in transmitting, over time, common values and interpretations. The socialization process, defined as an effort to reconcile conflicting values and meanings, is a process 26 involving value formation and change. As such it repre sents an ideal route to the eventual elimination of a particular source of conflict; but also, and perhaps more frequently, it plays a major though uncompleted role in compromise procedures, specifically by lowering the 27 intensity of the conflict. 25 Marion Levy, Jr., Modernization and the Structure of Societies: A Setting for International Affairs. Vol. I (Princeton: Princeton University Press, 1966), pp. 79-82. 26 See, for example Boulding, Conflict and Defense. p. 313. 27 See Sxmmel, Conflict and the Web of Group Affiliations, p. 117. 60 Simmel refers to compromise as one of mankind's 2 8 greatest inventions. Of all the procedural forms of conflict regulation, compromise is by far the most complex, but also in general the most desirable, since it may eliminate the necessity of total victory or total defeat. This is because compromise generally reveals a large base of commonly perceived meanings among the actors in a given situation, and usually presupposes a climate of ration- 1 - 4. 29 ality. In many cases where the government is involved in a conflict situation, it acts as mediator: that is, the government is there to see that the other actors play by the rules. Another key function of the mediator is to see that all actors correctly perceive and evaluate the structural and behavioral cues in the situation, to ensure that the actors will finally, as Coser has put it, "stop 28 Ibid., pp. 114-115. For another example, see Dahrendorf, who also places a high value on institutional ized compromise and bargaining. Dahrendorf, Class and Class Conflict in Industrial Society, pp. 228, 308. 29 For illustrations of the complexity involved in compromise and bargaining games, see the gaming sections in The Journal of Conflict Resolution, from Vol. IX, No. 1 (March, 1965) to present. 30 chasing the mirage of victory. ..." This is when socialization becomes increasingly relevant in making compromise possible. Once the prospect of total victory is abandoned, and given commonly perceived values, actors can pursue different goals in such a way as to facilitate a degree of goal-attainment for all involved, assuming the available structures for bargaining and compromise are effective."^ Another important phase involved in compromise and bargaining is competition. Here again, however, the structure of conflict regulation limits the means that can 32 be used in competitive situations. In the definition of violence in Chapter I it was argued that certain limits on aggression and force are implicit in any procedural approach to conflict. We also assume total victory is not 33 to be sought in any competitive situation. 30 Coser, "The Termination of Conflict," p. 352. 31 For a comparative analysis of moderate politics and conflict management, see Firestone, "Political Conflict A Theory and Comparative Analysis in Five Cultures," p. 66. 32 Simmel, Conflict and the Web of Group-Affilia tions. p. 78. 33 For a more indepth discussion of competitive struggles related to this context, see ibid., pp. 57-58. 62 When compromise is not reached and mediation is unsuccessful, then the third form of procedural conflict regulation may be useful. This is award, which, for the most part, assumes the form of arbitration and court 34 proceedings. In this context the term arbitration has a specific meaning. Dahrendorf has stated clearly the basic distinction between political or judicial arbitration and the more competitive types of arbitration used in labor disputes. The political and judicial forms qualify as award procedures in the present frame of reference. They presuppose an independent arbitrator's judgment of the merits of conflicting positions in terms of relatively 35 fixed canons, leading to a compulsory settlement involving the suppression of the conflict situation or the suppression of one of the actors in the situation. This sort of compulsory settlement can, on the one hand, prevent the selection of violence as a means, but.on the other hand it requires a high coercive potential on the part of the government actor and a willingness to use it. Repression however, of all procedural means, is considered least 34 See Boulding, Conflict and Defense, pp. 318-319. 35 Dahrendorf, Class and Class Conflict in Indus trial Society, p. 229. 63 likely to prevent violence, especially over the long term given the possibility of cumulative frustration and other types of preconditions for violence.^ Avoidance Avoidance of a conflict situation generally resolves nothing. If one or both actors leave the situa tion, it is probable that one of several other situations will arise If one or both parties leave because changes in their internal or external conditions render the issues unimportant, then the situation may in effect be resolved. If, on the other hand, one or both parties are forced to leave the situation or one actor defeats the other, then there is the possibility of a frustration build-up, which 37 may lead to either aggressive or apathetic behavior. Apathy could figure as a precondition for violence, but it is primarily the aggressive response that concerns this 38 study. It is assumed here that a high degree of 36 For example, see ibid., pp. 224-230; Gurr, "The Genesis of Political Violence: A Theoretical Analysis," chap. ix; Eckstein, "On the Etiology of Internal War," pp. 153-155. 37 For a discussion of this general tendency see Boulding, Conflict and Defense, pp. 91-92. 38 Both Gurr and Lupsha have reviewed the ways in 64 psychological or social frustration is one of the precon ditions for violence, and that it can result from invol untary avoidance of a conflict situation. Robert Merton has suggested several possible routes of avoidance— such as innovation, ritualism, re- 39 treatism, and rebellion. Merton's structural explana tion clearly indicates that avoidance frequently establishes a precondition for violence. Innovation and rebellion are by far the greatest threats to the established conflict regulating structures, while ritualism and retreatism are associated, to a degree, with the apathetic type of response mentioned above, and in the short run will have little effect on the government or its regulating struc tures . Preconditions for Violence Basically, I consider violence a common phenom enon that ruptures man's societal web. This rupture can which aggression is a response to frustration. Ted Gurr, "Psychological Factors in Civil Violence," World Politics. XX (January, 1968), 247-248; and Lupsha, "On Theories of Urban Violence," p. 12. 39 Robert Merton, Social Theory and Social Structure (New York: The Free Press, 1957), pp. 176-192. 65 take place at the most unexpected times and in the least expected places. Knowledge of the existence of precondi tions for violence can provide expectations as to the place of occurrence, but it is the trigger event that, for the most part, determines the precise time. Before listing the major types of preconditions suggested in the literature, I will clarify the distinction between preconditions and triggers of violence. A trigger is usually an event that suggests to potential deviants that they have a chance for attainment of goals if they resort to violence, or an event that increases anger or 40 frustration to an unbearable level. Preconditions for violence are situations or circumstances that make it possible for the trigger event to set off violence. With out the preconditions a trigger-type event would not spark any violence, though violence might appear within the boundaries of that particular event. Harry Eckstein has put his finger on the real theoretical necessity of making an analytical separation 40 For example, Boulding argues that frustration can result in violence only when there exists at least some faint hope of success. Boulding, Conflict and Defense. p. 92. Johnson suggests that a trigger can be any event that has discredited the authorities. Johnson, Revolution ary Change, pp. 98-99. 66 between trigger events and preconditions. He cites the need "to shift attention from aspects of internal war which defy analysis, to those which are amenable to sys tematic inquiry. Phenomema which precipitate internal war 41 are almost always unique and ephemeral in character." This study seeks to investigate further some of the preconditions for violence, and thus improve our basis for anticipating the likelihood that violence will break out if a trigger event takes place. Most of the precondi tions for violence fit easily into one of the general types of hypotheses generated by Eckstein. I have found his list useful and difficult to amend, and thus have 42 reproduced it below: Hypotheses Emphasizing "Intellectual" Factors: 1. Internal wars result from the failure of a regime to perform adequately the function of political socialization. Lupsha suggests, however, that there is not always ji precipitating event, but rather a series of es calating incidents, which are more susceptible to analysis: Lupsha, "On Theories of Urban Violence," p. 15. Given the macro-level of analysis used in this paper, I will use the Eckstein argument. Eckstein, "On the Etiology of Internal Wars," pp. 140-141. 42 Eckstein, "On the Etiology of Internal Wars," pp. 143-144. What Eckstein terms "internal wars" include what I have called violent events; see Chapter III. 2. Internal wars are due to the coexistence in a society of conflicting social "myths." 3. Internal wars result from the existence in a society of unrealizable values or corrosive social philosophies. 4. Internal wars are caused by the alienation (desertion, transfer of allegiance) of the intellectuals. Hypotheses Emphasizing Economic Factors: 1. Internal wars are generated by growing poverty. 2. Internal wars result from rapid economic progres 3. Internal wars are due to severe imbalances be tween the production and distribution of goods. 4. Internal wars are caused by a combination of long-term economic improvement and short-term setbacks. Hypotheses Emphasizing Aspects of Social Structure: 1. Internal wars are due to the inadequate circu lation of elites (that is, inadequate recruit ment into the elite of the able and powerful members of the non-elite). 2. Internal wars result from too much recruitment of members of the non-elite into the elite, breaking down the internal cohesion of the elite 3. Internal war is a reflection of anomie result ing from great social mobility. 4. Internal war is a reflection arising from little general social mobility— from general social stagnation. 5. Internal wars result from the appearance in societies of new social classes. 68 Hypotheses Emphasizing Political Factors: 1. Internal wars are due to the estrangement of rulers from the societies they rule. 2. Internal war is simply a response to bad govern ment (government which performs inadequately the function of goal-attainment). 3. Internal wars are due, not to the attacks of the governed on those who govern, but to divisions among the governing classes. 4. Internal wars are responses to oppressive government. 5. Internal wars are due to excessive toleration of alienated groups. Hypotheses Emphasizing No Particular Aspects of So cieties, But General Characteristic of Social Process: 1. Political violence is generated by rapid social change. 2. Political violence results from erratic and/or uneven rates of social change, whether rapid or not. 3. Internal war occurs when a state is somehow "out of adjustment" to society. The general type of precondition investigated in this study falls into the last category. The remainder of the list suggests the vast area of possible research defined away in this study, an area which must enter into the picture at some point or another before we can offer comprehensive explanations of the preconditions for violence. 69 Figure 1, a composite of generalizations and findings of authors cited above, provides an overview of the structure of conflict regulation within nations. Figure 1 suggests, among other things, what a larger project dealing with conflict regulation might have to contend with, and most importantly, it illustrates the 43 role of violence in politics. Figure 1 is set up as a functional system in order to suggest that conflict regu lation is a continual process involving feedback, and that the regulation or termination of one situation may generate another. The flows in Figure 1, as indicated by the arrows, generally suggest the possible relationships between the three types of conflict regulation (procedural, avoidance and violence) and the conflict situations. For example, a conflict situation may be directly amenable to either the procedural or avoidance methods of regulation, but the use of violence presupposes preconditions and trigger events. Also, the suppression type of award solution may produce the same effects as the avoidance method, which as indicated by the arrow frequently establishes or 43 For a definition of relevant violent behavior, see Chapter I. CONFLICT PROCEDURAL SITUATION Individual Socialization Value formation Group Value changing Organization Compromise Government With government as a party to the conflict International system With government as a mediator Award PRECONDITIONS OF VIOLENCE Constitutional settlement Legal settlement Rapid change Deprivation Suppression of conflict Monopoly of violence Frustration Etc. AVOIDANCE One actor leaves the conflict Both actors leave the conflict Trigger events Conquest by one actor ■> VIOLENCE Unacceptable behavior Opposed behavior Fig. 1.—The structure of conflict regulation Government policy prevails and/or is reinforced Government and opposition policies are reconciled Opposition policy prevails No influence on government policy Government creates change 71 reinforces preconditions for violence. The three forms of conflict regulation can influence subsequent government policy in a variety of ways. I have not indicated the probable manner in which one method of conflict regulation relates to one or another of the types of governmental responses, because to do so would open a whole area of problems and questions that are systematically related to, but not neccessary for the understanding of, the central problem under investigation here. Therefore, I have not attempted to link the methods of regulation to particular governmental responses with arrows. In terms of the frame of reference, the problem here is to find to what extent rapid rates of social change over-burden the procedural channels of conflict regulation, thus resulting in preconditions for violence. We must first arrive at precise empirical definitions of violence and social change, which I propose to do in the next two chapters. CHAPTER III VIOLENCE WITHIN NATIONS: DIMENSIONS AND DATA The universe of all possible conflict events is indeed vast— even when limited to those events which occur within the sample of 234 cases used in this study. My purpose in this chapter is to isolate the major units of analysis, and then to survey several studies which have analyzed events from the broader universe of behavior, and to isolate those conflict events that fall into the selected cases. I also discuss the basis for reducing the several types of violent events to three general dimen sions (turmoil, conspiracy and internal war). Then, the basis and method for determining the magnitude of violence within each case is discussed. The Universe of Cases The primary purpose of this section is to select comparable units of analysis. It is assumed that 72 comparability is increased when all units of analysis have similar sociocultural systems, and that to the extent they do, they will attach shared meanings to any given behavior. As explained in Chapter I, since violence is defined here as a form of unacceptable behavior, the occurrence or preconditions of violence will be directly tracable to attitudes toward certain types of behavior within a given sociocultural context. Thus we must select similar sociocultural units, units that assign similar meanings to any given behavior. The validity question is whether events we commonly consider indicators of violence— riots, coups, strikes, etc.— have the same meaning throughout a group of systems. If they do not, if they represent acceptable behavior in one system and unacceptable behavior in another, then we obviously cannot use our record of these events or be haviors as a basis for ascertaining whether violence exists or to measure degrees of violence. In comparing systems I apply two basic criteria. Arthur Kalleberg discussed the significance of these criteria, and summarized as follows: "In short, two objects being compared must be of the same class— they must either have an attribute or not. If they have it, 74 and only if they have it, may they fsicl be compared as to which has it more and which has it less."'*' The first stage in the comparative method is to classify, on an "either/or" basis: in the present study, cases without violence are dropped, and the cases in which violence occurred are retained. The second requirement is to draw the "more or less" comparison: since all cases exhibit some violence, and they may be measured as to how much violence has taken place in each case. After measurement, comparison takes place. The sociocultural system determines the particular meanings that prevail in a given system, and as explained in Chapter I, the meanings assigned to different kinds of behavior are essential in ascertaining when violence exists. For example, a coup d'etat in an African nation will have different causes and will indicate different kinds and degrees of violence than a coup d'etat in the United 2 States. A certain type of event might be classified as ■*"For a discussion of the basic assumptions under lying the comparison of these cases see Arthur Kalleberg, "The Logic of Comparison: A Methodological Note on the Comparative Study of Political System, " World Politics. XIX (October, 1966), 76, 81. 2 See for example, the discussion of the meaning of a coup in some African nations in Aristide Zolberg, "The 75 violent in one cultural context and not in another. Thus, when seeking to identify violence and the variables which indicate its preconditions, it is necessary to validate variables within the value system in which the behavior or condition exists."^ In terms of the above discussion, the best way to deal with the validity problem is to confine a study as far as possible to units of analysis, to be defined as nation-states, that have similar cultures. This can best be done by dividing the world into sociocultural regions, and by selecting all units to be compared from the same region. The idea of working exclusively within one region at a time is not new, but this approach has seldom been taken when aggregate data are used. Several studies have, however, implied that cross-national comparison is more useful and appropriate when all units are selected from the same cultural region. For example, Gurr1s observation Structure of Political Conflict in the New States of Tropical Africa, 1 1 American Political Science Review. LXII (March, 1968), 70-87. 3 For a more detailed discussion of this problem see Henry Teune, "Intra-System Comparisons in Cross-Systems Analysis" (paper presented at the 64th Annual Meeting of the American Political Science Association, Washington, D.C., September 3-7, 1968), p. 17. 76 of the commonality of Asian states led him to suggest that 4 the Arab states constitute a subcluster of nations. Tanter and Midlarsky also suggest this approach on empiri cal grounds: When the total number of cases is considered, domestic violence in Asian countries in general, and Middle Eastern countries in particular, may have an entirely different political meaning from the same degree of violence in Latin Ameri can countries. This division . . . is a first approximation to a control for cultural differ ences .J This increasing preference for regional over universalistic approaches has largely been stimulated by the need for more valid cross-national variables. Michael Haas has suggested that most authors who deal with aggre gate data at the universalistic level are aware of the validity problem, and anticipate the possibility that some 6 of their findings could be misleading. This study also assumes that some of the findings of universalistic studies 4 Ted Gurr with Charles Ruttenburg, The Conditions of Civil Violence: First Tests of a Causal Model. Research Monograph No. 28 (Princeton: The Center of International Studies, Princeton University, 1967), p. 98. 5 Raymond Tanter and Manus Midlarsky, "A Theory of Revolution," Journal of Conflict Resolution. XI (September, 1967), 273-274. 6 Michael Haas, "Aggreqate Analysis," World Politics, XIX (October, 1966), 117. may be based on the use of variables that have little or no cross-national validity, and that the best means of testing the validity of these variables is to replicate the studies using only nations from a common sociocultural region. For example, certain studies relating domestic and foreign conflict have found no significant relationship between the two types of conflict at the universalistic level. A more recent regional approach to the same ques tion, however, found "that domestic disorder is a sufficient 7 condition of foreign conflict behavior m Africa." The variables used in these studies seem to be valid for African nations, but not for a universal sample. The above mentioned suggestions and findings are considered an ade quate justification for testing the central hypothesis of this study in nations from a common sociocultural region. The Space and Time Dimensions I will now discuss means of selecting certain units of analysis from a definable sociocultural region. For an example of a universalistic approach see Rudolph Rummel, "The Dimensionality of Nations Project: A Five Year Research Program" (mimeographed); for a regional approach see John Collins, "Foreign Conflict Behavior and Domestic Disorder in Africa" (unpublished Ph.D. disserta tion, Northwestern University, 1967). 78 Each unit must be subject to the independent and dependent variable and may be an individual, a group, an organization, or a nation. The nation is the unit used in this study. In Chapter II it was stated that in this dissertation individuals, groups and organizations are the actors whose behavior is being analyzed. The behavior of these actors is considered in its national context. The nation, then, as a unit of analysis, provides a basis for categorizing events. Thus, the social context of all events is coinci dent with the nation-state. As Galtung has termed it, 0 the nation is primarily a unit in physical space. I deal with thirteen such space units, and I also include the dimension of time. The period of time over which each nation is considered is divided into equal time segments. The selec tion of units of time is to a degree arbitrary. The obvious possibilities are days, weeks, months, or years. The use of daily time segments would permit categorization of the data in a highly sensitive manner, and would permit meaningful time-lag analyses between the violence and rate of change variables. But since the data used to measure 8 Johan Galtung, Theory and Methods of Social Re- search (New York: Columbia University Press, 1967), pp. 24-26. 79 rate of data change are reported only yearly, in the major sources used time segments any shorter than one year are precluded. The violence data are therefore collapsed into yearly measures and correlated with the rate of change measures for each nation. Comparison across units of time is less of a problem than comparison across units of space, unless the units of time cover extremely long periods, or are accom panied by very rapid rates of social change. The eighteen- year time span adopted in this study is not considered excessive, and the rates of change in the nations included in this study, as measured by the independent variables, do not present a problem. In other words, the cases obtained through longitudinal analysis are considered highly comparable. The nation, then, is considered a "spacetime chunk."^ Within this frame of reference, the universe of all possible cases would equal the total number of nations multiplied by the number of years each nation has been in existence— or, more realistically, the total number of years in the history of each nation for which data are 9 See Galtung1s discussion of the logic underlying the selection of such units of analysis, ibid. 80 available. The sample of the universe used in this study is made up of thirteen space units, each of which is considered over eighteen one-year time periods. Each space unit is analyzed for eighteen time periods, each of which constitutes a separate case in which the dependent variables are correlated with the independent variables. Thus the total number of possible cases in this study is 234. The 234 cases are reduced on the basis of whether violence occurred in each case. These cases are classified as containing violence or not, which satisfies the first requirement (either/or categorization) for a comparative approach. Following classification on this basis, each remaining case can be measured in terms of magnitude of violence, and finally in terms of the degree to which rapid social change as a precondition is related to high magni tudes of violence. The Sociocultural Region The empirical basis for selecting a cultural region of nations in this study was drawn from Bruce Russett's "Regions of Socio-Cultural Homogeneity,"^ a 10 Bruce Russett, "Delineating International 81 delineation of regions based on factor analyzed data on 29 variables for 82 nations. These variables were selected from a larger group of 54 variables, as the most highly intercorrelated at a universalistic level, and for presumed relevance to sociocultural structures. The data used to measure these variables were those used in the World Hand- 12 book. These data generally represent the attained Regions," in Quantitative International Politics; Insights and Evidence, ed. by J. David Singer (New York: The Free Press, 1968), pp. 321-338. ■^The 29 variables used by Russett to determine regionalness are: GNP per Capita; Newspapers per 1000 Pop ulation; Nonagricultural Workers as Percentage of Wage and Salary Earners; Life Expectancy; Percentage of Labor Force in Agriculture; Radios per 1000 Population; Domestic Mail per Capita; Percentage of GNP from Agriculture; Inhabitants per Hospital Bed; Percentage of Labor Force in Industry; Primary and Secondary School Pupils as Percentage of Popu lation; Percentage of Adults Literate; Births per 1000 Population; Wage and Salary Earners as Percentage of Workers; Percentage of Population in Cities; Television sets per 1000 Population; Foreign Mail per Captia; Cinema Attend ance per Capita; Inhabitants per Physician; Investment as Percentage of GNP; Communist Votes as Percentage of all Votes; Central Government Revenue as Percentage of GNP; Non-communist Secular Votes as Percentage of all Votes; GNP per Capita, Percentage of Annual Increase; Roman Catholics as Percentage of Population; All Christians as Percentage of Population; Population Density; and Population per Hectare Agricultural Land. For the factor loadings on these variables see ibid.. pp. 323-326. 