INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.
UMI University Microtilms International A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor, Ml 48106-1346 USA 313/761-4700 800/521-0600
Order Number 9211220
A multivariate model of ethnic diversity and violent political behavior
Shih, Cheng-Feng, Ph.D. The Ohio State University, 1991
U-M 300N.ZeebRd. Ann Arbor, MI 48106
A MULTIVARIATE MODEL OF ETHNIC DIVERSITY AND VIOLENT POLITICAL BEHAVIOR
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of the Ohio State University
By
Cheng-Feng Shih, B.S., M.A.
*****
The Ohio State University
1991
Dissertation Committee: Approved by Chadwick F. Alger, Ph.D. Saad Z. Nagi, Ph.D. Adviser Goldie A. Shabad, Ph.D. Department of Political Science ACKNOWLEDGMENTS
I am deeply indebted to three teacher who have guided me at different stages of this study. I am grateful to Dr. Nagi for his reading the manuscript and offering methodological comments. I also extend my sincere appreciation to Dr. Shabad for her patient teaching and suggestions on theoretical issues. I am most deeply indebted to Dr. Alger, my adviser, who has encouraged and guided me during study at the Ohio State University. I still have a vivid image of the first day we met in the department reception, where he challenged: "What if Texas wants to get independence?" My strong commitment to the idea of self-determination led me to answer spontaneously: "Why not!" He has become my mentor since then. I can never appreciate him enough for his insights and stimulating suggestions spanning the content in all of the dissertation chapters.
I like to thank my parents for their love and support. To my wife Pei-Ing, I like to say without your encouragement and confidence in me, my study in the graduate school in this country would have not been possible.
ii VITA
February 24, 1958 .. Born - Taichung, Taiwan (Formosa) .. B.S., Department of Agricultural Economics, National Taiwan University, Taipei, Taiwan 1985-1986 .. Teaching and Research Assistant, Department of Political Science, Iowa State University. 1986 .. M.A., Iowa State University, Ames, Iowa 1986-1987 .. Graduate Teaching and Research Associate, Department of Political Science, the Ohio State University. 1989-1990 . .. Graduate Administrative Associate, University Libraries, the Ohio State University. 1990-1991 .. Graduate Research Associate, Mershon Center, the Ohio State University. MAJOR FIELD: Political Science
iii TABLE OF CONTENTS ACKNOWLEDGMENTS ii VITA iii LIST OF TABLES vi LIST OF FIGURES xii CHAPTER PAGE I. INTRODUCTION 1 A. Introduction 1 B. Objective 3 C. Study Organization 4 II. ETHNIC DIVERSITY AND VIOLENT POLITICAL BEHAVIOR 6 A. Wilsonian/Value Perspective 8 B. Psychological Perspective 11 C. Preliminary Model 14 III. INTERVENING VARIABLES 19
A. Economic Development (Ia) 21 B. Economic and Political Discrepancies (I2) 24 C. Governmental Coercion (I ) 27 D. Role of Ethnic Elites (IJ 29 E. Geographic Concentration of Ethnic Groups (I ) 32 F. External Links of Ethnic Groups (I6) 34 G. Process of Modernization (I7) 40 H. Favorable International Conditions (I8) 45 I. Historical Explanations (I9) 48 IV. TOWARD A MODEL OF VIOLENT POLITICAL BEHAVIOR 52 A. Revised Model 52 B. Testable Hypotheses 55 C. Methods and Sample 59 V. OPERATIONALIZATION AND DATA SOURCES 62 A. Defining Ethnic Groups 62
iv B. Ethnic Diversity 66 C. Violent Political Behavior 74 D. Intervening Variables 76 E. A Word on Aggregated Data 83 F. Univariate Statistics 87 VI. EMPIRICAL TESTS OF THE MODEL 91 A. Correlation Analysis 91 B. Bivariate Cross-tabulation 91 C. Multivariate Cross-tabulation 101 D. Bivariate and Multiple Regressions 117 VII. SUMMARY AND CONCLUSIONS 156 A. Summary of the Study 156 B. Findings ad Prospects 159 LIST OF REFERENCES 169 APPENDIXES A. Lists of Countries Included and Excluded in the Study 189 B. Sources of Indicators of Ethnic Diversity 194 C. Components of the Variables 197 D. Original Data Before Manipulations 198 E. Data After Manipulations 217 F. Cross-tabulation.and Regression Results 224 G. Statistical Results When the Indirect Measure of Governmental Coercion Is Used 263 H. Statistical Results With the Quadratic Forms of Economic Development and Governmental Coercion 270 I. Pearson Correlation Coefficients Among All Variables Including the Interaction Terms 273 J. Values for Each Variable in the Ascending Order 276
v LIST OF TABLES PAGE Univariate Statistics of All Variables 90 Pearson Correlation Coefficients Among All Variables 92 Cross-Tabulation of Ethnic Diversity with Riot 95 Cross-Tabulation of Ethnic Diversity with Armed Attack 96 Cross-Tabulation of Ethnic Diversity with Assassination 97 Cross-Tabulation of Ethnic Diversity with Death from Violence 98 Cross-Tabulation of Economic Development with Riot 225 Cross-Tabulation of Economic Development with Armed Attack 225 Cross-Tabulation of Economic Development with Assassination 226 Cross-Tabulation of Economic Development with Death from Violence 226 Cross-Tabulation of Economic and Political Discrepancies with Riot 227 Cross-Tabulation of Economic and Political Discrepancies with Armed Attack 227 Cross-Tabulation of Economic and Political Discrepancies with Assassination 228 Cross-Tabulation of Economic and Political Discrepancies with Death from Violence 228 Cross-Tabulation of Governmental Coercion vi with Riot 229 Cross-Tabulation of Governmental Coercion with Armed Attack 229 Cross-Tabulation of Governmental Coercion with Assassination 230 Cross-Tabulation of Governmental Coercion with Death from Violence 230 Cross-Tabulation of Role of Ethnic Elites with Riot 231 Cross-Tabulation of Role of Ethnic Elites with Armed Attack 231 Cross-Tabulation of Role of Ethnic Elites with Assassination 232 Cross-Tabulation of Role of Ethnic Elites with Death from Violence 232 Cross-Tabulation of Geographic Concentration with Riot 233 Cross-Tabulation of Geographic Concentration with Armed Attack 233 Cross-Tabulation of Geographic Concentration with Assassination 234 Cross-Tabulation of Geographic Concentration with Death from Violence 234 Cross-Tabulation of External Links with Riot 235 Cross-Tabulation of External Links with Armed Attack 235 Cross-Tabulation of External Links with Assassination 236 Cross-Tabulation of External Links with Death from Violence 236 Summary of the Cross-Tabulations 100 Cross-Tabulation of Ethnic Diversity with Riot Controlling for Economic Development 237
vii Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Economic Development 237 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Economic Development 238 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Economic Development 238 Cross-Tabulation of Ethnic Diversity with Riot Controlling for Economic and Political Discrepancies 239 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Economic and Political Discrepancies 239 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Economic and Political Discrepancies 240 Cross-Tabulation of Ethnic Diversity with Death from violence Controlling for Economic and Political Discrepancies 240 Cross-Tabulation of Ethnic Diversity with Riot Controlling for Governmental Coercion 241 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Governmental Coercion 241 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Governmental Coercion 242 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Governmental Coercion 242 Cross-Tabulation of Ethnic Diversity with Riot Controlling for Role of Ethnic Elites 243 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Role of Ethnic Elites 243 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Role of viii Ethnic Elites 244 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Role of Ethnic Elites 244 Cross-Tabulation of Ethnic Diversity with Riot Controlling for Geographic Concentration 245 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Geographic Concentration 245 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Geographic Concentration 246 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Geographic Concentration 246 Cross-Tabulation of Ethnic Diversity with Riot Controlling for External Links 247 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for External Links 247 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for External Links 248 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for External Links 248 Summary of the Cross-Tabulations 112 Regression Results of Violent Political Behavior on Ethnic Diversity 249 Regression Results of Violent Political Behavior on Ethnic Diversity in Quadratic Form 249 Regression Results of Violent Political Behavior on Ethnic Diversity and Economic Development 251 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Economic Development 251 Regression Results of Violent Political Behavior ix on Ethnic Diversity and Economic and Political Discrepancies 253 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Economic and Political Discrepancies 254 Regression Results of Violent Political Behavior on Ethnic Diversity and Governmental Coercion 255 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Governmental Coercion 256 Regression Results of Violent Political Behavior on Ethnic Diversity and Role of Elites 257 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Role of Elites 258 Regression Results of Violent Political Behavior on Ethnic Diversity and Geographic Concentration 259 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Geographic Concentration 260 Regression Results of Violent Political Behavior on Ethnic Diversity and External Links 261 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with External Links 262 Summary of the Regression Results 141 Cross-Tabulation of Governmental Coercion (Indirect Measure) with Riot 264 Cross-Tabulation of Governmental Coercion (Indirect Measure) with Armed Attack 264 Cross-Tabulation of Governmental Coercion (Indirect Measure) with Assassination 265 Cross-Tabulation of Governmental Coercion (Indirect Measure) with Death from Violence 265
x Cross-Tabulation of Ethnic Diversity with Riot Controlling for Governmental Coercion (Indirect Measure) 266 Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Governmental Coercion (Indirect Measure) 266 Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Governmental Coercion (Indirect Measure) 267 Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Governmental Coercion (Indirect Measure) 267 Regression Results of Violent Political Behavioron Ethnic Diversity and Governmental Coercion (Indirect Measure) 268 Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Governmental Coercion (Indirect Measure) 269 Regression Results of Violent Political Behavior on Ethnic Diversity, Economic Development, and Economic Development in Quadratic Form 271 Regression Results of Violent Political Behavior on Ethnic Diversity, Governmental Coercion, and Governmetal Coercion in Quadratic Form 272 Regression Results of Violent Political Bahavior on Ethnic Diversity and All Intervening Variables 151 Regression Results of Violent Political Bahavior on Ethnic Diversity, All Intervening Variables, and All Interaction Terms 152 Pearson Correlation Coefficients Among All Variables Including the Interaction Terms 274
xi LIST OF FIGURES FIGURES PAGE 1. Preliminary Model 15 2. Revised Model 53 3. Working Model 58 4. Final Model of Riot from Cross-tabulation 113 5. Final Model of Armed Attack from Cross-tabulation 114 6. Final Model of Assassination from Cross-tabulation 115 7. Final Model of Death from Violence from Cross-tabulation 116 8. Final Model of Riot from Regression 143 9. Final Model of Armed Attack from Regression 144 10. Final Model of Assassination from Regression 145 11. Final Model of Death from Violence from Regression 146 12. Final Model from Cross-tabulations 149 13. Final Model from Regressions 150
xii CHAPTER I INTRODUCTION
A. Introduction We are used to the popular misperception that the modern world is made up of "nation-states." The boundaries of sovereign states and those of ethnic groups are, however, rarely congruent in the real world. As Connor (1972: p. 320) notes, less than 10% of the 107 states surveyed qualified as ethnically homogeneous. And according to the study of Nielsson (1985: pp. 30-31), only 45 out of the 164 political units under investigation fell into the category of "single nation-group states," where, as he defines, the size of one ethnic group constitutes more than 95% of the population of the state its members reside in.
The failure of plural societies to achieve political consensus is generally explained in terms of their linguistic, religious or racial differences. Politicized ethnic sentiments are invariably perceived to be one of the most disruptive forces conducive to violent political behavior. A* Kurian (1979: p. 45) observes, "more civil wars have been fought in Asia and Africa in modern times on
1 2 issue of race, language or religion than on that of political ideology." Still, ethnic conflict is by no means a "tribal" phenomenon confined to the so-called Third World countries that came into existence after World War II. Even in the Western World, where nationalist leaders elsewhere are inclined to seek the classic paragon of "nation-states," ethnic strife resulting from politicized ethnic nationalism can not be avoided. The Flemish, the Scots, the Welsh, the Bretons, the Corsicans, the Basques, the Catalans, and the Quebequois are cases in hand (Allardt, 1979; and Esman, 1977). It is no wonder there exists a widely shared belief that ethnic diversity is to be blamed for most incidents of violent political behavior.
However widely shared the common sense or folk contention may be or however intuitively obvious it may be, it may be at times deceptive and illusory. In order to test such an assertion in a more rigorous manner, we must specify the ways in which differences in the degree of ethnic diversity as a national attribute are linked to violent political behavior. How has such a national attribute trait as manifested by ethnic diversity, which is the configuration of the objective cultural distinctions within a polity, been transformed into subjective ethnic consciousness, successfully politically mobilized, and subsequently led to violent political behavior? More 3 specifically, what are the contingent conditions that would stipulate the relationship between the degree of ethnic diversity and the level of violent political behavior?