12 Bruce Russett, Hayward Alker,, Karl Deutsch, and Harold Lasswell, World Handbook of Political and Social Indicators (New Haven: Yale University Press, 1964), pp. 13-258. values on each variable for each nation. Russett's 29 variables are similar to or include most of the variables used in this study to measure rate of change. The regions based on the 29 variables, however, are not necessarily homogeneous in terms of the other variables. Uncertainty as to cross-national validity is due to the fact that variables alter when measured for rate of change over time, rather than attained level at one point in time. Thus, even though Russett has found attained scores on these variables to be useful in outlining sociocultural regions, we can no more than assume that rate of change measures on these same variables will retain common meanings for all nations in the region. The thirteen nations selected here all fall into Russett's Afro-Asian group, and are further limited to the Middle East and North African states. Isreal is the only selected unit outside his region, and is included primarily on the basis of geography. Of the twelve nations in the region, Lebanon is least like the others, but nevertheless resembles the nations of this region more than it does other nations. This is, then, assumed to be a cross national study, not a cross-cultural analysis— save for the case of Isreal. With the exception of the latter, all 83 nations are from a common culture generally identified as the Muslim culture. To note how the selected nations generally group relative to nations from all regions of the world, see the World Handbook on the 29 variables. On the "Newspaper Circulation per 1000 Population" variable, for example, the nations cluster, with the exceptions that Israel and the Lebanon are much higher, and the Sudan and Libya much lower, than the majority of the nations in the region. The attained levels on "Items of Domestic Mail per Capita" and "School Pupils as a Percentage of Population Aged 5-19" are similar for all the nations except Israel. Certain vari ables, however, such as "Percentage of GNP Originating in Agriculture," show no apparent grouping. Measures on the cultural variables clearly suggest the regionality of the sample of nations, again with the exception of Israel— for example, the "All Christians as a Percentage of Total 13 Population" variable. The cross-national and longitudinal perimeters 13 Russett, et al.. World Handbook of Political and Social Indicators; see the relative measures of the vari ables discussed above on the following pages, respectively: pp. 108-110, 112-113, 219-220, 173-174, and 252-254. 84 have been established and operationally defined to include eighteen one-year time segments (1948-1965) for each of the following nations: Iraq, Syria, Lebanon, Jordan, Sudan, Algeria, Morocco, Tunisia, Libya, Egypt, Israel, Iran, and Turkey. In Chapter VI, by means of longitudinal analysis of the eighteen cases for each nation, the cross-national validity problem will be further controlled. Thus the present study is in part a further test of whether the thirteen nations do in fact constitute a homogeneous region. Selection of Variables The basic data for the dependent variables are drawn from measurement of the scope, duration and intensity of violent conflict in each case. Several studies have been involved in data collection with regard to conflict events, and constitute the universe from which my data are selected. These studies have also provided a basis for reducing the many types of conflict events to a managable number of dimensions. These dimensions reduce the sensi tivity of the dependent variables, but the reduction is necessary in order to ensure an adequate frequency of events in each case. The empirical justification for compiling events into dimensions of violent behavior is that when certain types of events correlate positively over a reasonable number of cases, then it is possible to speak of a cluster 14 of events or dimension of behavior. The statistical technique used to arrive at these clusters, or dimensions, is factor analysis. The several studies selected for presentation in Table 1 were factor analyzed in the works discussed below. The dimensions derived in each of the studies are compared with other proposed dimensions in order to determine whether those arrived at by Rudolph Rummel are reliable for data collections over longer periods of time. The studies listed in Table 1 have collected data on similar types of events, over a large number of nations, and from similar types of sources. Harry Eckstein1s data were collected for a fourteen year period (1946-1959) for 113 nations, and were drawn 14 See Rudolph Rummel, "Dimensions of Conflict Behavior Within Nations, 1946-59," Journal of Conflict Resolution. X (March, 1966), 65. For a more detailed dis cussion of the method, see Rudolph Rummel, "Understanding Factor Analysis," Journal of Conflict Resolution. XI (De cember, 1967), 444-479. TABLE 1 STUDIES OF CONFLICT EVENTS Types of Events Dimensions Years Covered Major Sources Number of Nations Eckstein Internal war, turmoil, riots, large-scale terrorism, small-scale terrorism, mutiny, coups, plots, administrative actions, quasi-private violence Turmoil Revolution Subversion 1946- 1959 New York Times Index 113 Rummel Assassinations, general strikes, guerrilla war, major government crises, purges, riots, revolutions, anti-government demonstrations, people killed Turmoil Revolution Subversion 1955- 1957 New York Times Index New International Yearbook Facts on File 77 Tanter Same events used by Rummel Turmoil Internal War 1958- 1960 New York Times Index Deadline Data on World Affairs 83 CO CD TABLE 1— Continued Types of Events Dimensions Years Covered Major Sources Number of Nations Feierabend Repressive action against specific groups, micro strikes, general strikes, micro or macro demonstrations, micro- or macro riots, arrests of significant persons, imprisonment of significant persons, arrests of few or many insignificant persons, imprisonment of insignificant persons, assassinations, martial law, execution of significant or insignificant persons, terrorism and sabotage, guerrilla or civil war, coups, revolts, exile Turmoil Internal War 1948- 1962 Deadline Data on World Affairs 84 Gurr Demonstrations, political strikes, riots, localized rebellions, banditry, plots, purges, assassinations, bombings, small-scale terrorism, small-scale guerrilla war, coups, mutinies,, large-scale terrorism, large-scale guerrilla war, civil war, private war, large-scale revolt Turmoil Conspiracy Internal War 1961- 1965 New York Times Index Newsvear African Digest 114 88 15 exclusively from the New York Times Index. Rudolph Rummel factor-analyzed Eckstein's data for comparison with 16 his own. Rummel found that his 1955-1957 data also reflected conditions during the longer period covered by the Eckstein data, and that his data supported the same 17 three dimensions. These three dimensions are composed of the following types of events, using Rummel's terms: turmoil (a non-organizational, spontaneous type behavior); anti-government demonstrations, riots, major government crises, assassinations, and general strikes. revolutionary (a covert, organized type behavior); revolutions, purges, general strikes. subversion (a covert, organized type behavior); guerrilla war, assassination. Eckstein1s types of events collapse into Rummel1s in more- or-less the following manner: His data are in Harry Eckstein, "Internal War: The Problem of Anticipation" (report submitted to the Re search Group in Psychology and the Social Sciences, Smith sonian Institution, Washington, D..C., January 15, 1962), Appendix I. 16 The Rummel data are available through the Inter- University Consortium for Political Research, Ann Arbor, Michigan; and are analyzed in the various studies by Rummel listed in the Bibliography of this study. 17 Rudolph Rummel, "Dimensions of Conflict Within and Between Nations," General Systems Yearbook. VIII (1963), 11-13. 89 guerrilla war; internal war, large-scale terrorism. assassinations; small-scale terrorism. riots7 riots, turmoil. revolution; mutinies, coups, plots. purges; administrative actions.^ Eckstein's quasi-private type event has no equivalent in the Rummel system. Raymond Tanter supplemented the 1955-57 data by using Rummel1s types of events and collecting data for the 19 1958-60 period. He then correlated the combined data collections (1955-60) with Eckstein's (1946-59) and 20 Feierabends' (1948-62) data. Tanter's factor analysis suggested there were only two major dimensions to conflict behavior— turmoil and internal war. His internal war dimension is based on the presence of organized violence, 18 Rummel, "Dimensions of Conflict Behavior Within Nations," p. 69. 19 See, for example, Raymond Tanter, "Dimensions of Conflict Behavior Within and Between Nations, 1958-60," Journal of Conflict Resolution. X (March, 1966), 41-44. His data are in Raymond Tanter, "Dimensions of Conflict Behavior Within and Between Nations, 1958-60" (Northwestern Univer sity, August, 1964), Appendix II (mimeographed). 20 Raymond Tanter, "Dimensions of Conflict Behavior Within Nations, 1955-60: Turmoil and Internal War" (paper presented at the Peace Research Conference, University of Chicago, November 16-17, 1964), pp. 14-15. 90 and reflects Rummel1s revolutionary and subversive dimen sions. The Feierabend data were found to have dimensions similar to the 1955-60 data. These factor analyses by Rummel and Tanter also provide reliability tests for these sets of data. In terms of intra-coder and inter-coder reliability the data were found to be highly reliable. They also determined that systematic error, as measured by press censorship and world interest in certain nations, did not distort the dimensions 21 of conflict behavior for the 1955-1960 and 1948-1962 data. The existence of the turmoil dimension is clear, while the dimensions that reflect organized violence remain uncertain. In the present study, however, Rummel's revolu tion and subversion dimensions will be used to insure a minimum loss of sensitivity, but they will be redesignated to correspond with those used by Ted Gurr in his data collection. Rummel's revolution dimension is comparable to Gurr's conspiracy category, and his subversion dimension becomes Gurr's internal war category. Gurr retains the three dimensions on the basis that it is useful to Ibid. In a separate analysis Rummel came to a similar conclusion: Rummel, "Dimensions of Conflict Be havior Within Nations, 1946-59," p. 71. 91 distinguish between large and small scale organized vio lence. The three dimensions of violent behavior used in 22 this study are as follows: turmoil (spontaneous and unstructured violence); demonstrations, political strikes, riots, politi cal clashes, and localized rebellions; conspiracy (intensively organized but small-scale violence): political assassinations, small-scale terrorism, small-scale guerrilla wars, coups, mutinies, and plots and purges (the last two if planned); internal war (organized, focused, and large-scale): large-scale terrorism and guerrilla wars, civil wars, private wars, and large-scale revolts. Method of Measurement A magnitude of violence has been calculated for each dimension of violence for all of the cases. For ex ample, the case of 1959 in Tunisia has a score for the magnitude of turmoil, the magnitude of conspiracy, and the magnitude of internal war. In addition, there is a fourth measure for each case, a composite including the scope, duration and intensity of all events occurring within the perimeters of each case. This fourth score is the total 22 Ted Gurr, "A Causal Model of Civil Strife.: A Comparative Analysis Using New Indices," American Political Science Review. LXII (December, 1968), 1107. magnitude of violence (TMV). The summary score (TMV) is not, however, the primary dependent variable. The exclusive use of the TMV would compound the loss of sensitivity already incurred by collapsing the various types of events into the three dimensions of violence. Accordingly all four measures are used in each analysis, the TMV and the more sensitive scores for the three dimensions of violence. Use of four dependent variables— turmoil, conspiracy, internal war, and the TMV— enables us to make separate observations of organized violence, small-scale organized violence, and the magnitude of violence in general. In order to arrive at a score for each of the four variables, each event contained in each of the dimen sions of violence was broken down into three major compon ents— scope, duration, and intensity. For example, all of the events under the heading of turmoil in each case were measured on these three components to obtain a single score. I arrive at this score by following Ted Gurr's scaling procedure and method for calculating the magnitudes, with only minor adjustments. Each event was measured on 23 the following three characteristics: 93 scope: the extent of participation by the affected population, operationally defined for this study as the sum of the estimated number of participants in all events under each type of violence for each case, expressed in terms of participants per 100,000 population; duration: the persistence of strife, indexed here by the sum of the spans of time of all events under each type of violence in each case, whatever the relative scale of the events, ex pressed in days; intensity: the human cost of strife, indexed here by the total estimated dead in all events under each type of violence for each case. Measures of scope, duration and intensity for each cluster of events were compiled into single scores for each dimension of violence in each case. The distri bution of scores was highly skewed, due to the very high magnitudes of violence in several cases. The compiling procedure, however, included necessary adjustments. A log (X+l) transformation was used to make the scores more nearly normal. The logged scores on each of the three components were added, divided by eight, and the anti-log used as the magnitude of violence score for each dimension of violence in each case. The scores for the cases ^Ibid., pp. 1107-1109. 94 reflecting the years of the Algerian war were still far from normal, and turned up as outliers in the regression analysis. They were finally dropped from the regression analysis (see Chapter V). Data Sources The Eckstein, Rummel and Tanter data collections (see Table 1) record frequency of occurrence for the different types of events, but do not contain the informa tion needed to calculate scope, duration and intensity. The Feierabend and Gurr data collections, however, do provide the necessary information, and are the data sources 24 used in this paper. Both collections record the data for each event on a single IBM card, which facilitated the initial selection of events to be considered in this study. Feierabend has recently found a high correlation (r = .7) between his data and Gurr's over the same period of time, which is the basis for my using the two different 24 Feierabends1 Data Bank of Political Instability Events and coding manual are available through the Inter- University Consortium for Political Research, Ann Arbor, Michigan. The 1961-1965 data and coding instructions used in this study were made available through the generosity of Ted Gurr from his Civil Strife Data Bank at the Center of International Studies, Princeton University. 25 sources of data side-by-side. Gurr's data, however, have been coded on different scales than those used by Feierabend. It was, therefore, necessary to work backwards with the Feierabend data and "decode" each event, taking the middle of each interval on the Feierabend scales as the raw numbers for the number of participants, number of days, and number killed in each event. This information was then recoded in Gurr's scales. Both Feierabend's and Gurr's coding requirements and scales for measuring scope, duration, and intensity are reprinted in Appendix A of this study. In general, recoding simply entailed collaps ing several of the Feierabend intervals into Gurr's; when one interval fell between two of Gurr's intervals, the interval giving the higher score was taken. After the 1948-1950 events were recoded, they were integrated with the 1961-1965 Gurr data. The Feierabend data covered 176 violent events within the perimeters of my 234 cases; Gurr's study measured 152 events. Thus the combined studies yield data 25 Ivo Feierabend, Rosalind Feierabend, and Betty Nesvold, "Social Change and Political Violence: Cross- National Patterns" (report prepared for the Task Force on History and Comparative Analysis of the National Commission on the Causes and Prevention of Violence, 1969), p. 26 (mimeographed). 96 on 328 events involving violence over an eighteen-year period in thirteen nations. The next step was simply to assign each event to one of the 234 cases (by nation and into proper yearly time segments). Then within each case the events were categorized in terms of turmoil, conspiracy or internal war. The final step in each of these subcategories was to break each event down into its scope, duration and inten sity components, and then to calculate the magnitudes for each dimension of violence in each case, as outlined above. These data are ordinal, and fall short of quali fying as interval data, which are presupposed by the statistical techniques used in this study. The use of ordinal data is rationalized on the following basis. Hubert Blalock suggests that real interval data seldom exist in the social sciences; he does, however, warn that if less sophisticated data are used though the techniques used presuppose interval data, then this flexibility must be reflected in any conclusions drawn from the data. In order to gain information, other social scientists have taken this step beyond their ordinal data, using techniques that call for interval data. Under these circumstances, as Ted Gurr points out, the alternative would be to forego 97 the type of analysis desired, and since this study is largely exploratory, it is deemed best to gain as much information as possible about the variables by treating 26 the data as interval, as have the authorities cited. On at least one of the four dependent variables violence had occurred in 124 cases. Thus, with the number of cases reduced to 124, and violent behavior categorized into and measured on four different dependent variables, we may now turn to the preparation of the independent variable before moving to the analysis. 26 Hubert Blalock, Social Statistics (New York: McGraw-Hill Book Co., 1960), pp. 274 and 16-17; Gurr with Ruttenberg, The Conditions of Civil Violence: First Tests of a Causal Model, p. 16. For examples of studies using techniques that assume an interval scale, see Russett, "Delineating International Regions," pp. 317-3 52; Phillip Gregg and Arthur Banks, "Dimensions of Political Systems: Factor Analysis of a Cross-Polity Survey," American Politi cal Science Review. LIX (September, 1965), 601-614; Rudolph Rummel, "Some Empirical Findings on Nations and Their Behavior," (Research Report No. 10, University of Hawaii, 1968) (mimeographed). CHAPTER IV RAPID RATE OF CHANGE: VARIABLES AND DATA Selection of Variables The universe of all possible variables on which the rate of social change can be measured is so vast that the selection of a sample is difficult. This difficulty is compounded by the fact that most of these variables have no precise meaning in terms of what is actually being measured. With so little known about these variables and so many to choose from, it is difficult to establish cri teria for selecting the independent variables. In this chapter, nonetheless, I attempt to move a step beyond the simple arbitrary selection of variables. The above-mentioned universe of possible variables of social change contains hundreds of variables, all of which, to some degree, reflect an aspect or condition of the social structure. There is no evidence, however, to 98 suggest that there is any single variable that is the sole indicator of a particular aspect or condition of a social system. For example, there are probably dozens of variables that could be used to measure the rate of change in family structures, and to further complicate the pro cess of selecting variables, it can also be argued that a variable which indicates change in family structure can also be an indicator of economic structure. Thus, given the present knowledge of the meanings of the variables surveyed below, I assume that no one variable is the sole indicator of a social phenomenon, nor is any one variable limited to only one phenomenon. The first step in selection of variables is to identify the universe of possible variables. In estab lishing perimeters for this universe in the present study, the first decision that can be made is to limit consider ation to variables that presumably reflect conditions of the social system— specifically the condition of rapid change. The second step will be to isolate any clusters of interchangeable variables. I will first select several possible variables and then determine whether or not they are interchangeable as measures of rate of change. As pointed out in Chapter I, my main concern 100 with rapid rate of social change is based on its assumed consequences— i.e., the disruption of social cues and orientation. In general terms, Collomb and Valantin sug gest that the following variables relate to the disruption of social structure (they base their determinations on studies of Senegal) The phenomena associated with development— urbanization, industrialization, acculturation, school attendance— are going to act like a traumatic aggression on group patterns, family structure and individual internal organization. The patterns will be scrambled, and the educa tional continuum disrupted. Parcellation and disruption are the correct terms to characterize the processes leading up to a more or less pain ful individuation lived out via new conflicts. What they consider measures of internal disruption, I propose to use as measures of social change. Their vari ables of disruption— urbanization, industrialization, and communication in general— are also relevant to the central problem in this study, and will serve as indicators of change. Feierabend and others, using a universalistic approach, have found evidence to support a proposition Collomb and Valantin, "Patterns of Mothering, Organization of the Personality and Rapid Social Change," p. 443. 101 that is related to the central hypothesis in this study. They also worked with structural variables (both social and economic).2 The results obtained from interrelating the percentage rates of change on the independent indices and levels of political instability indicate, in general, that the faster the rate of change, the higher the political unrest. The combined rate of change index shows a posi tive relationship to political instability. Their dependent variables relate to the broader concept of political instability, which is not directly comparable to the concept of violence used in this study. Their selec tion of independent variables is, however, useful and it is this aspect of their study that is relevant here. For the most part, they measure change on nine variables: 1. Urbanization (percentage of the population living in communities of 100,000 or more in habitants) . 2. Literacy (literate percentage of population). 3. Primary education (ratio of total school en rollment age 14 and under to toal population. 2 Feierabend, Feierabend, and Nesvold, "Social Change and Political Violence: Cross-National Patterns," p. 48. 102 4. Post-primary education (percentage of total population enrolled in all educational insti tutions beyond the primary school). 5. National income in local currencies. 6. Cost of living index. 