B. Objective Although the quantity of literature on ethnic relations is overwhelming, there have been surprisingly few attempts to provide a general theory and subject it to empirical test. And most discussions of ethnic relations have been limited to case studies of one state or comparisons of a few states1. Seldom has any systematic cross-national study of ethnic relationships in a global scale been suggested to test this thesis. As Horowitz (1985: p. xi) puts it,
"there is, in the main, too much knowledge and not enough understanding, too much evidence chasing after too few categories."
Furthermore, former studies are divided and inconclusive. We thus believe that the presumed relationships may be uncovered by a more sophisticated model tested by a more extensive data set. Our empirical study i3 the first to test such an assertion globally within a general theoretical framework. In order to answer the above questions in a more
x. There are some exceptions, such as the studies by Fishman and Solano (1990; and 1989), whose main deficiency, however, is lack of efforts at theory-building. 4 rigorous manner, a general framework is needed. We thus shall introduce a model of violent political behavior, explained mainly by ethnic diversity. It is hypothesized that there exists a positive relationship between the degree of ethnic diversity, the fundamental causal agent, and the level of violent political behavior, the dependent variable to be explained. Our fundamental proposition is thus that violent political behavior arises directly in magnitude with the degree of ethnic diversity. Nevertheless, as it is generally agreed that single- cause deterministic explanations are inadequately barren, ethnic diversity as a structural characteristic does not suffice to explain violent political behavior. As Geertz (1963: p. 124) rightly points out, "simple primordial sentiment is no more defensible a position than economic determinism." We would expect that the above relationship is invariably intermediated by some other variables. An adequate account of the genesis of violent political behavior thus demands the consideration of these factors.
C. Study Organization We would thus start with an inventory of some major factors and a probing sketch of their relationships. Once relevant variables and theories have been reviewed and investigated, we will propose a general conceptual framework of violent political behavior. The hypotheses to be tested 5 will then be developed from the discussions of those factors that we have already identified. These proposed hypotheses are then statistically estimated against empirical data. CHAPTER II ETHNIC DIVERSITY AND VIOLENT POLITICAL BEHAVIOR
There have been numerous attempts at explaining the upsurge, or resurgence, of ethnic nationalism after World War II. While some of them are specific to the time, the geographical region, or the issue, others are more general and abiding. In explaining the somewhat curious rise of ethnic nationalism in the Western World in the 1960s and 1970s, Connor1 (1977: pp. 23-24), for instance, lists no less than four popular explanations, in addition to his own historical one: (1) economic, political, or cultural relative deprivation, (2) psychological alienation resulting from cultural anomie in modern society, (3) the center-periphery relation, and (4) the erosion of the state in the present international context. Reviewing the literature on ethnic conflict in the same area, Lijphart (1977: pp. 55-64) proposes eight speculative explanations to be tested.
*. Earlier (1972: p. 332), he had delineated some other favorable conditions conducive to ethnic awareness: (1) the favorable international environment, (2) the spread of the appealing principle of self-determination, (3) successful precedents, and (4) the awareness of these factors as a result of the progress of communications.
6 7 Zolberg (1977: p. 35) also provides us three secondary variables that have contributed to the escalation of ethnic conflict there. Still more studies of ethnic conflict in other areas have been conducted (Young, 1976; Horowitz, 1985; and Esman and Rabinovich; 1988). Actually, a long inventory list of these myriad variables and theories may be compiled. A major drawback of merely listing the explanatory variables is that it fails to discriminate their relative weights. Without doubt, all the explanations or theories offered may shed some light on the nature of the subject matter concerned here. Still, we may find ourselves confused by the various interconnected variables.
For an understanding of the relationship between ethnic diversity and violent political behavior, we need a conceptual framework derived from these explanatory variables2. The first task of this study is to review these explanations by sorting out these variables and by evaluating their relative weights in a synthesizing text rightly seeking balance and completeness. That the literature lacks evidence of the theoretical
2. We notice that the often-used term "factor" is too vague to offer any explanatory utility than that of common sense. A distinction between primary and secondary factors is at least warranted. A satisfactory general perspective should at least assign relative weights to these factors and sort them out into causes, intervening variables, and facilitating factors. See Wai (1978) for a discussion of these distinctions. 8 direct linkage between ethnic diversity and violent political behavior is surprisingly true. Most writers have been content with implicit rather than explicit theoretical foundations of the causal mechanisms between ethnic diversity and violent political behavior, although there is a huge body of the literature on the probable causes of violent political behavior3. We have yet to move beyond the common sense prediction that a country with more ethnic groups is by nature inescapably characterized by violent political behavior. Broadly speaking, two nascent schools may be identified: Wilsonian/Value, and Psychological perspectives.
A. Wilsonian/Value Perspective
Wilsonian (Utopian) A Wilsonian/Utopian perspective, which can be traced back to John Stuart Mill, argues that ethnically homogeneous states tend to be more democratic (Claude, 1955: p. 81). For Mill (1958: p. 23):
"Free institutions are next to impossible in a country made up of different nationalities. Among a people without fellow-feeling, especially if they read and speak different languages, the united public opinion, necessary to the working of representative government, cannot exist."
3. For a thorough review of this body of literature, see Sanders (1981: pp. 1-20). 9
Taking the line of argument one step further, proponents of nationalism in the 19th century, arguing for self- determination and redrawing contemporary state boundaries, claimed that the establishment of nation-states was conducive to peace (Suhrke and Nobel: 1977: pp. 9-10). The major tenet of this line of argument is that ethnically homogeneous nation-states are the preconditions to peace. And the other side of the coin suggests that states made of heterogeneous ethnic groups are inevitably prone to communal conflict. Since ethnic homogeneity is neither a sufficient nor a necessary condition for democracy or peace4, this view is more an ideological exhortation than scientific explanation. In practice, for those ethnic elites, the ideology is that their group happiness can only be guaranteed in a state made up of all the members of their own nation. A revised version of the above view offers a better, though not satisfactory, explanation of the cause of violent political behavior. Theories of integration posit that cultural homogeneity is a prerequisite of political integration (Deutsch, 1966; and Kuper and Smith, 1969). For instance, Jacob and Teune (1964: p. 19) suggest that ethnic homogeneity is positively related to political integration
4. Here, peace is understood in a narrow sense as absence of direct violence, rather than structural violence. 10 and consequently positively related to domestic political stability. Almond and Verba (1965: p. 28) also generalize that England, the old Common Wealth, the US, and the Scandinavian countries were relatively stable because of their homogeneous political culture. Lijphart (1968: p. 31) similarly suggests that the stability found in the centripetal democracy is based on a homogeneous culture5. All these observations explicitly suggest that homogeneity is directly related to political stability and that ethnic diversity is inevitably detrimental. In fact, the deterministic argument that ethnic diversity is related to violent political behavior, although seemingly plausible, can not be hastily inferred from the above array of arguments.
Value Perspective A second school that provides a more satisfactory line of explanation may be termed a value approach. Davies (1969) contends that the underlying cause of political conflict is disagreement over political or social values among participants in the society. Bringing this argument to bear the question of ethnic diversity, it is argued that "irreconcilable value differences" may exist because of people's different, if not incompatible, ethnic backgrounds
5. It is noted that Lijphart (1969: pp. 207-25) later contends that ethnic homogeneity is not necessary for political stability. 11 (Nielsson, 1985: p. 45). This line of arguments share the following common assumptions: Conflicts arise from disagreement over value systems; people can only live peacefully with their own co- ethnics; and the existence of ethnic diversity provides the opportunity for violent political behavior. Why? The value perspective, being less fragmented and incomplete, explains it away with different values possessed by members of disparate ethnic groups. Gurr (1966: p. 67) is more specific: If the excluded or discriminated groups share with the dominant groups their value systems, there may not be any sense of deprivation. Under such a circumstance, conflict may be avoided. However, he points out, such acceptance of discrimination is unlikely to be accepted in the twentieth century. The study by Stavrinides (1976: pp. 5-10) illustrates how opposing self-images and attitudes and how two competing value systems has been one of the causes of conflict in Cyprus.
B. Psychological Perspective A psychological school provides a somewhat different but more sophisticated explanation of the causal linkage. Proponents of this school contend that domestic conflict is the result of dissatisfaction and dissent. Frustration could be the result of the relative deprivation of political power, economic interests, or cultural pride, such as the 12 persistent perception of cultural inferiority forcefully imposed by the dominant ethnic groups. It is noted that frustration is not necessary the result of unequal distribution of wealth among the composite ethnic groups or political subjugation of ethnic group by the other. Take Cyprus for example: it was the unfulfilled aspiration of unification with mainland Greek on the part of Greek Cypriots that had led to communal conflict, although Greek Cypriots were found to share a feeling of inferiority to Turkish Cypriots as a result of Ottoman Turkish colonization. Since their plea to the United Nations for self-determination could not be solved politically, conflict was the consequence of chronic deprivation and frustration.
The most prevalent theories in this school are Davies's J-curve theory (1962), Gurr's theory of relative deprivation (1968), and Feierabend and Feierbend's theory of frustration-deprivation (1966; and 1972). Whether in the form of Davies's discontent, Gurr's systematic frustration, or Feierabend and Feierabend's relative deprivation, the basic tenet of the psychological school is that it is the gap or discrepancy between aspiration and achievement that leads to revolution, aggression, or other forms of violent political behavior.
We must point out that the psychological perspective has its weakness when coming to cross-national empirical application. Breuilly (1982: p. 32), for instance, points 13 out the difficulty in testing such a psychological state in general. In their critical review of recent studies in the frustration/deprivation school, Finkle and Rule (1986: p. 56) have gone so far as to judge that they are "inherently flawed." In practice, however satisfactory they may be for students of ethnic relations in explaining the psychological mechanism of violent conflict behavior, they are not very empirically satisfactory instruments since we are unable to aggregate individual deprivation/frustration into a group frustration measure to be applied for subsequent cross- national studies6. Therefore, this intermediary stage, though retained in the revise model, will not be shown in the working model developed later.
Still, possible direct causal linkages have yet to be scrutinized although the above psychological explanation may indirectly support the hypothesis that ethnic diversity is conducive to violent political behavior. In other words, the above theories only explain, in the minimal sense, why deprivation/frustration would lead to violent political behavior. More immediate causes of deprivation/frustration must be present, such as inegalitarian ethnic distribution
6. Cooper (1974: p. 414) uses the measures of economic discrimination and of political discrimination provided by Gurr (1972: p. 191) to tap the concept of deprivation potential. See also Gurr (1968: pp. 1109-12) for various indirect measures of persistent deprivation. Still, the concept of realized deprivation is not directly measured. 14 of resources or power, which will be discussed in the following chapter.
C. Preliminary Model The preliminary model as shown in Figure l7 is based on the earlier discussions in this chapter. While ethnic diversity (X) is presumed the primary cause of violent political behavior (Y), eight intervening variables are considered: level of economic development (IJ, degree of
economic and political discrepancies (I2), governmental coercion (I3), role of ethnic elites (IJ, geographical concentration of ethnic groups (I5), external links of
ethnic groups (I6), process of modernization (I7), favorable
international conditions (I8), and historical explanations
(I9). In our conception, ethnicity begins with the identification of differences in such ascriptive attributes as national origin, race, language, religion, culture, and even historical experience. Without any slightest such distinctions, it would not be necessary to discuss ethnic conflicts. Rather, it may be conflicts between classes or between the center and the peripheries, although these conflicts may at times reinforce ethnic ones. Thus, our main generalization in this study is
7. For reason of easier recognition of the independent variables in the figure, each variable has been assigned an abbreviation. 15
Intervening Variables
Economic Economic and Governmental Development Political Coercion Discrepancies
Intervening Variablas
FIGURE 1. Preliminary Model 16 basically a structural one: It is the structure that leads to human behavior (Reed, 1990: p. 335). One of the major instrumental utilities of cultural differences is their use as an indispensable basis for political mobilization. Linguistic difference, for example, is one of such factors that tend to make a plural society conflict prone. As Urwin (1983: p. 242) suggests, Scottish nationalism resulting from economic change must ground itself on the ethnic base. Otherwise, ethnic mobilization would be impossible. Conflicts pertaining to linguistic differences may be more difficult to resolve, as ethnic elites may utilize linguistic distinction to rally ethnic identity. A famous phrase goes, "a state may become blind but not deaf" (Zolberg, 1977: p. 39). Not surprisingly, as McAllister and Mughan (1984: p. 241) discover in their analysis, Welsh- speaking, for instance, is the most important "social structural" variable of the Welsh nationalist movement. Urwin (1983: p. 233) even claims that language and ethnic identity are interchangeable.