7. Calories per capita per day. 8. Infant mortality rate. 9. Number of radios per 1,000 population. They argue, as I will also argue below, that change on these variables indicate that change is taking place in the social system. As indicated above, Feierabend, Feierabend, and Nesvold are also concerned with change that tends to con fuse social cues. They consider the above nine variables to be interchangeable, however, and combine them into a rate of change index. Their study then does not provide us with any information about how these variables are individually related to political instability. They assume that because these variables are interchangeable when measured for attained level of development, they are also interchangeable when used as measures of rate of change. I question this assumption at the universalistic level, 103 and investigate it at the regional level (see footnote 16 in this chapter). Some authors have been investigating the possibil ity that it is change in the relationship between particular variables, rather than overall structural change, that constitutes a major precondition for violence. This approach is altogether distinct from mine, but again their use of variables that are used in this study makes a com parison useful, because their analysis rests entirely on the basic assumption that the variables are valid cross- nationally— which is precisely the assumption that I test at the regional level. For example, Tanter and Midlarsky consider two possible causes of revolution: (1) changes 3 in economic development; (2) level of education. They observe the relationships between levels of achievement and of aspirations and expectations in order to measure need satisfaction. Changes in levels of achievement and aspiration are measured in terms of the rate of change in GNP per capita. Rapid change in expectations as a pre condition of violence is primarily indicated by rapid decrease or reversal in the rate of change in GNP per 3 Tanter and Midlarsky, "A Theory of Revolution," pp. 264-280. 104 capita as aspirations continue to rise. Educational level is defined as the primary school ratio, the number of children age 14 and under enrolled in primary school 4 divided by the total population ages 5-14. Along this same line, the above-mentioned work by Feierabend, Feierabend, and Nesvold is similar to that of Tanter and Midlarsky, but they refer to want formation and 5 want satisfaction. Education, like literacy, is regarded as a means of arousing awareness of modern goals and hence of rising aspirations. It is also possible, but not equally intuitively obvious, that education may raise expectations regarding the fulfillment of these aspirations. GNP/capita, on the other hand, is a want satisfaction indicator; it provides gratification of aroused aspirations. They found that a rapid increase in primary school enroll ment is positively related to political violence; and that a rapid increase in GNP per capita is negatively related to political violence. In these two studies and some of Ted Gurr's 4 The lack of data on school enrollment made it necessary to drop this indicator in the early stages of my data collection effort. 5 Feierabend, Feierabend, and Nesvold, "Social Change and Political Violence: Cross-National Patterns," pp. 65, 58. 105 research, an effort is made to establish measures of social conditions based on these same types of variables. Gurr, for example, defines perceived relative deprivation as the discrepancy between value expectations and the 0 environment's value capabilities. Many of the measures used to determine relative deprivation are variables 7 comparable to those used m the other two studies. In one way or another, these studies deal with frustration caused by conditions in the social environment.(or changes in the economic structure that have effects in the social system). Thus their approach is in basic accord with the general hypothesis under consideration here, but presupposes greater cross-national validity for the independent vari ables than is evident from preliminary analysis of the conditions in the Middle East and North Africa. These studies are probing at the frontiers of the efforts to explain the relationship between change and violence, and, of necessity, they make many assumptions which give their work a different purpose than this study. 6 Gurr, "Psychological Factors in Civil Violence," pp. 252-253. 7 Gurr with Ruttenberg, The Conditions of Civil Violence: First Tests of a Causal Model, pp. 52-53. For example, in the Tanter and Midlarsky study it is assumed that the change measured by the GNP per capita variable does reflect a rate of change taking place in society, change that goes beyond a simple measure of the relationship between a number of people and money. In addition, the assumption is made that this variable measures achievement and aspiration levels in society. This is then followed by the assumption that achievement, aspirations and expectations (as measured by a drop in the GNP per capita variable) all come together to measure the overall degree of need satisfaction in a society. All of these studies are making inferences which are, on the one hand, certainly difficult to support, but on the other hand may free the researcher to explore the effects of certain kinds of change and the relationship between change and violence. These studies suggest interesting uses for measurements taken on various social variables, and by doing so increase the significance and the need of studies which attempt to clarify the meanings of these variables. In relation to these efforts, I am still con cerned with the first assumption made about such variables as GNP per capita. More precisely the question is, do cross-national measures of change on these variables all 107 measure the same thing in each nation? Also, I use the social variables individually as measures of rate of social change because the substantive question dealt with in this study concerns the consequences of rapid social change rather than measurement of need satisfaction. Thus, I use variables such as GNP per capita as direct measures of some social change, rather than as indirect measures of some more complex social condition. Talcott Parsons has also considered the disruptive effects of change in the structure of society. He suggests that an important aspect of social change can be measured g on some kind of technology variable. It should above all be noted that technological change inevitably disrupts the informal human relationships of the members of working groups— relationships which have been shown to be highly important to the stability and working efficiency of the participants. Technological change, however, is still too imprecise to be measured, and as I understand Parsons1 technology variable, it is not unlike the general concept of modernization— especially as it relates to the disruption of social Talcott Parsons, "Certain Primary Sources and Patterns of Aggression in the Social Structure of the Western World," Psychiatry. X (1947), 167-181. 108 structure. Marion Levy considers the general process of modernization as disruptive if it proceeds too rapidly, and he suggests a specific and useful measure for the concept of modernization. Levy suggests that the extent of use of electrical power as opposed to manual power is 9 one of the best measures of modernization. (I would also consider this a measure of technological development, if one desires a distinction between the two concepts at this general level of concern.) Levy derives the electrical power variable in the following context: A society will be considered more or less modernized to the extent that its members use inanimate sources of power and/or use tools to multiply the effects of their efforts. Along these same lines, Michael Haas considers the disruptive effects of industrialization. He sees industrialization as a source of stresses "within societal systems that start in the economic subsystem and spill over into the social system."^ Haas and Levy both use rate 9 Marion Levy, Modernization and the Structure of Societies: A Setting for International Affairs. I, 11. ■^Michael Haas, "Social Change and National Aggres siveness, 1900-1960," in Quantitative International Poli tics: Insights and Evidence, ed. by J. David Singer (New 109 of change in per capita and production of electricity as a way of measuring industrialization and modernization. I use this variable as a measure of the disruption caused by rapid rates of modernization or industrialization. Exactly what aspect of these phenomena is being measured is not certain. We do assume, however, that it indicates significant social changes that might be taking place. The structure of social communication is another general area of concern. Communication is considered, in its broadest sense, to include the mass media as well as the whole socialization process. For example, Etzioni relates the development of mass communication to the idea of developing a "mass culture."'*''*' This phenomenon is frequently accompanied by significant changes in the structure of the village and family, tracable to disrupting 12 patterns of behavior and communication. Any significant changes in rate of urbanization and population growth, it is assumed, would indicate corresponding changes in village York: The Free Press, 1968), p. 221. "^Etzioni and Etzioni, Social Change: Sources. Patterns and Consequences, p. 181. 12 Collomb and Valantin, "Patterns of Mothering, Organization of the Personality and Rapid Social Change," p. 444. 110 and family structures, and thus disrupt traditional pat terns of communication. Two other variables are also considered to be indicators of changes in communication patterns— rates of change in number of radios in use, and rates of change in newspaper circulation. Next, I would like to suggest two indirect ways of measuring any changes that might take place in the structure of the distribution of a nation's wealth. First, the obvious and most used variable is GNP per capita. For example, in the above-mentioned work by Tanter and Midlarsky, they found a positive relationship between rate of increase in GNP per capita and domestic violence in the 13 Middle Eastern and Asian areas. The GNP per capita variable is also included in the Feierabend list of vari ables (see page 101). I am certain that this variable will indicate most major changes in the production of wealth in a nation, but I am not as certain as others, nor need I be for the purpose of this study, of exactly what is being measured. The second indirect measure of changes in the structure of the distribution of a nation's wealth is the 13 Tanter and Midlarsky, "A Theory of Revolution," p. 274. number of departures from a country. These departures presumably involve several types of migration: work forces leaving the country to seek employment, national minorities leaving because of economic discrimination, and so on. The former condition could reflect rapid diminution of employ ment opportunities, and the latter would more than likely be felt as an exodus of capital and/or skills. It is assumed that structural change due to migration will have disruptive effects throughout the social system. For example, when a rapidly increasing number of people leave the country for employment, I assume that the structure of the family and community will be disrupted. I also assume that a rapid exodus of capital and skills will have dis ruptive effects on the social system,as a whole. This migration variable represents an attempt to measure those disruptive effects of rapid movements of people that might not register as rate of change on variables such as popu lation and urbanization. There have also been several studies suggesting variables of economic development. Two of these studies have simplified the selection of variables of economic development by seeking clusters of interchangeable vari ables; some of these have been discussed above, and will 112 be used in this study. Again, I assume the measured economic changes have disruptive effects on the entire social system. The Dimensionality of Nations Project, in its first phase, came up with a list of about 600 variables, 14 later narrowed to 230 and then factor analyzed. The Yale Data Program started with a more select group of var iables, which was factor analyzed by Russett."^ Both analyses derived an economic development cluster of vari ables. These clusters include: GNP per capita; news papers; radios; and urbanization— all of which have been discussed above as variables of social change in other more specific areas. In terms of level of economic development the results of the Rummel and Russett studies indicate that the variables could be used interchangeably, because they all essentially measure the same thing. I factor analyzed several of these same variables, however, on the basis of rate of change over 124 cases, and no strong loadings 14 Rummel, "The Dimensionality of Nations Project: A Five Year Research Program," p. 6. 15 Russett, "Delineating International Regions, " pp. 317-3 52. 113 16 suggested any sort of cluster. The seven variables listed below did cluster when measured on level of devel opment; they did not cluster in any significant manner, however, when measured on rate of change. Thus it is assumed each of these variables measures a different aspect of change related to economic development. On the basis of the above discussion, I have selected the following variables for the purpose of mea suring rate of social change: 1. Population (percentage of change in total population); 2. Urbanization (percentage of change in popula tion of communities of 100, 000 or more); 3. Migration (percentage of change in numbers of departures); 4. Electrical production (percentage of change in electrical production in kwh); 5. GNP per capita (percentage of change in gross Three orthogonally rotated factors appeared with loadings equal to or greater than .50: Factor I included urbanization and electrical production; Factor II included population and radios; Factor III included migration, GNP per capita and newspapers. In order to retain as much in formation as possible, it was decided not to cluster these variables. 114 national product per capita); 6. Newspapers (percentage of change in esti mated circulation of newspapers); 7. Radios (percentage of change in number of radio licences issued and estimated number of receivers in use). As mentioned above, these variables are not treated interchangeably, but considered individually and as a group in relation to change on the dependent variables. The sensitivity gained by not lumping these variables into a composite is apparent in the chapters dealing with analysis. In addition, it is felt that too little is known about these variables to permit the establishment of a composite, or an index of rate of change. By treating the independent variables as distinct measures of change, we can develop twenty-eight sub hypotheses around the central hypothesis: each of the seven independent variables is analyzed for its relation ship to each of the four dependent variables. Thus the independent variable, rate of change of GNP per capita, has four possible associations, expressed as four sub- hypotheses : 115 A rapid rate of change as measured in terms of GNP per capita is positively related to: 1. a high magnitude of turmoil; 2. a high magnitude of conspiracy; 3. a high magnitude of internal war; 4. a high TMV. Only statistically significant positive or negative rela tionships of variables will be discussed in Chapters V and VI of this study. The remaining correlations of all vari ables are presented in Appendix B. Time factors affecting the relationships of the variables will be considered in analyzing the data. Michael Haas suggests the influence of rapid social change on behavior involves a time-lag; specifically, he recommends allowing for a three-year lag in order to measure the full impact of change.'*'^ Data Sources The sources of data for the independent variables are the United Nations Statistical Yearbook. United Nations Demographic Yearbook, and the United Nations Compendium of Haas, "Social Change and National Aggressiveness, 1900-1960," p. 226. 116 Social Statistics and United Nations Yearbook of National Accounts Statistics. The first two sources were by far the most useful, and contain most of the data used in this study. The most recent yearbooks were consulted first (beginning with 1966), and the most recent datum is used whenever information differs in different volumes. For example, if the 1960 yearbook reports a higher figure for urbanization in Tunisia in 1955 than the 1956 yearbook, then it is assumed that the later figure has been adjusted and corrected. As is generally done, I have accepted the above sources as the best and most useful available, and proceeded with analysis. The United Nations yearbooks themselves offer considerable warning about the probability of high degrees of error in certain data, but for a more specific discussion of the problem see the World Handbook of Polit- 18 ical and Social Indicators. 18 Russett, et al., World Handbook of Political and Social Indicators. The following variables are discussed: population data, pp. 15-17; urbanization data, pp. 49-50; migration data, pp. 231-232; GNP per capita data, pp. 149- 151, 158-159; newspaper and radio data, pp. 105-107, 118- 119. Also, for a brief discussion of the electrical pro duction variable, see Haas, "Social Change and National Aggressiveness, 1900-1960," pp. 219-221. 117 Twenty-six percent of the data for 234 cases are missing, but for the most part these missing data are on the migration and GNP per capita variables. If these two are discounted in the aggregate analysis in Chapter V, then the amount of missing data is 11 percent, which is more reasonable. These two variables, while providing no basis for generalization at the regional level, were nevertheless retained in national profiles for which enough data were available. There were data for Morocco, Algeria and Israel on the migration variable, and for Israel and Turkey on the GNP per capita variable. When the number of cases was reduced from 234 to the 124 exhibiting violence, the overall missing data were reduced to 23 percent. If the GNP per capita and migration variables are not considered, and the number of cases is reduced to 124, then the missing data are less than ten percent. The distribution of missing data for the 124 cases is illustrated in Table 2. In running the analysis, the missing data problem was handled by inserting means based on available figures. For example, in the case of Morocco there were only two missing data items; therefore the mean based on the other Moroccan data for the particular variable was used while 118 TABLE 2 PERCENTAGES OF MISSING DATA ON EACH INDEPENDENT VARIABLE BY NATION (124 CASES) (0 •p £ •r] O Ol £ £ ■ H H C id O o •p £ < 0 O U P •A 0 1 •rl id O O -A < D -P O ■ P N •H •A -p P Pi m •rH (0 •rH -p P O (D (d H p tJ r H £ £ -P £ ft f t £ < u m £ id P O 0 1 u b (4 A b tn < U O di £ u i . O P * rH rA P £ < ] } -P £ m Oi S M d) O a u £ O Morocco Algeria Tunisia Libya Egypt Sudan Iraq Israel Syria Iran Jordan Lebanon Turkey 17 45 33 45 45 33 50 67 89 100 42 100 100 40 67 25 91 80 22 50 100 100 78 54 100 36 100 100 100 62 83 45 49 33 119 we thus arrive at a more realistic picture of change in certain countries, there are still cases for which no data are available; see, for example, GNP per capita in Tunisia. Method of Measurement The data collected on each of the seven variables were used to calculate yearly percentage rates of change by subtracting the lowest from the highest value for each chronological time segment. For example, the 1958 datum on GNP per capita for Egypt is subtracted—from the 1959 value. This procedure yields the absolute amount of change taking place in each of the 234 cases on each indicator. Next the amount of change in each case was divided by the lowest value in the given segment to convert the figure into a percentage of change. Some studies take the highest and lowest figure for the entire period, calculate the total amount of change during the period, and then divide by the number of time segments. Such an approach, for example, would record the same change for 1949 as for 1954, precluding the pos sibility of detecting when change took place. Therefore, in order to retain as much information as possible, I have calculated the percentage of change on each variable in all of the 234 cases separately. The justification for using a percentage rather than the absolute amount of change in each case is based on the argument that percentages provide a sounder basis for comparison. For example, when the data contain error due to differences in methods and bases for data collection from one nation to another, this is not reflected in the percent age of change figures, as long as data collection in each nation is internally consistent. Use of percentage figures 19 is considered relatively safe for comparing aggregate data. Rapid change is determined on the basis of statis tical mean values. A mean rate of change based on the 124 cases is calculated for each variable. Those cases with a rate of change higher than the mean on a particular variable are measured for the difference between the mean and the actual rate. The degree to which the rate of change is higher than the mean is correlated, in each case, with the degree to which the magnitude of violence is higher than the mean magnitude of violence on the given variable. 19 For example, see Erwin Scheuch, "Cross-National Comparisons Using Aggregate Data: Some Substantive and Methodological Problems," in Comparing Nations: The Use of Quantitative Data in Cross-National Research, ed. by Richard Merritt and Stein Rokkan (New Haven: Yale University Press, 1966), p. 141. Russett and others also indicate the ad vantages of this kind of data. Russet, et. al.. World Hand book of Political and Social Indicators, p. 158. CHAPTER V CROSS-NATIONAL ANALYSIS In this chapter I introduce the analysis of the findings in this study, but, by implication, this chapter has a more far-reaching meaning. This is where I begin to evaluate what Karl Deutsch has called a "dialogue of data and theory."'*' In the context of this study it means that in Chapters I and II the theoretical definitions were stated, and in Chapters III and IV I responded with empirical definitions and data. In the final chapters, I shed some light on the degree to which the dialogue has been meaningful. General Approach and Problems The statistical techniques selected for analyzing the cross-national relationships between all possible "''Deutsch, "The Theoretical Basis of Data Programs, " pp. 33-44. 121 122 combinations of the independent and dependent variables are correlation and regression analyses, both of which presuppose the use of interval scales. The data in this study come together in what can be considered a good ordinal scale, but do not quite qualify for an internal scale, and this fact should be kept in mind when consider ing the significance of the findings in this chapter and Chapter VI. The correlational analyses are product moment correlations, and are evaluated on three basis: first, as to the possibility of accidental correlations; secondly, as to strength of correlations; and finally, as to whether correlations are positive or negative. The questions of accident or chance are answered by stating the level of significance for all relevant correlation coefficients. For example, a coefficient will be significant at the .05, .01, or .001 level, which means that it could be due to chance only five times in 100, one time in 100, or one time in 1000, respectively. Correlation coefficients indicate how strongly related given ariables are; a coef ficient approaching one is high, while the lowest possible is zero. We finally observe whether given variables are positively or negatively related. Correlational analysis gives considerable infor mation about the interrelationship of different pairs of variables, and it is not generally recommended that one go to more sophisticated techniques unless the correlations 2 are reasonably high. Most correlations in the social sciences are not, however, very high, and those in this study are not exceptions. As pointed out above, however, it is common to move on and use more complex techniques when a major purpose of the study is to investigate the dynamics of the independent variables themselves. Most social relationships are not one-to-one, and therefore I am also interested in determining the behavior and impact of the independent variables, as they come together to explain jointly the dependent variables— even if together they only partially account for the variance in the magnitude of violence. Even though the importance of variables has been determined either inductively by correl ation analysis or deductively by theoretical definition, or both, there is value in investigating their relative impact as preconditions for a type of behavior. Regression For a more indepth discussion of the problems involved in this decision see Blalock, Social Statistics. p. 299. 124 3 analysis will be used to gain this additional information. The stepwise regression takes each independent variable, one at a time, and relates it to a dependent variable. The first step takes the independent variable that is most highly correlated with the dependent variable and enters it into a predicted regression equation, and then makes a judgment of all other independent variables to determine which remaining variable will be able to explain more of the variance in the dependent variable at the second step, after the first variable that was entered at step one has explained all that it can. Each variable is entered in order of explanatory power in succeeding steps until they have explained all that they can. These actual entries form a regression equation which is at every step, evaluated against the predicted regres sion equation, until the actual values form a regression most like the predicted curve. The last variable entered will probably have little effect on making the actual 3 Draper and Smith suggest that the stepwise re gression is best for this sort of investigation, and they also suggest some guidelines for the sensible use of this technique, which will be pointed out as they are relevant. See N. Draper and H. Smith, Applied Regression Analysis (New York: John Wiley and Sons, 1966), pp. 171-172. 