The main thesis is thus that in a country where members of the ethnic group(s) perceive a higher level of deprivation/ frustration, violent political behavior is more likely to take place. At this point, we may arrive at our main thesis: It is logical to infer that, other things being equal, a country with a higher degree of ethnic diversity is more likely to have a higher level of violent political 17 behavior. In other words, we would anticipate that ethnic diversity and the degree of violent political behavior is positively and linearly related. Or more specifically, ethnic diversity increases the probability of violent political behavior. Hence, we have:
Hypothesis 1: The degree of ethnic diversity and the level of violent political behavior are positively and linearly related.
Nevertheless, not all students of ethnic politics agree with the presumed detrimental impacts of ethnic diversity. For instance, Wallerstein (1971: p. 669), analyzing the confrontation between tribalism and nationalism in Africa, concludes that the existence of ethnic groups may facilitate national integration by providing various functions that the state could not offer, especially in the transitional period. Smith (1983: p. 70) discovers that in the cases where ethnic cleavages and class conflict cut across each other, ethnic ties may lessen class conflict and therefore help to strengthen the authority of the state.
The most challenging anti-thesis is brought out by Coleman (1960: p. 368), who contends that the multiplicity of tribes within a state is not necessarily an obstacle to nation-building. On the contrary, he claims, the larger the number of tribes, the higher the chance for effective 18 amalgamation. Tanzania is such an extreme case that has escaped ethnic turmoil, since no tribe is dominant enough to monopolize political power and economic wealth. Horowitz (1985: pp. 37-40) also notices that the presence of many scatter ethnic groups, none of which is large enough to monopolize the power center, may actually facilitate inter- group collaboration At this point, we need to modify our original hypothesis and test whether there exists a curvilinear relationship between the degree of ethnic diversity and the level of violent political behavior. Hence:
Hypothesis 1.1: There exists a curvilinear (inverse U- shaped) relationship between the degree of ethnic diversity and the level of violent political behavior. CHAPTER III INTERVENING VARIABLES
Once again, it cannot be emphasized too much that ethnic diversity is but a necessary condition and by no means sufficient condition for ethnic consciousness. In other words, ethnic diversity may contribute to violent political behavior, but not necessarily. This is readily evident since ethnic awareness entails a subjective identification with the ethnic group. What, then, are the factors that set violent political behavior in motion? This involves exploring the nature of possible intervening variables.
As Shabad and Gunther (1982: pp. 453-54) conclude from their case study of regionalism in Spain, linguistic cleavage may be related to ethnic nationalism but is not its "cause." Rather, we may properly interpret it as a catalyst. Connor (1972: pp. 336-37; and 1984: p. 342) distinguishes between two dimensions of ethnic nationalism: the overt characteristics of cultural uniqueness and the psychological essence. He notes that tangible distinctions are usually mistaken as the explanatory variables because it is much easier to observe them and because they are actually
19 20 used by ethnic leaders for mobilization. Further, even if cultural attributes have been successfully distilled into ethnic loyalty or have been used as the basis of inequality, it is hypothesized, it may be still away from politicization, which demands the existence of such catalysts as economic discrepancies or unequal distribution of political power to intensify ethnic mobilization. Although cultural differences may constitute potential bases for political expression, this function is not deterministic. Thus, ethnic diversity or homogeneity alone as a structural factor does not suffice to explain variable degrees of violent political behavior across states.
From the psychological perspective, if there is a correlation between ethnic diversity and the occurrence of violent political behavior, it depends on the presence of whatever additional intervening variables required to account for the feeling of deprivation/frustration. Something other than ethnic diversity is thus being introduced to produce an analytically useful framework. The next step is to examine which variables may intervene in the hypothesized relationship between ethnic diversity and violent political behavior, by either reinforcing or inhibiting any violence potential.
More specifically, the role an intervening variable plays in the relationship between the degree of ethnic 21 diversity and the level of violent political behavior could be in the form of interpretation, explanation (spuriousness1), distortion, or suppression (Babbie, 1979: pp. 450-62). The following variables seem to intervene: level of economic development, degree of economic and political discrepancies (unequal distribution of political power and/or economic resources), governmental coercion, role of ethnic elites, geographic concentration of ethnic groups, external links of ethnic groups, process of modernization, and historical explanation.
A. Economic Development (I1) Increasing economic development is generally believed to have the function of alleviating the antagonism that ethnic diversity tends to generate. As Eichenberg et al. (1984: p. 17) postulate:
"developed societies are better able to meet material demands and thus to quiet conflict or prevent it from arising in the first place."
With the expansion of the economy, the government may exercise its economic instruments to "gain new supporters, to reinforce old ones, and to weaken opponents of the system." (Ake, 1974: p. 583). In other words, conflicts
1. A spurious relationship between variables X and Y is one that is explained away when a third variable Z is introduced. See Babbie (1979: p. 452). 22 abate as wealth increases. This line of argument may be applied to the group level. Coleman and Rosberg (1964: pp. 690-691) suggest that social and economic development will reduce the disruptive potential of tribal or regional separatism2. Conversely, during periods of economic recession, ethnic groups in distress, while competing for scarce resources, are less willing to make any cultural concessions (Mughan, 1985: pp. 295-97). It is thus inferred that ethnic conflicts tend to worsen when economic performance is poor. Taking an opposite position, Olson (1963: p. 533) submits that rapid economic growth is a source of conflict, which creates a class of "'nouveaus pauvres' who are much resentful of their poverty than those who have known nothing else." Taking a more determinative perspective, Wiener and Hoselitz (1961: p. 177) argue that economic development is bound to lead to conflict: no matter how the distributive capacity of the government may increase, some groups would invariably feel dissatisfied, which would then increase the possibility for conflict. Although focusing on different factors, they share the view that there is a positive relation between the rate of economic development and the
2. Some students of ethnic conflict maintain that economic buy-off tends to be useless as long as political demands are not fulfilled, which challenges the thesis that the relationship between level of economic development and ethnic conflict is secondary as long as political aspirations are not fulfilled. See Urwin (1985: p. 165). 23 frequency of violent political behavior. For the purpose of empirical testing, it is thus hypothesized that in a country with a lower level of economic development, ethnic diversity will be more likely to produce violent political behavior than it would be in a country with a higher level of economic development. In other words, there is a combined, or interaction, influence of economic development and ethnic diversity on violent political behavior. Accordingly, we have:
Hypothesis 2: In a country with a lower level of economic development, ethnic diversity is more conducive to violent political behavior than in one with a higher level of economic development.