125 regression match the ideal predicted regression, and thus will make an insignificant contribution to explaining the variance in the dependent variable. This procedure results in a multiple correlation coefficient (R) which shows how much variation the inde pendent variables can collectively explain, and the procedure itself tells us which independent variables were most important. If each variable is explaining the same portion of variance, then after the most important one is entered at step one, the remaining variables will not add significantly to the R. This technique also provides a residual value for each case, which is the difference between the pre dicted and actual regression curve. The higher the magni tude of the residual the higher the error for that case. When a case is not at all typical of the rest of the data it is called an outlier. After careful consideration such a case can be removed from the data, but only if 4 reasons can be determined for its outlier status. This becomes necessary for the cases affected by the Algerian war. 4Ibid.. pp. 26, 94-95. 126 Before going into the data analysis, one final word of caution is called for. While regression analysis is generally used for predictive or causal models, I am only analyzing the extent to which the independent variables are related to the dependent variables. In other words, I am asking whether a high magnitude on the dependent vari ables is a correlate of a rapid rate of change on the independent variables collectively. Certainly it is hoped that eventually more knowledge of these variables will lead to their use in some causal model. A further restric tion here is the fact that by removing all those cases in which no violence occurred, I have eliminated any pos sibility of looking for causal relationships. Another point related to the above problem has been brought out by Haas 1 suggestion that we build a time element into correlational analysis, in order to move a step closer to causal type relationships. He proposes that this be done by establishing a time-lag between the inde pendent and dependent variables, and recommends a three-year 5 time lag. I will first explore this possibility, and then move into the analysis of the data. 5 Haas, "Social Change and National Aggressiveness," P. 226. 127 The rate of change data in this study were correlated with the magnitude of violence with no time-lag, a one-year time-lag, a two-year time-lag, and a three-year time-lag within each nation. For example, the rate of change in 1948 was observed in relation to violence present in 1949 for the one-year time-lag, and for a two- year time-lag correlated with the violence appearing in 1950, and so on. This was done for all the cases exhibit ing violence. The relevant correlation coefficients from the four sets of data are presented in Table 3. The first or no time-lag set of data measures the degree of association between rate of change and mag nitudes of violence within a given year. Using this approach, there are six significant relationships present (those marked with an asterisk in Table 3). The question is whether these associations become stronger or more significant relationships appear as we move through the three time-lag runs. The highlights of this time-lag investigation are briefly discussed below. With no time-lag there are four relationships significant at the .05 level, and two at the .01 level. Rate of change in electrical production appears as a pre condition for a high magnitude of turmoil, internal war 128 TABLE 3 TIME-LAG CORRELATION COEFFICIENTS Population Urban izat ion Migration Electrical Production GNP per Capita Newspaper Circulation Number of Radios No Time-Lag a a Turmoil -.13 -.17 -.17 -.21 .20 -.05 -.10 Conspiracy -.04 .13 .13 -.08 -.06 .16 -.01 Internal war - .20a .12 .00 - .24a .07 -. 19a .10 TMV -.16 .06 -.05 - .28a .11 -.12 .08 1-Year Time-Lag Turmoil -.11 -.09 -.15 - .24a • 26a -.03 -.14 Conspiracy -.08 .08 .04 -.13 -.10 .10 . 52 a Internal war - .2 3a .03 -.07 - .2 0a .14 .2 0a .11 TMV -.17 i 0 u> 1 .11 - .2 5a .16 -.13 .10 2-Year Time-Lag Turmoil -.02 .03 -.13 -.15 .26a .01 -.05 Conspiracy -.09 -.13 -.10 -.13 .11 .07 .15 Internal war -.12 -.05 .09 -.16 .10 -. 18a -.10 TMV -.12 -.09 .01 -. 19a .14 -.12 -.08 3-Year Time-Lag Turmoil -.13 -.04 -.13 -.12 .26a -.09 -.08 Conspiracy -.09 -.19a -.10 -.06 -.15 -.00 .15 Internal war -.12 -.01 .13 -.06 .10 - .2 0a -.09 TMV -.14 -.06 .06 -.05 .16 -. 18a -.06 aSignificant at least at the .05 level. Product moment correlation coefficients above .24 are significant at the .01 level. The levels of significance for the No and I-, 2-, and 3-Year Time-Lags are based on runs with 124, 121, 116 and 114 cases respectively. 129 and TMV. To a lesser degree, rate of change on newspapers is associated with internal war, rate of change in GNP per capita with turmoil, and rate of change in population growth with internal war. Moving to the one-year time-lag, the same six relationships are significant, with the relationships between rate of change in electrical production and tur moil, and between rate of change in GNP per capita and in population and internal war, becoming more pronounced. The correlations of electrical production with both internal war and TMV are weaker, while the relationship between newspapers and internal war remains about the same. None of these changes occurring from no time-lag to the one-year time-lag would suffice to justify use of the one- year time-lag. But appearance of a strong correlation between rate of change in radio use and the magnitude of conspiracy is interesting, and could prove useful if one were considering a future study using only this radio use indicator to measure change. For the most part, the two- and three-year time- lag sets of data reveal increasingly less correlation between rates of change and magnitudes of violence; the two- and three-year lags yield only three and four 130 ' ■ significant relationships, respectively. While no substantial basis for using the one- two- or three-year time-lag results from this exercise, several new significant relationships do appear as the different runs are profiled. At this point, however, these relationships should be considered only as areas for more extensive future investigation, and not as grounds for saying that certain variables suggest that a time-lag exists. In fact, Table 3, if anything, suggests that noth ing is gained by using the time-lags. Thus I have elected to use the no time-lag data, assuming that whenever a rapid rate of change seems to involve a precondition for high magnitudes of violence, the violence will appear in the same year that the rapid change takes place. The six significant correlation coefficients based on the no-time-lag data set illustrate an obvious lack of evidence of the relationship proposed between rapid rate of change and high magnitudes of violence. As illus trated in Table 3, five of the six significant coefficients indicate a negative relationship, and only the correlation between rapid rate of change of GNP per capita and high magnitude of turmoil suggests any evidence in support of the proposed positive relationships. The three strongest 131 coefficients, -.21, -.24 and -.28, are negative, and correlate rapid rate of change in electrical production to each of three violence measures, turmoil, internal war and TMV, respectively. I will not go into the correlation coefficients represented in Table 3 any further at this point because adjustments in the data base are necessary and these adjustments entail changes in both the nature and the degree of the coefficients. The adjustments in the data base, however, did not provide any evidence indicating that the time-lag question should be reopened. Table 3 has primarily been useful in determining whether the decision should be made to use a data set based on a time-lag, and the decision to use no time-lag has been sustained. Before discussing the results of the cross-national correlational analysis any further, a final elimination of data should be explained. The no-time-lag data with 124 cases were used on the first cross-national regression analysis, and an analysis of the residuals indicated that a large amount of error between the estimated and observed curve was due to the Algerian war. Fifteen cases were then removed for the cross-national analyses, resulting in a data set with 109 cases. The correlation and following 132 regression analyses are based on this refined set of data. Correlation Analysis Table 4 contains the correlation matrix for the cross-national study based on 109 cases. There are several aspects of the correlation matrix that are interesting and should be pointed out briefly. The correlation coefficients in the box in Table 4 represent the associations between all independent vari ables and each type of violence. The first thing that is obvious is that the use of composite measures, such as the TMV, excludes much information, and can only increase the uncertainty of the findings. This is most apparent in comparison of all negative or positive relationships for turmoil, conspiracy and internal war with the combined TMV score. For example, the correlations between rate of change in electrical production and turmoil, conspiracy and in ternal war are all negative and less then significant. When these negative relationships are compiled they result in a significant TMV score. Another example of this cumulative effect illustrates the opposite tendency. Rate of change in GNP per capita and turmoil are nearly sig nificantly related, but this fact is not at all reflected TABLE 4 CROSS-NATIONAL CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population -.08 -.02 .05 .15 -.04 .13 „ a .20 -.07 .07 .13 2. Urbanization .18 .37 -.23 -.00 .17 -.22 .13 .35 •21 3. Migration .02 -.24 -.22 .14 -.17 .11 .16 .05 4. Electrical production -.18 .01 -.02 -.14 -.11 -.15 -.23 5. GNP per capita .19 -.01 .18 -.05 -.07 -.00 6. Newspaper circulation .20 .18 .14 -.09 .04 7. Number of radios -.07 -.01 .28 .24 8. Turmoil -.12 .01 .39 9. Conspiracy .14 .28 10. Internal war .88 11. TMV aThe underlined product moment correlation coefficients are significant at the .05 level for 109 cases; those above .24 are significant at the .01 level. The coeffi cients in the box are the 7 independent variables (change) correlated with each of the 4 dependent variables (violence). H U) w 134 in the TMV score as a result of the two low negative relationships between GNP per capita and both conspiracy and internal war. Given the relatively low correlation coefficients in this study, the composite TMV score is too costly, in terms of information loss, and I have therefore adhered to the decision not to use the TMV score as a single composite measure for the dependent variables. The TMV score, then, should be considered as a most general summary type measure, and the primary emphasis placed on the three different types of violence. In terms of the degrees of the relationships profiled in Table 4, the rate of change in population, taken by itself, could be a precondition for a high magni tude of turmoil, but is not significantly associated with the other types of violence. The rate of change in urbanization is negatively related to turmoil, indicating a rapid rate of urbanization alone is not a precondition for turmoil. Rate of urbanization is, however, positively correlated with high magnitudes of internal war. This could be a result of the fact that most internal war takes place in rural areas, forcing rural inhabitants to seek safety in urban areas; thus urbanization by itself would not necessarily constitute a precondition for violence. 135 The rate of change in radio use is positively correlated with the internal war type violence. It seems likely that a' greater number of radios would contribute to the main tenance of an internal war, since this type of violence tends to be the least spontaneous and the longest in dur ation. In general, the summary measure (TMV) indicates that there is a positive relationship between the magnitude of violence and the rate of change in urbanization and radio use, while there is a negative association between violence and electrical production. Setting aside the possible usefulness of the TMV measure for the moment, we are left with only four significant correlation coefficients. Three of the four provide evidence in support of my proposed positive rela tionship between rapid rate of change and high magnitudes of violence. The two strongest coefficients, in fact, are positive, and suggest that rapid rates of change in urbanization and radio use do correlate with high magni tudes of internal war within the perimeters of this study (r = .36, .28 respectively). The third strongest coefficient (r =-.22) relates rate of change in urbanization to magnitude of turmoil. This third coefficient, however, is negative, indicating 136 that a rapid rate of change in urbanization is not a precondition for high magnitudes of turmoil, although it may be for internal wars (r = .36) or for large scale types of violence in general. Further analyses of the relationships between the dependent and independent vari ables are necessary, however, before any complete inter pretation is possible. The above discussion of Table 4 illustrates the difficulties in gaining an understanding of the precondi tions of violence when each independent variable is considered by itself. Thus, before further investigation of the correlation coefficients, I would like to explore the advantages of using a more complex technique. My primary reason for doing this is based on the assumption that changes on the independent variables do not in reality occur in a vacuum and do not separately influence political behavior. Rapid change on several variables might have a cumulative impact, converting what alone might be an unimportant indicator of change to an important part of a set of variables that indicate the general condition of the social structure. The next step, then, is to investi gate the cumulative role of the independent variables as preconditions for the three types of violence, and to 137 indicate their relative importance. Four stepwise regression analyses were run on the data. The first took each of the independent variables separately against the magnitude of turmoil. The second, third and fourth subjected conspiracy, internal war and the TMV, respectively, to the same process. These regres sions are in relation to the correlation coefficients in Table 4. Stepwise Regression on Turmoil The results of the first stepwise regression analysis are summarized in Table 5. These results illus trate the relative importance of the seven independent variables as preconditions for the variance in the magni tude of turmoil. The rows in Table 5 indicate what took place at each step in the regression analysis. Columns one and two list the independent variables in order of importance. The third column contains the multiple correlation coefficient (R) at each step in the analysis; in the fourth the 2 squares of the multiple correlation coefficients (R ) 2 . appear. R is the cumulative percentage of variance in the dependent variable at each step. The last column 138 TABLE 5 SUMMARY OF THE STEPWISE THE MAGNITUDE OF REGRESSION TURMOIL3 FOR Step Number Variable Entered R R2 Increase in R^ 1 Urbanization .22 .05 .05 2 Population .29 .08 .03 3 Newspaper circulation .34 .12 .04 4 Number of radios .36 .13 .01 5 Migration .37 .14 .01 6 Electrical production .38 .14 .01 7 GNP per capita .38 .14 .00 aThe R2 (.14) is significant at the .05 level for 109 cases. 139 contains the percentage of variance due to each independent 2 variable. The R of the last row or seventh step is the cumulative percentage of variance accounted for by all the independent variables taken together. Rate of change in urbanization was entered first, and by itself accounted for only 5 percent of the variance in magnitude of turmoil. In the second step, rate of change in population was entered and explained an addition al 3 percent of the variance in the dependent variable. The rate of change in newspaper circulation was entered into the regression equation at the third step and added about 4 percent, bringing the total explained variance to about 12 percent. Thus the first three variables were entered in the order suggested by the correlation coeffi cients presented in Table 4. The fourth step, however, recorded the rate of change in radio use rather than GNP per capita. Rates of change in migration and electrical production were entered as anticipated for the fifth and sixth steps. Change in GNP per capita was entered at step seven and made an insignificant contribution to explaining the variance in turmoil. No attempt can be made at this point to explain why the relative contribution of GNP per capita should move down three places; it may simply be noted that after rates of change in urbanization, population and newspaper circulation have accounted for all the variance they can, change in GNP per capita can account for only so much additional variance. The other interesting aspect of this situation is that the rate of change in radio use moved from the seventh place in order of relative importance to the fourth place. At the first step (when the urbanization variable was in) the radios variable was still the least significant one, but when population was added to the equation it moved up past electrical production, while GNP per capita dropped in significance but remained above the radios variable. It was at the third step when the rate of change in newspaper circulation was entered that the importance of GNP per capita took its greatest drop, and the radios variable made its greatest gain. Thus the total variance accounted for in terms of change in GNP per capita had at that point been recorded. It is equally interesting to note that once the newspapers variable had accounted for variance, a portion of the rate of change in the radios variable accounted for additional variance in turmoil. All seven of the rate of change variables taken together account for about 14 percent of the variance in 141 the magnitude of turmoil. This is not a great amount of explained variance, but it could be important. This 14 percent result may supplement further explanations of vari ance in some of the works cited above, such as the recent findings of Gurr (he has explained 28 percent of the 0 variance in turmoil for 114 nations). The rate of change in migration, electrical production and GNP per capita together account for only slightly over 1 percent of the variance, and suggest that data collection efforts on these last three variables would not be worth the effort and expense. Before these variables are set aside, however, studies on rates of change and turmoil should be conducted in other sociocultural areas. It should also be remembered that the lack of data on migration and GNP per capita precludes an evaluation of these variables. Stepwise Regression on Conspiracy The magnitude of conspiracy is the dependent variable for the second regression analysis. In this analysis the seven independent variables do not account for a significant amount of variation in the conspiracy type of 6 Gurr, "A Causal Model of Civil Strife: A Compara tive Analysis Using New Indices," p. 1116. 142 violence. Together they only explain about 9 percent of the variance in the dependent variable. This low level was anticipated, due to the fact that none of the independent variables were strongly cor related with conspiracy (see Table 4). No single variable was able to explain more than about 2 percent of the variance regardless of its point of entry into the equation. Thus any attempt to explain the relative contributions of the independent variables would prove less than meaningful. The data in Table 6 provide no reliable indication of the main preconditions for conspiracy. For example, recorded change on the migration variable, at step two one of the three major contributors to the multiple correlation coefficient, is based on large adjustments for missing data, and should be dismissed as meaningless. These results and those of the correlation analysis suggest that there are still major problems in handling conspiracy type events in cross-national comparisons. Ted Gurr, on the other hand, has explained about 40 percent of 7 the variance in conspiracy for 114 nations. His inde pendent variables include some measures that could be ^Ibid. 143 TABLE 6 SUMMARY OF THE STEPWISE REGRESSION FOR THE MAGNITUDE OF CONSPIRACY3 Step Number Variable Entered R R2 Increase in R2 1 Newspaper circulation .14 .02 .02 2 Migration .20 .04 .02 3 Electrical production .23 .05 .01 4 Urbanization .28 .08 .02 5 Number of radios .30 .09 .01 6 GNP per capita .30 .09 .00 7 Population . 30 .09 .00 3The R2, based on 109 cases, is not significant at the .05 level. 144 considered more political in nature than any adopted in this study, and therefore, perhaps, more directly relevant to conspiracy type events. The national profiles discussed in Chapter VI, which provide an indepth analysis of the proposed relationships, add nothing to our understanding of the preconditions of conspiracy. In other words, there is little evidence to suggest that rapid rates of change in social structure are preconditions for conspiracy. Stepwise Regression on Internal War In the analysis of internal war, the proposed correlations accounted for about 29 percent of the vari ance. Thus we may realistically discuss the relative importance and dynamics of the independent variables. The correlation matrix in Table 4 indicates the four most relevant independent variables, starting with the most important, are rates of change in urbanization, number of radios, migration, and electrical production. It should be kept in mind that electrical production is negatively correlated; that is, the absence of a high rate of change in such production is correlated with high magnitudes of internal war. Table 7 illustrates the regression process evolved 145 TABLE 7 SUMMARY OF THE STEPWISE REGRESSION FOR THE MAGNITUDE OF INTERNAL WARa Step Number Variable Entered R 2 R Increase in R^ 1 Urbanization .36 .13 .13 2 Electrical production .47 .22 .10 3 Number of radios .51 .26 .04 4 Newspaper circulation .53 .28 .02 5 Population .54 .29 .01 6 Migration .54 .29 .00 7 GNP per capita .54 .29 .00 a 2 The R (.29) is significant at the .001 level for 109 cases. in this analysis of internal war. At the first step, rate of change in urbanization accounted for about 13 percent of the variance in the dependent variable. As this took place the electrical production variable moved up in potential influence on variance, and the rate of change in number of radios moved to a lower position and entered at step three. The variables steps two and three accounted for additional variance (about 14 percent), while the rate of change in migration variable increasingly lost its explanatory potential. In the end result, the original potential variance that might have been accounted for by rate of change in migration was nearly all accounted for by the variables entered at the first four steps. Possibly the relatively high correlation coefficient for migration and internal war is misleading, and only in the regression analysis did the missing data problem on this variable become noticeable. At the outset, rate of change in number of radios was in second place in terms of potential, but once urban ization had accounted for all the variance it could, then number of radios dropped to third place and rate of change in population moved to fourth place. When the radios variable was entered at the third step then the newspaper 147 variable increased in potential to account for an addition al amount of variance, and the population variable moved down to enter at the fifth step. The rate of change in newspaper circulation finally accounted for nearly 2 percent of the variance, but only after the radios variable had accounted for about 4 percent of the variance. Since the rates of change in newspapers and radios account for dif ferent amounts of variance in the dependent variable, they must measure different types of preconditions for internal war. Given the degree of association and the level of significance found in this analysis, there is reason to consider dropping the variables that indicate the rate of change in population, migration and GNP per capita. These three variables each accounted for less than 1 percent of the variance in the magnitude of internal war. It should be recalled, however, that missing data on migration and GNP per capita could account for the lack of influence of these two variables. The rates of change in urbanization, electrical production, and number of radios accounted for about 13 percent, 10 percent, and 4 percent of the variance in the dependent variable, respectively, and thus they are likely 148 indicators of social preconditions for internal war in the Middle East and North Africa. Stepwise Regression on the Total Magnitude of Violence It was pointed out above that any generalizations about the relationships involving the total magnitude of violence should be made with important reservations. While the three types of violence— turmoil, conspiracy and internal war— are all aggregates of several types of events, the TMV moves further in this direction and is an aggregate of all types of events, including a few events that did not fit any one of the three types of violence. With some risk we can assert that the TMV is a measure of the total intensity, duration and scope of violence, and is not concerned with what the original events might have been; we therefore need not explain why the TMV does not always reflect the nature or degree of violence in each case. It should, however, be interpreted with caution, since the need to distinguish types of violence has only been set aside, not eliminated. This stepwise regression analysis does not substantially alter the order of importance of the inde pendent variables as they appear in the correlation 149 analysis illustrated in Table 4. It does, however, suggest that there are four variables that account for a signifi cant portion of the variance in TMV: Table 8 illustrates the relative importance of the rates of change in number of radios, electrical production, urbanization, and population as preconditions for a greater intensity, duration, and scope of violence (TMV). As the rates of change in number of radios and electrical production were entered at the first two steps, a situation was created in which the rate of change in urbanization accounted for an increasing amount of the variance in the TMV. At step three the urbanization vari able came to account for 9 percent of the variance, more than either of the first two variables accounted for. This is a typical illustration of the way in which independent variables shift in their capacity to explain variance, once they interact statistically with other variables. Steps five, six and seven added almost no explan ation of the variance in the dependent variable. The variables entered at these last three steps qualify for elimination. GNP per capita and migration were nearly useless in this analysis, as anticipated due to missing data. 150 TABLE 8 SUMMARY OF THE STEPWISE REGRESSION FOR THE TOTAL MAGNITUDE OF VIOLENCE3 Step Number Variable Entered R R2 Increase in r2 1 Number of radios .24 .06 .06 2 Electrical production .32 .11 .05 3 Urbanization .47 .20 .09 4 Population .47 .22 .02 5 Newspaper circulation .47 .22 .00 6 GNP per capita .47 .22 .00 7 Migration .47 .22 .00 3The R2 (.22) is significant at the .001 level for 109 cases. 