Apparently combining the above two seemingly opposing perspectives, Gurr and Duvall (1973: p. 141) propose that there exists a curvilinear relation between the level of economic development and stress, which is linked to political conflict. Stated more directly, economic development would intensify political conflict in early stages and lessen antagonism in latter stages. It is suggested that there is a curvilinear (inverse U-shaped) relationship between the level of economic development and the level of violent politcal behavior. However, since the concentration of this study is to discern the interaction relationships among the independent, dependent, and intervening variables rather than the relationships between the intervening and dependent variables, we shall not formally list the hypothesis regarding the curvilinear relationship. Instead, we shall test it and report the result in Appendix H.
B. Economic and Political Discrepancies (I2) Here, two types of ethnic inequality are under scrutiny: political and economic discrepancies among ethnic groups. Imbalance of power is generally recognized as potentially destabilizing (Bagley, 1972: p. 348) An ethnic group having more access to political power or economic resources in disproportion to its group size would definitely block the upward mobility of other ethnic groups. As Inglehart and Woodward (1967: p. 28) note:
"The likelihood that linguistic division will lead to political conflict is particularly great when the language cleavages are linked with the presence of a dominant group which blocks the social mobility of members of a subordinate group, partly, at least, on the basis of language factors."
Therefore, language differences are politically divisive in so far as the elites use them to unite the individuals for political actions in improving the political or economic status of the ethnic group as a whole (p. 35). Empirical studies of the relation between group discrimination and 25 domestic conflict have consistently shown that the relation is strong (Gurr and Gurr, 1983: p. 51, especially note 7). However, Gurr and Gurr (1983: p. 51) take a more balanced stand: "Social cleavages and ethno-linguistic diversity are necessary for group discrimination, but one is not directly a function of the other." They contend: "Societies that are homogeneous with respect to ethnicity and religion offer less opportunity for group discrimination. ..." The line of economic discrepancy explanation3, be it in the form of Hecter's (1975) internal colonialism or Nairn's (1977) uneven development, basically emphasizes that ethnic tensions are the result of uneven level of economic development. As a catalyst, economic discrepancies may exacerbate the preexisting perception of cultural distinction. In the case of the French communities in Canada, the original concern may have been over minority rights. But it was later translated into the concern over economic inequality only after the inception of industrialization (Fenwick, 1981: p. 204).
As the case of Belgium has illustrated, Walloon nationalism did start with economic discontent although Flemish nationalism may have originated from cultural discontent. The difference between the Canadian and the
3. See Grove (1974: pp. 305-7) for a nice comparison of the Marxist/radical internal colonial model and the pluralist model. 26 Belgian cases is that linguistic conflict now takes on a new economic dimension. The above transformation demonstrates that manifestations of ethnic discontent may not only be manifold but also highly fluid and malleable, that is, manipulable. On the contrary, Connor (1984: p. 345) argues that even though economic discrepancies may exacerbate ethnic conflict, regions with a higher standard of living may also nourish ethnic discontent. He gives us the examples of the Basques and the Catalans as opposed to the Castilians, of the Croats and the Slovenians to the Serbs, and of Flanders to Walloons in the 1960s (1977: p. 37). Urwin (1983: p. 241) suggests that economically superior regions, such as Flanders, Catalonia, the Basque region, and Scotland, would have more resources to challenge the powerful political center. Murphy (1988: pp. 139-40) holds a similar stand. In the 19th century, both Wallonia and Flanders had similar economic positions. It is noted that in the post-war period, when Flemish nationalism was on the upsurge, Flanders did actually receive a disproportional amount of investment.
Accordingly, the above line of argument suggests that there may have been other causes of ethnic nationalism and that economic factors only exacerbated it. While Walloon nationalism may be exacerbated by economic decline, ethnic nationalism in the Basque region, Catalonia, and Scotland 27 are actually related to economic improvement. Accordingly, it is economic change generally, in addition to economic inequality, that may act as catalyst of ethnic nationalism. For the sake of empirical test, we arrive at:
Hypothesis 3: In a country where there is a higher degree of economic and political discrepancies, ethnic diversity is more conducive to violent political behavior than in a country with a lower degree of economic discrepancies.
C. Governmental Coercion (I3) Ascriptive ethnic differences may be complicated — strengthened or weakened — by the policy adopted by the centralizing state, which in turn plays a crucial role in deciding the rise and the wane of ethnic nationalism and delimiting its content and even strength. There is a continuum of governmental reactions to ethnic nationalism4. On the one end is genocide or expulsion of the minority groups, if the center decides that the group is not desirable. On the other extreme, the state may allow outright secession of the ethnic group from the status quo state. If it is decided that the ethnic group is to remain in the territory, the decision will be whether to induce, or
4. See Claude (1955) especially Chapters 7 and 8 for the various measures. 28 to impose, assimilation of the minority, or whether to permit cultural and/or political autonomy. In all cases, arrangements would require prudent policy-making on the part of the government. The degree of governmental coercion comes into the picture here. Feierabend and Feierabend (1966: p. 25) point out that punishment and coerciveness of political regimes would inhibit aggressive behaviors. Gurr (1968: p. 1105) also suggests the inhibiting effect of punishment or coercion, actual or threatened. These observations lead us to reach the following:
Hypothesis 4: In a country with a low degree of governmental coercion, ethnic diversity will be more likely to produce violent political behavior than it would be in a country with a high degree of governmental coercion.
However, the interaction relationship among ethnic diversity, violent political behavior, and governmental coercion may be the opposite. According to the observation of Iran's ethnic unrest by Khalizad (1984-85: p. 671), it was the effort at nation-building by Reza Khan that had provoked local revolts against the central government in Tehran. The harsher the punitive measures adopted by the government are, the more violent the ethnic resistance is. Therefore, there may exist a curvilinear relationship between the degree of governmental coercion and the level of violent political behavior. However, since our focus is to ascertain the existence of any interaction relationship among the independent, dependent, and intervening variables, we shall not formally propose any hypothesis regarding the expected inverse U-shaped relationship between governmental coercion and violent political behavior. Nonetheless, the results of the tests will be presented in Appendix H.
D. Role of Ethnic Elites (I4) As ethnic elites are more exposed to outside influence, they cannot escape the idea of nationhood. In the absence of leading ethnic elites, the mass largely remain indifferent to ethnic nationalism. Further, ethnic sentiment may be less salient at the mass level than at the elite level, as the mass generally lacks interest in political issues. As Greenwood (cited in Urwin 1983: p. 25) has observed, "ethnicity is highly malleable and responsive to the circumstances in which groups find themselves." We may thus posit that ethnic conflict is the result of the mobilizing efforts of elites.
As Urwin (1983: p. 226) points out, the concern with ethnic identity may be universal, but the support of nationalist movements is not. Shabad and Gunther (1982: pp. 454-56) similarly observe that the elite may intensify ethnic conflict by articulating their agenda, which may in 30 turn strengthen the ethnic loyalty. In the case of Belgium, the mass showed little interest in linguistic issues even when they had grown prominent in elite competition. Tiryakan (1980: p. 10) also observes that those who vote for the nationalist party in Scotland, Quebec, and Wales are fewer than those who share the nationalist sentiment. As the Case of Cyprus has vividly demonstrated, the elites of the two communal groups had never encouraged their followers to pursue the goal of national integration. Rather, their loyalties had been to Greece and Turkey respectively. For the majority Greek Cypriot community, enosis (union with Greek) was implied by the British for their cooperation in the war. On the other hand, the prospective Greek domination had made the minority Turkish Cypriots antagonist to the idea of enosis. The Turkish Cypriots welcomed the continued British colonial rather than immediate independence if the goal of partition could not be obtained. Their mutual suspicion and hostility were spread to the grass-roots and finally contributed to the breakdown of the constitution and flare-up of ethnic conflict three years after the declaration of independence in 1960 (Salin, 1978: pp. 5-6; and Joseph, 1985: pp. 55-56).
But why are elites, more willing to take the lead? In a way, it may be due to their perceived relative deprivation. See (1986: pp. 3-4) finds two functions of ethnic sentiment: a primordial one that may fulfill the psychological need for 31 collectivity, and an instrumental one that utilizes the collectivity to advance material interests. She suggests that the former works more for the mass while the latter works more for the elite. Taking an instrumental perspective5, Inglehart and Woodward (1967: p. 21) observe that "ambitious members" of the subordinate ethnic groups may perceive "the opportunity to make careers for themselves by fanning a large potential group into consciousness of its separate identity. ..." Therefore, calculating from the rational choice perspective, ethnic elites expect to ascend to leaders of the nascent ethnic groups which have heretofore not been mobilized. Covell (1985) similarly describes how elites in Belgium made use of consociational mechanisms to protect their personal interests.
Taking a primodial perspective, Smith (1986: p. 46) argues that the real concern of the elite are cultural and social relations and that economic and political interests are only used for rational justification. Geertz (1963) also emphasizes the indispensable essence of the primodial ties that lies behind competition between ethnic elites.
We argue that all elements may be involved. As Barth (1991: pp. 9-10) justly makes the judgement: The strategic choices made by the ethnic elites are confined by the value
5. See also Bell (1975: pp. 169-71) and Barth (1985) for this school of reasoning. 32 system inherently shared by members of the ethnic group. Therefore it is fruitless for us to portray the elite as exclusively utilitarian or non-materialistic. The ethnic elites in the peripheries may manipulate ethnic sentiment in their competition with the national elites for political power and economic resources by persuading the mass that their interests are ethnic ones. In addition, they may also come to the lead out of primordial sentiments. According to the above discussions, we thus would anticipate:
Hypothesis 5: in a country where ethnic elites play a more active role in ethnic mobilization, ethnic diversity is more conducive to violent political behavior than in one where ethnic elites play a less active role in ethnic mobilization.
E. Geographic Concentration of Ethnic Groups (I5) One spatial facilitating factor is whether the ethnic group is geographically concentrated or interspersed. In other words, we need to juxtapose cultural distinction with territory in order to duly assess the impacts of ethnic diversity on violent political behavior. Urwin (1983: p. 239) observes that groups not geographically concentrated would be difficult to mobilize. One such example at hand is the Swedish Finns, who are scattered in separate parts around the country. 33 On the other hand, when the ethnic group is geographically concentrated, that is, where there is a congruence between the ethnic identity and the homeland, such as the cases of Scotland, Wales, and the two Belgian regions, it would be easier for ethnic elites to undertake political mobilization, especially for the cause of separatism. Furthermore, it would be more difficult for the central government to assimilate these geographically concentrated ethnic groups. To borrow the words of Inglehart and Woodward (1967: p. 29), when the assimilating attempts look "unreasonable" in the eyes of members of the minority groups, political resistance is more likely. Overseas Chinese in Southeast Asia, who mostly reside in urban areas, also fall into this category. The Karens in Burma, the Moros in the Philippines, and the Malays in Thailand all fare the same in their power wrestling with the central government (Leng, 1983: p. 215). And when members of other ethnic groups intrude upon the ethnic homeland, ethnic hostility may come about more easily. Under such a condition, ethnic antagonism would make politicization of ethnic awareness much easier. We thus have:
Hypothesis 6: ethnic diversity is more conducive to violent political behavior in countries where there are more geographically concentrated ethnic groups than in countries where there are fewer or no such ethnic groups. 34
F. External Links of Ethnic Groups (I6) Another facilitating factor is external links of the ethnic groups concerned (Gurr, 1968: pp. 1114-15), which are almost ignored by students of ethnic conflict. The most commonly seen picture is when the homeland of a border ethnic group is separated into two or more countries. The Kurds, the Baluchis, and the Pathans are familiar cases (Phadnis, 1984: p. 201). According to Nielsson (1985: pp. 34-39), 42% of the ethnic groups in his investigation (248 out of 589) are dispersed over several states. Taking an ethnic-centric perspective, Nielsson (1985: p. 33) finds that there are 1357 ethnic units if we count those ethnic groups separated in several countries as separate units.
Oftentimes, an oppressed ethnic group may seek support from their co-ethnics living across the borders, which in turn may exacerbate the domestic conflict (Said and Simmons, 1976; Bartelsen, 1977; Suhrke and Noble, 1977; and Shields, 1984). The ethnic conflict in Cyprus, Sri Lanka, Northern Ireland, and Israel are the most prominent cases. To better understand the possible causal linkage between ethnic diversity and international conflict, we need to scrutinize the interactions among the parties involved in the conflicts. And a comprehensive model of ethnic conflict should include relevant external factors. We propose two dimensions for facilitating the following analyses: (1) the initiator of the spill-over of ethnic conflict, and (2) the structure of ethnic diversity in terms of the ethnic relationships among the three possible parties.