151 Summary In summarizing the results of the four regression analyses I will consider only those variables that accounted for at least 1 percent of additional variance when entered in the regression equation. Table 9 therefore does not include rate of change in GNP per capita, while the rate of change in migration appears only under magnitude of conspiracy. Conspiracy, however, was the least explained dependent variable. This and the missing data problem suggest that, in fact, the migration variable should also be dropped. It is interesting to note that these two variables (migration and GNP per capita) are the ones for which Table 2 shows the greatest percentage of missing data— a fact that has obviously affected their usefulness in the analyses. The most important variable is the rate of change in urbanization. It explained more of the variance in both turmoil and internal war than any of the other independent variables, and made a significant contribution, at one step or another, to the explanation of all measures of violence. In addition, its major contributions were to the two types of violence that are, perhaps, the most relevant types: turmoil and internal war. The importance 152 TABLE 9 SUMMARY OF REGRESSION ANALYSES® Urbanization Population Newspaper circulation Number of radios Electrical production Migration u < 0 >i & 0 3 1 —1 1 —1 M r o ■ r H g 0 ft 0 U 1 < u u G •p 3 0 G o H 14 13 2 4 3 1 4 4 5 3 1 3 2 2 2 ®The numbers represent the steps at which the independent variables entered each of the four regression analyses for the dependent variables. 153 of the turmoil measure lies in the fact that it embraces those types of disruptive events which occur most fre quent ly, and are least organized and therefore most difficult to deal with. What should be kept in mind, however, is the fact that according to Table 4, the rela tionship between rate of change in urbanization and high magnitudes of turmoil is, while strange, a negative one. Internal war ranks high as a type because of the fact that when it does occur it is usually at an extremely high magnitude. The usefulness of the urbanization variable is also enhanched by feneral availability of the necessary data. The second most useful independent variable was the rate of change in number of radios. This variable did not come in at the first step in any of the analyses of violence, but it appears to be most useful in explaining the variance in the TMV. The importance of communication as a precondition for increased intensity, duration and scope of violence has been discussed above, and only needs to be recalled at this point. In terms of data availa bility, the radio use variable ranks above all others and thus should obviously be considered for future studies on the influence of rapid rates of social change on political 154 behavior in this region of the world. Also, it is important to note at this point that the newspaper circu lation variable did not significantly contribute to the explanation of the TMV— which suggests that the number of radios variable has special significance among communica tion variables. The next most important variable is the rate of change in newspaper circulation, which adds somewhat significantly to the explanation of both turmoil and internal war. Its contribution to the explanation of conspiracy must, however, be considered in light of the fact that the total explanation of this dependent variable was not significant. Again, however, this independent variable is high in terms of data availability. Rate of change in newspaper circulation, then, has served as a moderately useful variable. The fourth most useful independent variable is rate of change in electrical production. Its most signif icant contributions are to the explanations of the variance in internal war and the TMV. What should be kept in mind, however, is the fact that the correlation coefficients in Table 4 indicate that the relationships are negative. In other words, it is possible that a low rate of change in 155 electrical production is a likely precondition for both internal war and violence in general. This is not the anticipated direction of the relationship, however, and thus does not support the central hypothesis of this study. This negative association also held for the three time-lag runs, and thus indicates a need for an indepth analysis of the relationship between modernization and violence— or, at any rate, further inquiry into what else this variable might be measuring. Data on the electrical production variable are available— we need now to understand what they mean. The rate of change in population is related to turmoil and the TMV. Its most substantial contribution is to the explanation of the variance in the turmoil variable. In concluding the stepwise regression analyses some observations need to be stated and others restated. First of all, the information gained by retaining the separate measures of both change and violence proves that we know far too little about the behavior of these vari ables to arbitarily combine them into composite measures— or to consider any one or two of the independent variables the primary measure or measures of social change. In this regard, the stepwise regression has proven useful in 156 distinguishing the behavior and relative importance of each of the several independent variables. The regional approach has suggested that some rethinking of the nature and relationships of variables is needed. The analyses reveal that the independent vari ables differ by country in significance and in the nature of their associations with the different types of violence: which suggests the need to consider the proposed relation ships within still smaller universe of nations. This conclusion is not meant to imply that the universalistic approach has over-extended itself at this point, but rather stresses the need to investigate the foundations on which these studies are based, the purpose being to provide a better basis for selecting valid variables. Thus, just as there was, and still is, a need to move from the universal- istic approach to the regional approach in order to learn more about the variables being used, there is also the need to move back another step to the national level and investigate the behavior and relative importance of the variables at that level. This study began by isolating part of a problem that has been analyzed at the universalistic level of aggregation. It moved down to a selected homogeneous 157 region of the world in search of more information about the variables. In keeping with this direction^ I will now go down another level for a brief look at the associations between the independent and dependent variables for each of the nations included in this study. CHAPTER VI LONGITUDINAL ANALYSES OF NATIONAL PROFILES Introduction The purpose of surveying the national profiles is to get a more in-depth look at the degrees and directions, positive or negative, of correlations between the indepen dent and dependent variables. I will first deal with the directions of the correlation coefficients for each nation, then I will reduce the number of nations to those having one or more correlation coefficients significant at least at the .05 level. The complete correlation matrices for the nations without significant coefficients are presented in Appendix B.1 The matrices in Appendix B are for the following nations: Morocco, Tunisia, Libya, Egypt, Israel, Iran, and Turkey. Iran and Libya were most affected by large amounts of missing data. 158 Directions of the Correlations The direction of correlations within variable-sets for each nation is debermined by a longitudinal analysis based on eighteen cases or time segments, in terms of nations, there are thirteen possible frequencies for each variable-set. Significant amounts of missing data on either the independent or the dependent variable for any nation produce a zero correlation, and reduce the frequency by one for each nation in which a particular variable-set is assigned a zero correlation. Assuming the variables are all valid in the longitudinal analyses, I will first investigate the cross-national validity of the variable-sets in terms of the direction of their relationships, as recorded in Table 10. By multiplying the number of possible variable-sets (twenty-eight) by the number of nations (thirteen), we obtain a possible frequency of 364. Each of the twenty- eight variable-sets is essentially an exploratory hypothesis to be tested for each of the thirteen nations in an eighteen-case longitudinal comparison. Of the 364 possible variable-sets, 117 are excluded due to missing data, leaving a frequency of 247. 160 TABLE 10 DIRECTIONS OF CORRELATIONS WITHIN VARIABLE-SETS, IN FREQUENCIES Variable-Sets Correlations Positive Negative None Turmo i1/Populat i on 4 6 3 Turmoil/Urbanization 5 8 0 Turmo i1/Migrat ion 4 6 3 Turmoil/Electrical production 4 8 1 Turmoil/GNP per capita 3 3 7 Turmoil/Newspaper circulation 4 9 0 Turmoil/Radios 4 9 0 Conspiracy/Population 2 6 5 Conspiracy/Urbanization 4 8 1 Conspiracy/Migration 5 2 6 Conspiracy/Electrical production 2 9 2 Conspiracy/GNP per capita 1 5 7 Conspiracy/Newspaper circulation 5 7 1 Conspiracy/Radios 6 6 1 Internal war/Population 2 0 11 Internal war/Urbanization 4 0 9 Internal war/Migration 1 2 10 161 TABLE 10— Continued Variable-Sets Correlations Positive Negative None Internal war/Electrical production 2 2 9 Internal war/GNP per capita 1 1 11 Internal war/Newspaper circulation 0 4 9 Internal war/Radios 2 2 9 TMV/Population 4 6 3 TMV/Urbanization 5 8 0 TMV/Mi gr a t i on 5 7 1 TMV/Electrical production 4 8 1 TMV/GNP per capita 2 4 7 TMV/Newspaper circulation 4 9 0 TMV/Radios 5 8 0 Totals 94 , 153 117 162 At the macro-level, looking at the general hypothe sis that rapid structural changes are positively associated with high magnitudes of violence, I find that this proposi tion is supported 38 percent of the time. Sixty-two percent of the time (153 frequencies) the two basic concepts are negatively associated. Indicating that perhaps there is a greater tendency, in a given time-unit, for violence to be associated with a low rate of structural change— as measured by the seven independent variables used in this study. Table 10 indicates, for example, that on eight of the twenty-eight variable-sets, rapid change is negatively related to violence. More specifically, I am referring to the following variable-sets: turmoil/urbanization; turmoil / electrical production; turmoil/newspapers; turmoil/radios; conspiracy/population; conspiracy/ urbanization; conspiracy/electrical production; and TMV/newspapers. Only in the case of the internal war/urbanization variable-set is there any suggestion that the anticipated positive hypotheses is supported. This, however, is based on information from only four nations and, while it leaves the question open, it is certainly not decisive. In fact, if anything, the evidence suggests that within the 163 perimeters of this study the central hypothesis should be rejected. The real question that is reopened by this overview, however, is related to the question of just what structural changes are being measured by the independent variables. When the general proposition is broken down into the twenty-eight empirical propositions represented by the variable-sets, no general patterns are obvious, and the basic question of cross-national validity becomes increas ingly relevant. For example, I surveyed several studies above, each of which, in one way or another, used measure ments of change on indicators such as GNP per capita, and assumed that whatever it was measuring, it had the same meaning for large numbers of nations. This assumption, how ever, seems less defensible in light of the fact that, in Table 10, GNP per capita is positively related to turmoil in three nations, and negatively in another three. Another example is the variable-set conspiracy/radios, positively correlated in six nations but negatively in six others. This, in view of the fact that I have sought a common sociocultural region in my sample of nations, is less than encouraging. 164 The only thing that can be said with some degree of certainty at this point is that, in terms of the nature of the associations, these variables are not valid cross- nationally. It is necessary, however, to take a closer look at some of the stronger correlation coefficients. The remainder of this chapter considers only the significant correlation coefficients derived from longitudinal analyses. Selected National Profiles Iraq In the case of Iraq, as illustrated in Table 11, there are four correlation coefficients significant at the .05 level of confidence. There were no missing data on any of the variables involved in these four correlations. Rate of change in urbanization is significantly related to two types of violence, turmoil and conspiracy. Urbanization is negatively related to turmoil, which is a relatively spontaneous, unstructured type of violence. This relationship, then, does not support the general hypothesis. In fact, the violent events, such as strikes, riots, and demonstrations, which make up the turmoil dimen sion correlate with a drop in the normally high rate of urbanization in Iraq. Figure 2 profiles this relationship TABLE 11 IRAQ: CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population .01 .00 -.02 .00 -.01 .00 .16 -.07 .06 .09 2. Urbanization .08 .07 .00 .48 .22 -55a .50 .14 .12 3. Migration .07 .00 -.15 .11 .09 .24 .07 .08 4. Electrical production .00 .10 -.23 -.28 -.10 -.45 -.46 5. GNP per capita .00 .00 .00 .00 .00 .00 6. Newspaper circulation -.34 -.33 -.06 -.40 -.44 7. Number of radios -.12 -.08 .52, .51 8. Turmoil -.27 .15 .22 9. Conspiracy .41 .44 10. Internal war .99 11. TMV aThe underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. H cn U1 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Turmoil Urbanization r = -.55 H H H H H H H H H H H H H H H H H H VD VO VO VO vo VO VO VD VO VO VO VO VD VO VD VO VD VO Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul 01 01 Ol 01 01 01 CD VO O H to U) Ul 01 ■O 00 VO O H to <jJ Ul Fig. 2.— Iraq: The magnitude of turmoil and the rate of change in urbanization, over time. 166 167 over the eighteen-year time span. The horizontal axis rep resents the time dimension, the vertical axis both the mag nitude of turmoil and the percentage of change on the 2 independent variable. It should be noted that the indepen dent variable curve denotes rate of change and not absolute values; in Figure 2, for example, the curve represents the rates of change in urbanization and not absolute levels of urbanization. In the years 1948, 1952, and 1956 turmoil events occurred, while at the same time the rate of change in urbanization was also relatively high, but after 1956 the rate of change dropped and remained low during a three-year period of high turmoil. After this period of high turmoil, which accounts for the negative correlation, the rate of change increased dramatically for the last five years covered in this study. The indication here is that a sharp drop in rate of change in urbanization could be a precondi tion for turmoil in Iraq. The rate of change in urbanization correlates in the anticipated direction with conspiracy, characterized as 2 Since two distinct types of variables are scaled on the vertical axis, the areas enclosed by the change and violence curves are not comparable. This cautionary note applies to Figures 3-19 as well. 168 intensively organized and relatively small-scale violence. The rapid change in urbanization correlates with, and slightly precedes, the period of high conspiracy which peaks in 1963 (see Figure 3). The correlations reveal that conspiracy and turmoil are so different that use of a composite measure of violence would have resulted in much lost information. In Table 11, if we note the correlation between rate of change in urbani zation and the TMV, we find a very low coefficient, which by itself would not indicate that this independent variable might be one precondition for conspiracy type events in Iraq. The rapid rate of change in number of radios is very positively related to internal war, as can be seen in Figure 4. Internal war involves organized and large-scale violence, exemplified in guerrilla wars and large-scale revolts. The presumed disruption in communication patterns, as measured by the high rate of change in number of radios, is a likely precondition for internal war, and could also facilitate the outbreak of internal war by providing the new communication channels necessary for the conduct of large-scale, organized violence. In either event, this independent variable does appear to explain some of the 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Conspiracy — Urbanization r = .50 H H H I-1 H M H H H H I-1 H H H M M H VO VD VO VO VO VD VO VO VD VO VD VO VD VD VO VO VD Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul 01 Ol Ol Ol 01 00 VO O H to co 4^ Ul cn •O 00 VD O I-1 NJ CO 4^ Fig. 3.— Iraq: The magnitude of conspiracy and the rate of change in urbanization, over time. 1965 Number of radios H H 1 H H H H H M H* H H H H H H* H H VD VD VD VD VD VD VD VO VO VO VD VO VD VO VO VD VO VD Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul Ol 01 Ol Ol 01 Ol 00 VO O H to CO Ul Ol 00 VD O H to CO Ul Fig. 4.— Iraq: The magnitude of internal war and the rate of change in number of radios, over time. 170 171 variance in the internal war variable in Iraq, and is cer tainly worth further investigation in order to find out more clearly what structural changes it is measuring, and in what types of nations it does or does not measure certain changes. Figure 5 illustrates the high positive association between increase in number of radios and the TMV. In Iraq, however, the TMV is largely a reflection of the violence that took place as internal war, though it does not contra dict the findings on the relationship between increase in number of radios and the magnitudes of turmoil and conspira cy for 1948, 1952, 1956, 1959, 1960, 1961, and 1950, 1958, 1963 and 1964, respectively. Syria There are three significant correlation coefficients for Syria, of which two involving the newspapers variable are especially interesting. Two independent variables, pop ulation and newspaper circulation, appear as positive corre lates of conspiracy. Rate of change in population correlates positively, but the missing data problem makes it difficult to place confidence in the strength of this relationship. It is, 10.0 Number of radios r = .51 H H H H H H H H H I-1 H H H H H H H VO VO VO VO VO VD VO VD VD VO VD VD VO VO VO VO VO VO £> Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul Ol Ol Ol 01 Ol Ol 00 VD O H co Ul cn 00 VO O M fO CO Ul Fig. 5.— Iraq: The total magnitude of violence and the rate of change in number of radios, over time. 172 TABLE 12 SYRIA: CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population -.39 .00 -.15 -.05 .37 .13 -.03 .50,a .00 .12 2. Urbanization .00 -.28 .41 .13 .42 .21 .18 .00 .23 3. Migration .00 .00 .00 .00 .00 .00 .00 .00 4. Electrical production -.15 -.39 -.15 -.35 -.22 .00 -.36 5. GNP per capita -.45 .15 .00 -.07 .00 -.01 6. Newspaper circulation .10 .36 .65 .00 .51 7. Number of radios .15 .13 .00 .09 8. Turmoil .34 .00 .92 9. Conspiracy .00 .66 10. Internal war .00 11. TMV 3 The underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. H u> 174 however, clearly positive. Although 50 percent of the data are missing, data from 1955 to 1965 are nearly complete. Clearly it is between the years 1961 and 1965 that the rate of change in population appears as a possible precondition for conspiracy. Thus the correlation is especially note worthy for years after 1955. The mean rate of change in population for Syria was used in place of missing data, and has the effect of lowering the correlation coefficient, as may be observed in Figure 6. The rate of change in newspaper circulation seems a far likelier precondition for conspiracy in Syria. The rate of change in newspaper circulation is consistently higher in Syria than in other nations in this study, and increases rapidly during and just prior to the outbreak of conspiracy. The correlation coefficient for the rate of change in newspaper circulation and the TMV is, of course, slightly lower than that recorded for the same variable and conspira cy, primarily due to the low correlation between turmoil and the newspaper variable. Figure 8, when compared with the conspiracy curve in Figure 7, clearly reflects the 1953 and 1964 turmoil events. If we are concerned with the corre lates of the general intensity, duration and scope of vio lence as measured by the TMV, then the findings need not be 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Conspiracy Population r = .50 M H H H H H H H H H H H H I-1 H M H VO VO VO VO VO VO VO VD VO VO VD VO VO VD VO VO VO VO 4^ Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul Ol Ol Ol 01 01 Ol 00 VD O H to U> 4^ Ul Cl -J 00 VO O H to Ul 4^ Ul Fig. 6.— Syria: The magnitude of conspiracy and the rate of change in population, over time. 175 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 /conspiracy 1 * , • I Newspaper circulation l H H H H H H H H H H H H H H H H H H VD VD VO VD VO VO VD VD VD VD VD VD VD VD VD VD VD VD 4^ 4^ (Ji Ul Ul Ul Ul Ul Ul Ul Ul Ul Ol Ol 01 Ol 01 Ol 00 VD O H to U 4^ Ul 01 CO VO O H to U) 4s> Ul Fig. 7.— Syria: The magnitude of conspiracy and the rate of change in newspaper circulation, over time. I \ 10.0 TMV ------------------------ Newspaper circulation --- r = .51 .0 8.0 .0 .0 .0 4.0 .0 0 1.0 0.0 H VO I -1 VD Ul H VD Ul U> H VO Ul Ul H VD H VD Ul Fig. 8.— Syria: The total magnitude of violence and the rate of change in newspaper circulation, over time. 177 178 qualified; but for the purposes of this type of study it should be kept in mind that the degree and direction of this overall correlation primarily reflects the strong relation ship between the rate of change in newspaper circulation and conspiracy events in Syria. Lebanon As illustrated in Table 13, there is only one sig nificant correlation coefficient for Lebanon. The data are complete for both the variables, but since they record only one outbreak of internal war (see Figure 9), we cannot ascribe much importance to the positive correlation. I have added another correlation coefficient to this discussion, not because it is relatively high, but on the contrary because it is so low. I am referring to the rela tionship between electrical production and the TMV (r= .02). The importance of the frequency of events is illustrated by the fact that this coefficient, based on many more events than the high internal war score, is so unusually low. Examination of Figure 10 might lead one to deduce that the relationship between rate of change in electrical production and the TMV is much higher than the statistically derived .02. Thus the correlation coefficient can be very TABLE 13 LEBANON: CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population .01 .00 -.01 .00 -.01 -.00 .15 .15 .06 .17 2. Urbanization -.17 • to o .00 -.30 .41 -.20 .06 .27 -.08 3. Migration -.13 .00 .00 -.42 .14 -.35 -.00 -.10 4. Electrical production .00 .05 .08 .08 -.09 .58a .02 5. GNP per capita .00 .00 .00 .00 .00 .00 6. Newspaper circulation .15 -.26 -.44 -.14 -.36 7. Number of radios -.32 -.17 -.11 -.24 8. Turmoil .65 .51 .91 9. Conspiracy .43 .90 10. Internal war .54 11. TMV The underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. H l £ > 10.0 9.0 8.0 Internal war --------- Electrical production r = .58 7.0 6.0 0 4.0 0 0 1.0 .0 H H H H H H H H H H H H H H H H H VO VO VO VO VO VO VO VD VO VD VD VD VO VO VO vO VO Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul Ol 01 01 01 01 00 VD O I-1 to CJ Ul 01 00 VD O H to U) .P* Fig. 9.— Lebanon: The magnitude of internal war and the rate of change in electrical production, over time. 1965 10.0 TMV ------------------- Electrical production r = .02 4.0 H VD H ( - ■ VD Ul H VD Ul H VO H VD Ul Ul H VD Ul H VD H VD 00 Ul Fig. 10.