Initiators The first dimension is identified by the initiator who attempts to gain from the spill-over of internal ethnic conflicts into the international arena. Three parties are identified here: the first party is the host state (or status quo state); the second party is the ethnic group(s); and the third party is the neighboring state(s). Since what we present here are ideal types, multiple initiators are possible, which will make the picture more complicated. We shall begin with the discussion of the ethnic group, then the neighboring state, and finally the host state.
As interest groups, dissatisfied ethnic groups are easily mobilized to seek alternative processes outside the state boundary in order to pursue goals that cannot be reached otherwise (Said and Simmons, 1976: p. 38). By widening the scope of the conflict, they may identify potential allies (Bartelsen, 1977: p. 3). The terrorist means adopted by the PLO is meant to enlist the sympathy if not assistance from the neighboring Arab states.
By linking internal and international conflicts, ethnic groups may actively manipulate foreign policy and may be passively manipulated by the host-state and the neighboring state (Jensen, 1982: p. 138). Greek lobbying activities in 36 American foreign policy in the Cyprus crisis is a good example of active manipulation of foreign policy by an ethnic group. On the other hand, the Kurds between Iraq and Iran are the best example of how an ethnic group may be played by the host and neighboring state as pawn in their conflict. Conversely, the escalation of domestic ethnic disputes may also start from the neighboring state. There is no end of the list of the manipulation of ethnic conflict in one state by another state. In a modest form of intervention, Albania gave verbal encouragement of its kin in Yugoslavia. But many cases of meddling take the form of providing sanctuary or arms, or of taking outright military action.
Domestic ethnic conflicts tend to draw external intervention from neighboring states for all sorts of reasons. Suhrke and Noble (1977) differentiate two type of external intervention in domestic ethnic conflict: (1) affective intervention, based on cross-boundary ethnic ties, and (2) instrumental intervention in the absence of such ties.
The neighboring state may provide material assistance to its oppressed brethren out of humanistic consideration or seek unification of the territory occupied by that kin group. But the state involved needs not be geographically adjacent to the status quo state. Nor is the assistance necessarily in any material form. As General de Gaulle 37 spoke his famous slogan "vive le Quebec Libre" in Montreal in 1967, no other form of foreign assistance could have helped more to arouse the ethnic sentiment among the Quebequois. Thus Suhrke and Noble (1977) posit that affective interventions are more prone to violent conflict and that instrumental involvements are more apt to mediating. Shields (1984: pp. 11-12) also hypothesizes that ethnic separatists are more likely to get aid from neighboring states if the states are controlled by their ethnic kin. Domestic ethnic conflict may offer traditional enemies opportunity and incentive for intervention, as the dissident ethnic group may be perceived as a convenient instrument for disrupting the internal order of another state. The Indian intervention in the Pakistani civil war in 1971 is a successful example. But India itself is not immune from external manipulation of its complicated ethnic composition: It has complained that the Nagas were trained and armed by China. In neighboring Burma, it was complained that China harbored Kachin dissidents and other ethnic secessionists. We may posit that the more diverse the ethnic composition is, the more opportunity for enemies to take the advantage of ethnic conflict.
Nevertheless, foreign state involvements are not invariably on the negative side of the ethnic grove. Said and Simmons (1978: p. 29) find that another state may lend 38 covert or overt assistance to the state besieged with ethnic conflict. For instance, France sold armory to Spain for cracking down on the Basque dissidents harbored in the border areas. Finally the initiator may also be the government of the host state, which is termed as the diversion hypothesis (Jensen, 1982: pp. 60-61): elites engaged in external conflicts in order to divert attention from domestic difficulties. Therefore, it is not necessarily the ethnic group that hopes to gain from the escalation of internal war. Bhutto's hostility to India was to defuse the demand of independence by East Pakistan; Sukarno's dispute with Malaya was intended to divert domestic attention to external conflict. In this regard, ethnic diversity may be treated as the cause of external conflicts.
Ethnic Configuration The other dimension is the structure of ethnic diversity judging from the relation of the ethnic group with the host and neighboring states. Three types of ethnic configuration are delineated. The first type is characterized by a minority group in a host state and its cross-boundary ties, which creates ethnic diversity in the host state. The ethnic group may seek assistance from its kin ruling in other states or even seek unification. The neighboring state may also seek unification with the ethnic 39 group and thus dispute with the host state. International conflicts here are the result of irredentist movements. Somalia's disputes with Ethiopia and Kenya can be found in its desire to recover Great Somalia. This type of ethnic configuration seems most conducive to internationalize domestic ethnic conflict. The second form of ethnic configuration involves a stateless ethnic group whose home land is divided by two or more states. Ethnic diversity can thus be found in both states. The ethnic group may fight against both states but may also contribute to cooperation between them. It may also be utilized as a tool by states against one another. In the last case, where international conflict may be present, domestic conflict is not the cause but one facet of the international conflict.
The third configuration is composed of the host state and a third state with no ethnic tie with the ethic group. The ethnic group may seek external help from any third party; and the third may also use the ethnic group as a tool. Accordingly, the possibility of spill-over of domestic conflict is decided by the willingness of the ethnic group to spread the conflict and by the relationship between the first and the third parties.
We may find that an attribute of ethnic diversity renders a state more vulnerable to ethnic dispute in the contemporary world, where self-determination is well 40 accepted through demonstration effect and the viability of small states is higher than it used to be. From the above discussion, we may suggest that ethnic diversity is generally conducive to internal and international conflict as it provides all three parties more opportunity to spread or to exploit domestic conflict. We would thus reach:
Hypothesis 7: In countries where ethnic groups have more external ties, ethnic diversity is more conducive to violent political behavior than in countries where ethnic groups (or group) have (or has) fewer external ties.
H. Process of Modernization (I7) The process of modernization includes several interrelated sub-processes: industrialization, urbanization, standard education, the progress of communication and transportation facilities (Deutsch, 1953; Lerner, 1959). There is little agreement as to whether the process of modernization serves as an integrative or disruptive force in the formation of ethnic nationalism and in nation- building6. While some argue that modernization would promote assimilation and facilitate integration, others argue that modernization would increase ethnic consciousness. Up to this point, it is argued that
6. For a nice theoretical treatment of the destabilizing effects of the process of modernization in general, see the cogent review by Ake (1974). 41 modernization would increase inter-group contacts through effective communications. The focus now turns to whether increased contacts will dissipate or increase ethnic loyalty. The structural school7 deals with how strains and tensions, arising in the process of modernization, may disrupt domestic tranquility. The key concept here is social mobilization, which is defined by Deutsch (1971: pp. 385-86) as:
"the process in which major clusters of