— Lebanon: The total magnitude of violence and the rate of change in electrical production, over time. 181 182 misleading when considered at an aggregate level for a nation. The single incident of internal war in Lebanon would have appeared impressive, statistically, with its r= .58. but obviously we need further data to substantiate such findings. The findings do not, therefore, warrant gen eralization about internal wars in Lebanon. Jordan The matrix in Table 14 reveals four significant cor relations for Jordan. The electrical production variable is positively associated with turmoil, as reflected in the TMV coefficient, and the radio variable is also positively related to both turmoil and the TMV. It must be noted that the correlations with electri cal production are derived largely from mean values in place of nearly 100 percent missing data on production. The years for which data exist, 1962, 1963, and 1964, happen to coin cide with the most violent occasions of turmoil, which accounts for the two misleadingly high coefficients. Both of the correlations have therefore been dropped from the analysis. The remaining significant relationships involve turmoil, the TMV, and change in the number of radios. TABLE 14 JORDAN: CORRELATION MATRIX 1 1. Population 2. Urbanization 3. Migration 4. Electrical production 5. GNP per capita 6. Newspaper circulation 7. Number of radios 8. Turmoil 9. Conspiracy 10. Internal war 11. TMV 2 3 4 5 6 .2 7 .02 -.44 .00 .39 -.06 .01 .00 -.14 .31 o o • -.51 .00 -.08 .00 7 8 9 10 11 .02 .09 -.36 o o • 1 H o I o -o -.17 i to H o o 1 to o o I -.20 1 O o o -.29 00 •59a .33 o o .59 o o • o o o o o o o o .18 .36 00 o • o o .32 .59 o 00 o o .53 .50 .00 .96 .00 .70 .00 £ The underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. H oo oj 184 Figure 11 records a strong and positive correlation between rate of change in number of radios and the turmoil type of violence, supporting the likelihood that rapid rate of change in number of radios is a precondition for turmoil. The rate of change in number of radios is not quite as strongly associated with the TMV, primarily due to a conspiracy event in 1960, during a period of unusually low rate of change in number of radios (see Figure 12). In gen eral, however, this TMV coefficient is largely a reflection of the turmoil/radios relationship. This is obvious when a comparison is made of Figures 11 and 12. The turmoil/radios variable-set, then, provides the only really important cor relation for Jordan. Sudan The correlation matrix for the Sudan reveals only two significant relationships (Table 15), both involving the rate of change in urbanization, for which over 50 percent of the data are missing; data are available, however, for the last several years, during which most of the internal war type violence occurred. Figure 13 indicates that, while the correlation co efficient is relatively high (.59), it should be interpreted 10.0 I Turmoil ! Radios - 9.0 .59 8.0 0 .0 4.0 0 .0 .0 .0 H V D H V D Ul H V D Ul Ul H V D Ul H V D Ul H V D Ul H V D H V D H V D H V O H VO 01 Ul I-1 V D to Ul Fig. 11.— Jordan: The magnitude of turmoil and the rate of change in number of radios, over time. 185 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Radios H H H H H H H H H H H H H H H H H M V D V O V O V D V O V D V D V O V D V O V O V O V O V O V O vo V D V O U l U l U l U l U l U l U l U l U l U l 0 1 0 1 0 1 0 1 0 1 0 1 00 V O O H NJ C O 4 ^ U l O l 00 V O o H t o C O 4 ^ U l Fig. 12.— Jordan: The total magnitude of violence and the rate of change in number of radios, over time. 186 TABLE 15 SUDAN: CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population .00 .00 .00 .00 .00 .00 .00 .00 .00 .00 2. Urbanization .00 -.29 .27 .05 -.23 -.04 -.23 •59a •52. 3. Migration .00 .00 .00 .00 .00 .00 .00 .00 4. Electrical production .28 -.30 .01 -.12 .18 -.40 -.38 5. GNP per capita .13 .02 .08 -.37 -.04 .00 6. Newspaper circulation .01 .34 .02 -.02 .14 7. Number of radios -.30 .25 -.15 -.26 8. Turmoil .08 .32 .61 9. Conspiracy .22 .16 10. Internal war .93 11. TMV aThe underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. o o <! 10.0 Internal war 9.0 Urbanization 59 8.0 1.0 0.0 H H H H H H VD Ul H H H H H H H H H H H Ul Ul Ul Ul U> Ul Ul Ul Ul Ul Ul cn to Ul Fig. 13.— Sudan: The magnitude of internal war and the rate of change in urbanization* over time. 189 with caution. The rate of change in urbanization did rise sharply in 1963, but missing data for 1964 and 1965 caused the curve to drop back to the mean. Also, the outbreak of violence in 1959 does not correlate with a very high rate of change. The correlation coefficient for the rate of change in urbanization and the TMV, profiled in Figure 14, is some what lower than the coefficient for internal war and the same independent variable, primarily due to inclusion of magnitude of turmoil events in the TMV scores. The turmoil occurring in 1951, 1954 and 1955 was correlated with the larger segment of missing data on the urbanization variable. Algeria In the case of Algeria, turmoil and internal war correlate significantly and positively with urbanization. In Table 16, we note negative correlations of turmoil and internal war with the newspaper circulation variable; thus a low rate of change in newspaper circulation may be a like ly precondition for both turmoil and internal war in Algeria. Though the relationships are not significant, we should also note that both the urbanization and newspaper variables reverse the directions of their correlations for the con spiracy type of violence. 10.0 9 . 0 8.0 7 .0 6.0 5 .0 4 . 0 3 .0 2.0 1.0 0.0 TMV --------- U r b a n iz a t io n r = .52 H H H I-1 H M H I-1 H M H (-* H H I-1 VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO 45* 4^ Ui <Ji Ul UI UI Ui Ul Ul Ul Ul cn Ol cn cn cn 01 03 VO O to U> 4^ Ul cn 00 VO o H to Ul 4^ Ul Fig. 14.— Sudan: The total magnitude of violence and the rate of change in urbanization, over time. 190 TABLE 16 ALGERIA: CORRELATION MATRIX 1 2 3 4 5 6 7 8 9 10 11 1. Population .17 -.07 .02 .08 .09 -.00 .00 -.00 .00 .00 2. Urbanization .34 .13 .39 -.66 -.35 •iia -.35 • 52. .52 3. Migration -.27 .38 .21 -.30 -.12 .14 -.18 -.17 4. Electrical production -.10 -.07 .43 .41 -.33 .42 .38 5. GNP per capita -.14 .08 .09 -.05 .08 .08 6. Newspaper circulation .17 -.53 .33 -.47 -.45 7. Number of radios .13 -.11 .15 .15 8. Turmoil -.65 .92 .91 9. Conspiracy -. 72 -.56 10. Internal war .98 11. TMV 3The underlined product moment correlation coefficients are significant at the .05 level for 18 cases. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. 192 Figure 15 illustrates the positive relation between urbanization and turmoil, which as in other cases we must place in the context of the Algerian war for independence. Given our hypothesis, the rise in rates of urbanization between 1953 and 1959 should have created a condition which disrupted social patterns and facilitated increase in the magnitudes of violence. Indications are indeed that the rising urbanization reflects the movement of the French seg ments of the population into urban areas for security, a move we may assume was sufficient to disrupt social rela tions among the French, thus theoretically increasing the possibility of violence. The drop in the rate of change in urbanization during the conspiracy phase is a reflection of both the French exodus and Algerian immigration, both of which signaled a sharp drop in turmoil. The same explanation holds, for the most part, for the relationship of urbanization to both internal war and the TMV (see Figures 16 and 17). Figure 17 illustrates this reversal in direction of the urbanization variable when cor related with conspiracy. As pointed out above, most vio lence of the turmoil and internal war types ended by 1962, but 1964 marked a highpoint of conspiracy type violence, though change in rates of urbanization was low. 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Turmoil Urbanization r = .59 / H H - * H H H H H1 H H H H H H H H H l - 1 VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul Cl cn cn Cl cn cn (X) VO O I-1 NJ OJ Ul Cl -J 00 VO o H to 00 Ul Fig. 15.— Algeria: The magnitude of turmoil and the rate of change in urbanization, over time. 193 10.0 9.0 Internal war Urbanization 52 4.0 H VO H VO cn H VO H H H H - * H H H H H H H Ul Ui Ul Ul Ul Ul cn CO Ul Fig. 16.— Algeria: The magnitude of internal war and the rate of change in urbanization, over time. 194 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 TMV --------- Urbanization r = .52 H H H H H H H H H H M H H I - 1 H H H VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul cn cn cn cn cn cn 00 VO O to U> •F* Ul CTi 00 VO o H to oo Ul Fig. 17.— Algeria: The total magnitude of violence and the rate of' change in urbanization, over time. 195 196 Figures 18 and 19 illustrate the low rate of change on the newspaper variable during the war years. A four-year period with nearly no change on the independent variable accounts for the negative correlation with the turmoil and internal war types of violence. Also, the positive rela tionship between this independent variable and conspiracy results from the fact that by 1964 the rate of change in newspaper circulation was high, coinciding with the 1964 high for the occurrence of conspiracy type violence. In the case of Algeria, then, significant correla tions indicate a positive relationship between rate of change in urbanization and each of three measures of violence: turmoil, internal war, and the TMV. The antici pated positive relationships, however, do not appear between the rate of increase in newspaper circulation and both turmoil and internal war. Summary The longitudinal analyses cover eighteen cases or yearly time segments for each nation, affording a reasonable basis for observing the relative importance and the behavior of variables within each nation. It was assumed that if a particular independent variable behaved generally the same 10.0 9.0 8.0 7.0 6.0 5 .0 4.0 3.0 2 .0 1.0 0.0 Turmoil Newspaper circulation r = -.53 H H H H H H H H H H (-* H I-1 H H VO VO VO VO vo VO VO VO VO VO VO VO VO VO VO VO VO VO 4^ 4*> ui Ul Ul Ul Ul Ul Ul Ul Ul Ul 01 cn cn cn 01 01 00 VO o I-1 to U) 4*> Ul cn <! 00 VO O to U) 4*. Ul Fig. 18.— Algeria: The magnitude of turmoil and the rate of change in newspaper circulation, over time. 197 10.0 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Internal war Newspaper circulation r = -.47 H H H H H H H H H H H H H H H H H H VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO VO Ul Ul Ul Ul Ul Ul Ul Ul Ul Ul 01 O! Ol Ol 01 Ol 00 VO O H to 00 4^ Ul 01 -0 00 VO O H to 00 4^ Ul Fig. 19.— Algeria: The magnitude of internal war and the rate of change in newspaper circulation, over time. 198 199 in cases of all nations, there would then be evidence sup porting the assumption that these nations do in fact form a common sociocultural region, assuming that in any true sociocultural region the above variables would be valid cross-nationally. No attempt was made to observe whether the variables appeared to be valid for other regions of the world. The first check on the homogeneity of my sample of nations concerned the direction of correlations within variable-sets. If the variables had been valid throughout the region, then for all nations in the region the correla tion on any one variable-set would be either uniformly positive or uniformly negative. Looking at all possible cases, including nonsignificant correlations, we find no indication of a common meaning for the variables. In fact, on many variable-sets the relationships were positive for half of the nations and negative for the others (see Table 10). Only three of the twenty-eight variable-sets correlated in the same direction throughout the region. These were, however, sets that involved internal war, which only four or fewer nations had experienced; the three variable-sets were inapplicable to most of the nations. In any event, there seems to be sufficient evidence, based 200 solely on the direction of the relationships, to question the cross-national validity of all the variables used in this study. The probe into national profiles, which involved only those variable-sets with statistically significant associations, was somewhat more productive in terms of supporting the central hypothesis of this study. The pro files provided no basis for rejecting the above conclusion. They did, however, reveal specific relationships that could, within certain perimeters, be stated as hypotheses related to the general hypothesis stated above, and usable in future studies directed to an understanding of the basic problem under investigation here. These specific relationships are stated below. In the substantive area, the goal of this study has been to isolate certain correlations within pairs of dependent and independent variables for the region as a whole. The find ings of the analyses, however, have indicated that any regional propositions based on the present study would be meaningless. The significant findings emerge from the longitudinal analyses, and pertain to single nations. These findings may be stated as follows: In Iraq, a rapid rate of change in urbanization is a precondition for conspiracy. In Iraq, a rapid rate of change in the number of radios is a precondition for internal war. In Syria, a rapid rate of change in population is a precondition for conspiracy. In Syria, a rapid rate of change in newspaper circulation is a precondition for conspiracy. In Lebanon, a rapid rate of change in electrical production is a precondition for internal war. In Jordan, a rapid rate of change in the number of radios is a precondition for turmoil. In the Sudan, a rapid rate of change in urbani zation is a precondition for internal war. In Algeria, a rapid rate of change in urbaniza tion is a precondition for turmoil. In Algeria, a rapid rate of change in urbaniza tion is a precondition for internal war. CHAPTER VII CONCLUSIONS The substantive contribution of this study to the understanding of preconditions for internal violence is lim ited to nine conclusions based on longitudinal analyses ap plied to six nations (see p. 201 above, 206-207 below). The major contribution, however, as was anticipated, is methodo logical: the study has elucidated problems involved in at tempting cross-national analyses of the preconditions for violence within nations. The following conclusions are based on the relatively limited time-space units used in this analysis, and inferences drawn from these findings for testing on other regions of the world should be viewed with considerable caution— especially since the findings suggest that future investigations should be confined to single nations. The significance of the violence category in the structure of conflict regulation is attested by the fact that out of a possible 234 cases, at least 124 contained 202 203 violent events. This is a rough indication of the extent to which violence supplants procedural types of control, rein forcing the assumption that domestic violence exists to an undesirable degree. The breakdown of the types of violence for the region was as follows: turmoil occurred in ninety- two cases, conspiracy in sixty-one, and internal war in only thirteen. This suggests that the occurrence of the turmoil type of violence, which is primarily unorganized and sponta neous, is the most frequently troublesome for the nations in the Middle East and North Africa. The fact that the least organized violence is the most frequent type suggests that there is still a need to look for societal preconditions for the outbreak of turmoil, primarily because the major nongov ernmental actors are usually not identifiable, well-orga nized actors. The fact that they lack formal organization makes it difficult if not impossible to focus on the actors themselves— their capacities, their intents, their goals, or any other characteristics. The significance of the turmoil type of violence thus indicates the increasing need to look beyond the actors themselves for the preconditions of violence, as I have tried to do in this study. 204 The findings of the stepwise regression analyses lead to the following conclusions: 1. The least useful independent variables were rates of change in migration and GNP per capita. 2. The most useful independent variable was rate of change in urbanization for both the turmoil and internal war types of violence. 3. The second most useful variable was rate of change in number of radios for the total magni tude of violence. 4. Rate of change in newspaper circulation was moderately important in explaining both turmoil and internal war. 5. A low rate of change in electrical production is a correlate of internal war and the total magni tude of violence. The use of stepwise regression analysis has provided much useful information regarding the irregular dynamics among the independent variables. This procedure has been useful in two respects. First, because it illustrates the lack of any pattern of relationships among independent and dependent variables, it seems to preclude the possibility that a composite measure of rate of change would be 205 meaningful in this region of the world at this time. Second, it has demonstrated the utility of the stepwise regression as a technique for selecting independent variables to explain a dependent variable. The latter was, perhaps, the most important contribution of the regression analyses, largely because the process clearly exposes the gap between the level of sophistication at which the tech nique is applied, and the level of our present knowledge regarding the variables being used. The sensitivity of this technique was illustrated by the fact that the rates of change on the migration and GNP per capita variables emerged as the least useful measures. This is, of course, as it should logically have been, since they were the variables with the most missing data. At the aggregate level the correlation analysis did not reflect the lack of data; these two variables were from time to time included in high correlation coefficients, suggesting that a significant relationship existed, when in fact there were not enough data to confirm this. The question of missing data is too complex to be fully explored here. I would, however, like to note once more the use of the statistical mean for each nation on each variable in place .of zeros in cases of missing data. The 206 resultant absence of correlation should not be interpreted as a real lack of interrelatedness between the variables. The national profile of the Sudan illustrates the sort of misinterpretation possible at the aggregate level of anal ysis when missing data are involved. The survey of directions of correlations on the variable-sets for each nation reinforced the above indica tions that the variables were not valid across nations (see pp. 16.0-161) . The nation-by-nation approach did yield the following limited propositions concerning the particular nations. These hypothetical statements, based directly on results of the longitudinal analyses, may be considered potentially reliable and useful within the perimeters established by the statements themselves. 1. In Iraq, a rapid rate of change in urbanization is a precondition for conspiracy. 2. In Iraq, a rapid rate of change in the number of radios is a precondition for internal war 3. In Syria, a rapid rate of change in population is a precondition for conspiracy. 4. In Syria, a rapid rate of change in newspaper circulation is a precondition for conspiracy. 207 5. In Lebanon, a rapid rate of change in electrical production is a precondition for internal war. 6. In Jordan, a rapid rate of change in the number of radios is a precondition for turmoil. 7. In the Sudan, a rapid rate of change in urbani zation is a precondition for internal war. 8. In Algeria, a rapid rate of change in urbaniza tion is a precondition for turmoil. 9. In Algeria, a rapid rate of change in urbaniza tion is a precondition for internal war. Although it was hoped this regional study would establish the comparability of the selected variables, the findings afford no indication of general cross-national validity, which means we must suppose that the problems encountered at the present regional level would be com pounded in studies carried out at the universalistic level. The study has not resolved the issues by focusing on the regional universe, but instead cast new doubt on the viabil ity of the universalistic approach. There are two obvious possible explanations for the lack of support of the central hypothesis. The first is, of course, that rapid social change is simply not a precondi tion for domestic violence. The second focuses on what I 208 consider to be the major reason, the general methodological or operational problems involved in this study. If the latter is indeed the chief reason, then we need not reject the hypothesis, since methodological difficulties have prevented the effectual testing of the hypothesis. The methodological and theoretical questions involve many assumptions that had to be made in order to test the central hypothesis. I will focus on what I consider to be the several assumptions most likely to result in linkage breakdowns. I am primarily concerned with the assumptions used in the operationalization of the concept of rapid rate of change, since it is fairly apparent that events defining the dependent variable are in fact clear measures of violence. The several areas that have raised doubts about whether the central hypothesis has actually been tested can best be isolated by looking at some of the assumptions underlying this study. These several assumptions are: 1. That rapid rate of social change can be measured by structural change at the aggregate level of analysis rather than the more in-depth analysis that would look at the effects of change on individuals. That the data used to take measures on the seven independent variables are sensitive enough to indicate the amounts of change that have taken place. That the seven independent variables do in fact provide measures of how much disruptive change has taken place. That each of the seven independent variables are in fact measuring the same thing in all of the nations used in this study. That structural change temporarily disrupts cues and makes it difficult for actors to determine the meaning of behavior (their own behavior as well as that of others). That when actors have trouble giving common meanings to the behavior of others, then they will have difficulty in determining what their own behavior should be. That when actors have trouble giving common meanings to a type of behavior then they will have difficulty in judging behavior as acceptable or unacceptable. 210 8. That when actors have difficulty in determining what is acceptable behavior there is a greater likelihood that unacceptable— that is, violent— behavior will occur. These eight assumptions indicate the major methodo logical or theoretical links, weaknesses in which could individually lead to, or collectively contribute to, the failure to support the central hypothesis of this study. The evidence derived from the longitudinal analyses of national profiles, however, suggests that the most signifi cant problem is with the fourth assumption listed above. This assumption is that each of the independent variables measures the same thing in all of the nations included in this study. For example, the assumption is that the rate of change measured by the urbanization variable indicates the same kind and degree of social disruption within each of the several nations used in this study. I feel, however, that this is the least tenable of the several major assump tions that provide the foundation of this study— i.e., these independent variables are not comparable across the nations used in this study, because the way in which they correlate with different types of violence from nation to nation, as analyzed in the national profiles in Chapter VI, suggests 211 that no one of these variables is measuring the same thing in all of the nations. Given the lack of validity and comparability of these variables, one course seems clear: we could attempt to determine inductively the validity and comparability of variables; rather than deduce what logically seem to be valid variables at the universalistic level, we could follow the approach suggested in the analyses of the national pro files. This would entail longitudinal analyses in order to determine empirically the preconditions for each type of violence within each nation. The purpose of using the na- tion-by-nation approach to establish the validity of vari ables is to accumulate a body of propositions such as those arrived at in Chapter VI (see pp. 201, 206-207 above). Once an adequate number of such propositions are available, and have been tested over time in each nation, then it will be possible to compare nations on the degree and direction of the relationship between rate of change and domestic vio lence. More precisely, rapid rate of change might well be a precondition for violence in all nations, but the rate of change clearly has no universal indicators for the universe of cases used here. Once we have valid variables for nations, any number of them can be compared. This may be 212 the longest route toward developing the capacity to gener alize about the problem, but the present analysis does not suggest an alternative. The problem will persist, and we will increasingly need to know what social structures can be changed without creating a precondition for violence. Or, at any rate, we should know where violence is most likely to break out if a trigger event occurs. The opening quote by H. L. Nieburg, while perhaps hyperbolic, conveys the urgency implicit in the pursuit of answers to these questions. A P P E N D I X E S 213 APPENDIX A CODING INSTRUCTIONS AND SCALES 214 APPENDIX A The following coding instructions and scales were used to generate the violence data in Ted Gurr1s Civil Strife Data Bank (1961-1965). I have reproduced only those sections of the coding sheets and coding instructions that relate to the scope, duration, and intensity of each violent event. Scope: Number of Initiators, Proximate; Those who carry out the reported event are counted here, not their victims or police who attempt to suppress them. If the report distinguishes between continuously and sporadically active initiators, use the total figure. People said to give frequent assis tance to initiators also are counted as initiators. If two private groups engage in a riot, each attacking the other, the total membership of both groups is used. Code the proximate-scale interval designation in which the estimated number of initiators of the event falls, if suf ficiently precise. If more than 110,000 persons are said to be involved, record their approximate number. If you are unable to distinguish the number of initiators as closely as this scale requires, check "00000" for cols. 22 to 26 and code instead the estimate-scale interval, cols. 27-30, which in your judgment is most likely to compromise the number of initiators. (Note: The proximate 215 216 scale is a rounded geometric progression with a base of 2. Multiplying the interval designation by 20 gives the approximate mid-point of the interval.) Number of Initiators. Proximate 00000 (go to next heading) bbbbl 1 to 40 bbbb3 41 to 80 bbbb 8 81 to 240 bbbl6 241 to 400 bbb32 401 to 900 bbb64 901 to 1, 700 bbl28 1, 701 to 3, 500 bb250 3, 501 to 6, 500 bb500 6, 501 to 14,000 blOOO 14,001 to 27,000 b2000 27,001 to 55,000 b4000 55,001 to 110,000 bbbbb no basis for judging Number on Initiators, Estimate; If the estimate has been fitted to the proximate scale, code "0000" here. If there is no basis for even a rough judgment about the number of initiators, code "bbbbb" for cols. 22-26 and "bbbb" for 27-30. In most cases, however, it should be possible to make an informed guess. Convention: If much or all of a community (territorial, ethnic, religious, communal) is said to be engaged in a rebellion or civil war, and no better estimate of active and supportive participants can be made, the number of initiators is 15% of the community's population (as determined by reference to statistical or other source material). Number of Initiators. Estimate 0000 (previous heading coded) bbb4 less than 100 bb40 101-1,000, "hundreds," "many" b400 1,001-10,000, "thousands" 4000 10,001-100,000 bbbb no basis for judging other 217 Duration; Indicate how much time elapsed in the period covered by the report(s) being coded, whether the event(s) are contin uous or discontinuous. If the entire event is being coded, specify the time elapsed from its onset to end. If the report(s) coded are of either "current period, continuing event" or "all of continuing event, to date" specify the time-span covered, not the total duration of the event(s). If the event is composed of discontinuous incidents, as in small-scale terrorism or a wave of bombings, specify the entire period, from first to last incident. If the period covered falls exactly at a category boundary, for example one month, code it for the lesser of the two categories, "two weeks to one month" in this example. If an event lasts less than 24 hours but spans two calender days, for example a riot beginning one afternoon and ending the next morning, classify it as "1 to 2 days." Code "1/2 day or less" for such events as a single bombing, and riots and demonstrations said to last less than 12 hours. (Note; The duration scale is an adjusted geometric progression. Multiplying the interval designation by 0.3 gives the approximate midpoint of the interval in number of days. By this procedure scale category 1217 is precisely equivalent to one year.) Convention: For plots the duration of the event is "no basis for judging." For purges, specify the length of time the series of government arrests continues, if known. Duration of Action bbbb no basis for judging bbbl 1/2 day or less bbb3 1/2 to 1 day bbb5 1 to 2 days bblO 2 to 4 days bb20 4 days to 1 week bb40 1 to 2 weeks bb80 2 weeks to 1 month bl60 1 to 2 months b320 2 to 4 months b640 4 to 9 months 1217 9 to 15 months 2000 15 months to 2 years 4000 2 to 4 1/2 years 218 Intensity: Code the number of deaths suffered by all the above groups under this heading if the reported data are sufficiently percise to be placed unambiguously within a given category. For example, 310 deaths would be coded "bb256" and "about 100 deaths would be coded "bbb64." Write the number or estimate given in the report under "No. Reported." If more than 16,000, specify under "other." If no deaths are reported and the character of the event, or the otherwise- explicit detail of the report, suggest that none are likely, code "zero, none likely." If data are not precise enough for coding here, check "00000 (go to next heading)." If several estimates of deaths are given, eliminate those said by the source to be unreliable (if any), record the remaining estimates under "No. Reported, " and code their average. If the sources states unequivocally that many more or less deaths occurred than the reported estimate, but makes no corrected estimate, record the reported number and ’ how much in error, in what direction, it is thought to be, but do not code; corroboration should be sought in other sources. Convention: If "casualties" only are reported, without distinguishing between deaths and injuries, 20% are coded as deaths, 80% as injuries. (Note: The deaths scale is an adjusted geometric progression. Multiplying the interval designation by 1.5 gives the approximate midpoint of the interval.) Number of Deaths, Proximate 00000 (go to next headi: bbbbO zero, none likely bbbbl 1 to 2 bbbb3 3 to 6 bbbb 8 7 to 16 bbbl6 17 to 32 bbb32 33 to 64 bbb64 65 to 130 bbl28 131 to 250 bb256 251 to 500 bb500 501 to 1,000 blOOO 1, 001 to 2,000 b2000 2, 001 to 4, 000 219 b4000 4,001 to 8,000 b8000 8,001 to 16,000 other____________ bbbbb no basis for judging No. Reported____________ Number of Deaths, Estimate. If no information could be coded under the preceding heading, attempt a judgment on the basis of contextual materials. If information is coded or recorded under cols. 60-61, above, code here "0000 (previous heading coded)." If the report is too vague or ambiguous for any judgment about number of deaths, code "no basis for judging" under both headings. Number of Deaths. Estimate 0000 (previous heading coded) ^ bbb4 10 or less, "few" bb40 11 to 100, "scores, many" b400 101 to 1,000, "hundreds" 4000 1,001 to 10,000 bbbb no basis for judging The following coding instructions and scales were used to generate the violence data in Feierabend's Cross- National Data Bank of Political Instability Events (1948- 1961). Again, I have only reproduced those sections of the coding instructions that relate to the scope, duration, and intensity of each violent event. Scope; Number Involved. The number of people who participated in each of the following events is recorded in this field: resignations, dismissals, appointments, strikes, demonstra tions, boycotts, arrests, assassinations, executions, coups, 2 2 0 revolts, and exiles. Again, it should be noted that both columns are never used for the same event. The number of people in absolute terms, if given. 0 = no data or zero 1 = 1 2 = 2-10 3 = 11-50 4 = 51-100 5 = 101-500 6 = 501-1,000 7 = 1,001-10,000 8 = 10,001-100,000 9 = 100,001 + The number of people involved, based on an intuitive rating of the particular event. 0 = not enough data for rating 1 = individual 2 = few (between 2-50 persons on resignations, dismissals, appointments, arrests, executions, and exiles. Between 2-10,000 persons on strikes, demonstrations, boycotts, coups, and revolts). 3 = mass (over 50 persons on resignations, dis missals, appointments, arrests, executions, and exiles. Over 10,000 persons on strikes, demonstrations, boycotts, coups, and revolts). Duration: This field describes the length of time during which the event took place. Duration is considered pertinent only for the following events: strikes, demonstrations, boy cotts, martial law, terrorism, sabotage, guerrilla war, civil war, coups, revolts, and exiles. It should be noted that cols. 17 and 18 are never used together; one or the other is used, depending upon the availability of data. The duration, in absolute terms, if given. 0 = no data 1 = 0 to 1/2 day 2 2 1 2 = 1/2 to 1 day 3 = 1 to 2 days 4 = 2 days to 1 week 5 = 1 week to 2 weeks 6 = 2 weeks to 1 month 7 = 1 month to 6 months 8 = 6 months to 1 year 9 = 1 year + The relative duration based on an intuitive rating of the particular event. 0 = not enough data for rating 1 = short (a week or less for all events except civil war and exiles, for which six months or less is used as the gauge) 2 = long (over a week for all events except civil war and exiles, for which the breaking point is over six months) Intensity; The number of people killed in each of the events is recorded in this field. Once again, both cols, are not used for the same event. Number of people killed, in absolute terms, if given. 0 = no data or zero 1 = 1 2 = 2-10 3 = 11-50 4 = 51-100 5 = 101-500 6 = 501-1,000 7 = 1,000-10,000 8 = 10,001-100,000 9 = 100,001 + Number of people killed, based on an intuitive rating of the particular event. 0 = not enough data for rating 1 = individual 2 = few (2-50 persons) 3 = mass (over 50 persons) APPENDIX B CORRELATION MATRICES FOR SEVEN NATIONS 222 TABLE 17 TURKEY: CORRELATION MATRIX 3 1 2 3 4 5 6 7 8 9 10 11 1. Population I o 0 1 • u> .39 .02 -.10 .02 -.11 -.35 .00 -.11 2. Urbanization .28 .26 .58 .43 .57 -.30 -.00 .00 -.30 3. Migration .08 -.08 .18 .10 .16 .34 .00 .14 4. Electrical production .26 -.23 .15 .15 -.10 .00 .07 5. GNP per capita -.06 .10 -.10 -.30 .00 -.16 6. Newspaper circulation .39 -.16 -.02 .00 -.17 7. Number of radios -.36 -.23 .00 -.38 8. Turmoil .55 .00 .97 9. Conspiracy .00 .66 10. Internal war .00 11. TMV aNone of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to to TABLE 18 MOROCCO: CORRELATION MATRIX3 1 1. Population 2. Urbanization 3. Migrat ion 4. Electrical production 5. GNP per capita 6. Newspaper cimilation 7. Number of radios 8. Turmoil 9. Conspiracy 10. Internal war 11. TMV 2 3 4 5 6 .23 .26 -.51 .29 -.27 -.27 .69 -.05 .15 -.45 .55 -.13 -.29 .10 .01 7 8 9 10 11 -.16 o I —1 i —1 0 1 o o .09 .56 -.32 -.33 .00 -.34 c r * CM I -.10 .01 .00 -.11 CD I T ) -.28 i H .00 -.32 00 CM .33 .15 .00 to 00 r - ~ o 1 l o to -.20 .00 -.02 -.01 -.17 .00 -.07 .56 .00 .98 .00 .68 .00 3None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to TABLE 19 ISRAEL: CORRELATION MATRIX 3 1 1. Population 2 . Urbanization 3. Migration 4. Electrical production 5. GNP per capita 6. Newspaper circulation 7. Number of radios 8. Turmoil 9. Conspiracy 10. Internal war 11. TMV 2 3 4 5 6 ,24 .51 .40 -.38 .52 -.08 -.05 -.22 -.10 .36 .14 .05 -.19 .52 -.50 7 8 9 10 11 .26 I to o o o o 1 to .19 CM CM 1 o o o o -.22 o o 1 O CD o o o o -.09 .05 00 O 1 o o o o 00 0 1 .44 0 1 —1 1 o o o o -.10 .23 -.13 o o o o -.13 -.31 o o o o -.31 .00 .00 .99 .00 .00 .00 None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to U1 TABLE 20 IRAN: CORRELATION MATRIX3 1 2 3 4 5 6 7 8 9 10 11 1. Population o o 0 C M 1 .00 .00 -.27 -.13 -.02 -.01 .00 -.00 2. Urbanization .00 .00 .00 -.12 -.06 .06 .15 .00 .09 3. Migration .00 .00 .00 .00 .00 .00 .00 .00 4. Electrical production .00 .00 .00 .00 .00 .00 .00 5. GNP per capita .00 .00 .00 .00 .00 .00 6. Newspaper circulation .08 -.09 -.31 .00 -.18 7. Number of radios -.32 .06 .00 -.24 8. Turmoil .05 .00 .94 9. Conspiracy .00 .33 10. Internal war .00 11. TMV 3None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. NJ N3 c r * TABLE 21 EGYPT: CORRELATION MATRIX3 1 1. Population 2. Urbanization 3. Migration 4. Electrical production 5. GNP per capita 6. Newspaper circulation 7. Number of radios 8. Turmoil 9. Conspiracy 10. Internal war 11. TMV 3 4 5 6 o o • .34 -.19 .04 .19 -.13 -.03 .30 -.06 -.74 .00 < x > i —i i -.17 -.02 7 8 9 10 11 -.38 -.17 -.33 .00 -.25 ■ —I m I -.09 -.08 .00 -.12 -.02 .22 .00 .00 .22 I • H 00 -.18 -.02 .00 -.19 .16 -.17 -.07 .00 -.18 -.15 -.11 .02 .00 -.00 .15 .10 .00 .08 .43 .00 .93 .00 -.68 .00 None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to TABLE 22 LIBYA: CORRELATION MATRIX3 1 2 3 4 5 6 7 8 9 10 11 1. Population .00 .00 .00 .00 .00 .00 .00 .00 .00 .00 2. Urbanization .31 .08 .00 .51 -.35 .03 -.12 .00 -.02 3. Migration -.03 .00 .37 -.43 -.14 .00 .00 -.14 4. Electrical production .00 .18 .01 -.04 -.06 .00 -.06 5. GNP per capita .00 .00 .00 .00 .00 .00 6. Newspaper circulation -.18 -.26 -.16 .00 -.29 7. Number of radios -.25 .28 .00 -.17 8. Turmoil .20 .00 .98 9. Conspiracy .00 .35 10. Internal war .00 11. TMV 3None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to o o TABLE 2 3 TUNISIA: CORRELATION MATRIX3 1 2 3 4 5 6 7 8 9 10 11 1. Population -.66 .36 .31 .00 -.27 -.34 -.15 .00 .00 -.12 2. Urbanization .00 -.05 .00 .21 .18 .02 -.09 .00 -.06 3. Migration .36 .00 -.26 -.56 -.15 .02 .00 -.10 4. Electrical production o o • -.25 -.18 -.23 -.02 .00 -.20 5. GNP per capita .00 .00 .00 .00 .00 .00 6. Newspaper circulation .26 .23 -.11 .00 .10 7. Number of radios -.36 -.11 .00 I • CO 00 8. Turmoil -.24 .00 .61 9. Conspiracy .00 .62 10. Internal war .00 11. TMV 3None of the product moment correlation coefficients are significant at the .05 level. The coefficients in the box are the 7 independent variables correlated with each of the 4 dependent variables. to to B I B L I O G R A P H Y 230 SELECTED BIBLIOGRAPHY Books Apter, David. The Politics of Modernization. Chicago: University of Chicago Press, 1965. Aristotle. Aristotle's Politics and Poetics. Translated by Benjamin Jowett and Thomas Twining. New York: Viking Press, 1957. Bienen, Henry. Violence and Social Change. Chicago: University of Chicago Press, 1968. Blalock, Hubert, Jr. Social Statistics. New York: McGraw-Hill Book Co., I960. Boulding, Kenneth. Conflict and Defense: A General Theory. New York: Harper and Row, 1963. Buckley, Walter. Sociology and Modern Systems Theory. Englewood Cliffs, N. J.: Prentice-Hall, 1967. Cohen, Albert. Deviance and Control. Englewood Cliffs, N. J.: Prentice-Hall, 1966. Coser, Lewis. The Functions of Social Conflict. New York: The Free Press, 1964. Dahrendorf, Ralf. Class and Class Conflict in Industrial Society. Stanford: Stanford University Press, 1959. Deutsch, Karl. The Nerves of Government. New York: The Free Press, 1966. 231 232 Deutsch, Karl. "The Theoretical Basis of Data Programs." Comparing Nations; The Use of Quantitative Data in Cross-National Research. Edited by Richard Merritt and Stein Rokkan. New Haven: Yale University Press, 1966. Drake, St. Clair. "Urban Violence and American Social Move ments ." Urban Riots: Violence and Social Change. Edited by Robert Connery. New York: The Academy of Political Science, Columbia University, 1968. Draper, N. R. and Smith, H. Applied Regression Analysis. New York: John Wiley and Sons, 1966. Eckstein, Harry. A Theory of Stable Democracy. Research Monograph No. 10. Princeton: The Center of Interna tional Studies, Princeton University, 1961. _________. "Toward the Theoretical Study of Internal War." Internal War: Problems and Approaches. Edited by Harry Eckstein. New York: The Free Press, 1964. Etzioni, Amitai. The Active Society: A Theory of Societal and Political Processes. New York: The Free Press, 1968. Etzioni, Amitai, and Etzioni, Eva, eds. Social Change: Sources. Patterns and Conseguences. New York: Basic Books, 1964. Feldman, Arnold. "Violence and Volatility: The Likelihood of Revolution." Internal War: Problems and Approaches. Edited by Harry Eckstein. New York: The Free Press, 1964. Fogelson, Robert. "Violence as Protest." Urban Riots: Violence and Social Change. Edited by Robert Connery. New York: The Academy of Political Science, Columbia University, 1968. Galtung, Johan. Theory and Methods of Social Research. New York: Columbia University Press, 1967. Gurr, Ted. New Error-Compensated Measures for Comparing Nations: Some Correlates of Civil Violence. Research Monograph No. 25. Princeton: The Center of Interna tional Studies, Princeton University, 1966. Gurr, Ted, with Ruttenberg, Charles. The Conditions of Civil Violence: First Tests of a Causal Model. Re search Monograph No. 28. Princeton: The Center of International Studies, Princeton University, 1967. Haas, Michael. "Social Change and National Aggressiveness, 1900-1960." Quantitative International Politics: In sights and Evidence. Edited by J. David Singer. New York: The Free Press, 1968. Janos, Andrew. "Authority and Violence: The Political Framework of Internal War." Internal War: Problems and Approaches. Edited by Harry Eckstein. New York: The Free Press, 1964. Johnson, Chalmers. Revolutionary Change. Boston: Little, Brown and Co., 1966. Lasswell, Harold. Politics: Who Gets What. When, How. Cleveland: The World Publishing Co., 1958. _________. World Politics Faces Economics. New York: McGraw-Hill Book Co., 1945. Leiden, Carl and Schmitt, Karl. The Politics of Violence: Revolution in the Modern World. Englewood Cliffs, N. J.: Prentice-Hall, 1968. Levy, Marion, Jr. Modernization and the Structure of Societies: A Setting for International Affairs. Vol. I. Princeton: Princeton University Press, 1966. _________. The Structure of Society. Princeton: Princeton University Press, 1952. Merritt, Richard, and Rokkan, Stein, eds. Comparing Nations: The Use of Quantitative Data in Cross-National Research. New Haven: Yale University Press, 1966. Merton, Robert. Social Theory and Social Structure. New York: The Free Press, 1957. Moore, Wilbert. Social Change. Englewood Cliffs, N. J.: Prentice-Hall, 1963. 234 Parson, Talcott. The Social System. New York: The Free Press, 1951. Pye, Lucian. Aspects of Political Development. Boston: Little, Brown and Co., 1966. Report of the National Advisory Commission on Civil Dis orders . Otto Kerner, Chairman of the National Advisory Commission on Civil Disorders. New York: Bantam Books, 1968. Rights in Conflict: Report to the National Commission on the Causes and Prevention of Violence. Daniel Walker, Director of the Chicago Study Team. New York: Bantam Books, 1968. Russett, Bruce. "Delineating International Regions." Quantitative International Politics: Insights and Evidence. Edited by J. David Singer. New York: The Free Press, 1968. Russett, Bruce; Alker, Hayward; Deutsch, Karl; and Lasswell, Harold. World Handbook of Political and Social Indi cators . New Haven: Yale University Press, 1964. Scheuch, Erwin. "Cross-National Comparisons Using Aggre gate Data: Some Substantive and Methodological Prob lems ." Comparing Nations: The Use of Quantitative Data in Cross-National Research. Edited by Richard Merritt and Stein Rokkan. New Haven: Yale University Press, 1966. Scott, John. Aggression. Chicago: University of Chicago Press, 1958. Simmel, Georg. Conflict and the Web of Group-Affiliations. Translated by Kurt Wolff and Reinhard Bendix. New York: The Free Press, 1964. Smelser, Neil. "Toward a Theory of Modernization." Social Change: Sources. Patterns and Conseguences. Edited by Amitai Etzioni and Eva Etzioni. New York: Basic Books, 1964. 235 Sorel, Georges. Reflections on Violence. New York: Collier Booksj 1961. _ Tocqueville, Alexis de. The Old Regime and the French Revolution. Garden City: Doubleday, 1955. Zawodny, J. K. Man and International Relations: Contri butions of the Social Sciences to the Study of Conflict and Integration. Vol. I: Conflict. Vol. II: Inte gration. San Francisco: Chandler Publishing Co., 1966. Articles Alexander, Franz. "A World Without Psychic Frustration." American Journal of Sociology. XLIX (1944), 465-469. Birnbaum, N. "Conflicting Interpretations of the Rise of Capitalism: Marx and Weber." The British Journal of Sociology. IV (June, 1953), 125-141. Blumer, Herbert. "The Idea of Social Development." Studies in Comparative International Development. II, No. 1 (1966), 1-11. Boulding, Kenneth. "Organization and Conflict." Journal of Conflict Resolution. I (1957), 122-134. Collomb, H., and Valantin, S. "Patterns of Mothering, Organization of the Personality and Rapid Social Change." International Journal of Social Science. XX (1968), 431-447. Coser, Lewis. "Social Conflict and the Theory of Social Change." The British Journal of Sociology. VIII (September, 1957), 197-207. _________ . "The Termination of Conflict." The Journal of Conflict Resolution. V (1961), 347-353. Dahrendorf, Ralf. "Toward a Theory of Social Conflict." Journal of Conflict Resolution. II (1958), 170-183. 236 Eckstein, Harry. "On the Etiology of Internal Wars." History and Theory. IV (1965), 133-163. Feierabend, Ivo, and Feierabend, Rosalind. "Aggressive Behavior within Politics, 1948-1962: A Cross-National Study." Journal of Conflict Resolution. X (1966), 249-271. Frenkel-Brunswick, Else. "Social Tensions and the Inhibi tion of Thought." Social Problems. II (October, 1954), 75-81. Galtung, Johan. "Institutionalized Conflict Resolution: A Theoretical Paradigm." Journal of Peace Research. II (1965), 348-397. _________. "On the Meaning of Nonviolence." Journal of Peace Research. II (1965), 228-257. _________. "A Structural Theory of Aggression." Journal of Peace Research. I (1964), 95-119. Gregg, Phillip, and Banks, Arthur. "Dimensions of Politi cal Systems: Factor Analysis of a Cross-Polity Survey." American Political Science Review. LIX (September, 1965), 601-614. Gurr, Ted. "A Causal Model of Civil Strife: A Comparative Analysis Using New Indices." American Political Science Review, LXII (December, 1968), 1104-1124. _________. "Psychological Factors in Civil Violence." World Politics. XX (January, 1968), 245-278. Haas, Michael. "Aggregate Analysis." World Politics. XIX (October, 1966), 106-122. Holzman, Mathilda. "Theories of Choice and Conflict in Psychology and Economics." Journal of Conflict Reso lution. II (1958), 310-320. Kadt, Emanuel de. "Conflict and Power in Society." International Social Science Journal. XVII (1965), 454-471. 237 Kalleberg, Arthur. "The Logic of Comparison: A Method ological Note on the Comparative Study of Political Systems." World Politics. XIX (October, 1966), 69-82. Mack, Raymond, and Snyder, Richard. "The Analysis of Social Conflict— Toward an Overview and Synthesis." Journal of Conflict Resolution. I (June, 1957), 212-248. Montgomery, John. "The Quest for Political Development." Comparative Politics. I (January, 1969), 285-295. Murphy, Robert. "Intergroup Hostility and Social Cohesion." American Anthropologist. LIX (1957), 1018-1035. Nieburg, H. L. "Uses of Violence." Journal of Conflict Resolution. VII (1963), 43-54. Nordlinger, Eric. "Political Development: Time Sequences and Rates of Change." World Politics. XX (April, 1968), 494-520. Olson, Mancur, Jr. "Rapid Growth as a Destabilizing Force." Journal of Economic History. XXIII (December, 1963), 529-552. Parsons, Talcott. "Certain Primary Sources and Patterns of Agression in the Social Structure of the Western World." Psychiatry. X (1947), 167-181. Rose, Arnold. "Theory for the Study of Social Problems." Social Problems. IV (January, 1957), 189-199. Rummel, Rudolph. "Dimensions of Conflict Behavior within Nations, 1946-59." Journal of Conflict Resolution. X (March, 1966), 65-73. _________. "Dimensions of Conflict within and between Nations." General Systems Yearbook. VIII (1963), 1-50. _________. "Understanding Factor Analysis." Journal of Conflict Resolution. XI (December, 1967), 444-479. Sjoberg, Gideon. "The Comparative Method in the Social Sciences." Philosophy of Science. XXII (April, 1955), 106-117. 238 Sjoberg, Gideon. "Contradictory Functional Requirements and Social Systems." Journal of Conflict Resolution. IV (1960), 198-208. Stagner, Ross. "Personality Dynamics and Social Conflict." Journal of Social Issues. XVII (1961), 28-44. Stone, Lawrence. "Theories of Revolution." World Poli tics, XVIII (January, 1966), 159-176. Tanter, Raymond. "Dimensions of Conflict Behavior within and between Nations, 1958-60." Journal of Conflict Resolution. X (March, 1966), 41-64. Tanter, Raymond, and Midlarsky, Manus. "A Theory of Revolution." Journal of Conflict Resolution. XI (September, 1967), 264-280. Verba, Sidney. "Some Dilemmas in Comparative Research." World Politics. XX (October, 1967), 111-127. Wallace, Anthony. "Mazeway Disintegration: The Individ ual's Perception of Socio-Cultural Disorganization." Human Organization, XVI (Summer, 1957), 23-27. Wright, Quincy. "The Nature of Conflict." The Western Political Quarterly, IV (June, 1951), 193-209. Zolberg, Aristide. "The Structure of Political Conflict in the New States of Tropical Africa." American Political Science Review. LXII (March, 1968), 70-87. Unpublished Materials Collins, John. "Foreign Conflict Behavior and Domestic Disorder in Africa." Unpublished Ph.D. dissertation, Northwestern University, 1967. Eckstein, Harry. "The Incidence of Internal Wars, 1946- 1959." Paper presented to the Research Group in Psy chology and the Social Sciences, Smithsonian Institu tion, Washington, D.C., January 15, 1962. 239 Feierabend, Ivo; Feierabend, Rosalind; and Nesvold, Betty. "Social Change and Political Violence: Cross-National Patterns." Paper presented to the Task Force on History and Comparative Analysis of the National Com mission on the Causes and Prevention of Violence, November, 1968. (Mimeographed.) Firestone, Joseph. "Political Conflict: A Theory and Comparative Analysis in Five Cultures." Unpublished Ph.D. dissertation, Michigan State University, 1965. Gurr, Ted. "The Genesis of Political Violence: A Theo retical Analysis." (Mimeographed.) Halpern, Manfred. "Conflict, Violence, and the Dialectics of Modernization." Paper presented at the 64th Annual Meeting of the American Political Science Association, Washington, D.C., September 3-7, 1968. Lupsha, Peter. "On Theories of Urban Violence." Paper presented at the 64th Annual Meeting of the American Political Science Association, Washington, D.C., September 3-7, 1968. McClelland, Charles, and Hoggard, Gary. "Conflict Patterns in the Interactions Among Nations." University of Southern California, July, 1968. (Mimeographed.) Riggs, Fred. "Structure and Function: A Dialectical Approach." Paper presented at the 63rd Annual Meeting of the American Political Science Association, Chicago, September 5-9, 1967. Rummel, Rudolph. "Dimensionality of Nations Project: Orthogonally Rotated Factor Tables for 236 Variables." July, 1964. (Mimeographed.) _. "The Dimensionality of Nations Project: A Five Year Research Program." Research Report No. 9. Uni versity of Hawaii, 1968. (Mimeographed.) _. "Some Empirical Findings on Nations and Their Behavior." Research Report No. 10. University of Hawaii, 1968. (Mimeographed.) 240 Tanter, Raymond. "Dimensions of Conflict Behavior within and between Nations, 1958-60." Northwestern University, August, 1964. (Mimeographed.) _________. "Dimensions of Conflict Behavior within Nations, 1955-60: Turmoil and Internal War." Paper presented at the Peace Research Conference, University of Chicago, November 16-17, 1964. Teune, Henry. "Intra-System Comparisons in Cross-Systems Analysis." Paper presented at the 64th Annual Meeting of the American Political Science Association, Washing ton, D.C., September 3-7, 1968.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Nationalism And Communication Among Dispersed Peoples: The Jewish And Palestinian Diasporas

PDF

Models In The Study Of Political Development: A Critique

PDF

The Politics Of The German Occupation In The Protectorate Of Bohemia And Moravia: A Case Study Of A Totalitarian "Breakthrough"

PDF

Liberalism And Conservatism: A Study Of Mass-Media Foreign Policy Attitudes

PDF

Management Of Change In Organizations--A Process Analysis

PDF

An Examination Of Economic Expectations As A Determinant Of Political Behavior

PDF

Representation Theory And Apportionment Models: A Study Of California Constituencies

PDF

The Effectiveness And Credibility Of Arab Propaganda In The United States

PDF

Totalitarianism, groups, and technology: a case study of professionalism in the East German military: 1949-1971

PDF

Patterns In Political Violence And War, 1751-1960

PDF

Some Effects Of Monetary And Fiscal Policy On The Distribution Of Wealth

PDF

Social Change And Development: A Sociological Analysis

PDF

A Theoretical Analysis Of The Dynamic Aspects Of Economic Integration Of Less Developed Countries

PDF

A Developing Pattern Of Unenumerated Individual Rights In The United States Constitution

PDF

Perception Of The Power Structure By Social Class In A California Community

PDF

Contemporary Theological Approaches And The Political Role Of The Anglo-American Protestant Denominations In An Urban Complex

PDF

Valuation Adjustment In Input-Output Tables, Economic Structure, And Economic Systems: A Comparative Empirical Analysis

PDF

A Comparison Of Contemporary Conservatism In Great Britain And The Unitedstates

PDF

External Trade And Economic Growth In Developing Countries With Special Reference To Iraq

PDF

A Monte Carlo Evaluation Of Interactive Multidimensional Scaling

Asset Metadata

Creator Harrell, Roger Herman (author)

Core Title Domestic Violence And Indicators Of Social Change Within Nations: A Regional Perspective

Degree Doctor of Philosophy

Degree Program Political Science

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

Tag OAI-PMH Harvest,political science, general

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Murray, Ray (committee chair), Beling, Willard A. (committee member), Thompson, Ken (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-404370

Unique identifier UC11360953

Identifier 7016866.pdf (filename),usctheses-c18-404370 (legacy record id)

Legacy Identifier 7016866.pdf

Dmrecord 404370

Document Type Dissertation

Rights Harrell, Roger Herman

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 au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA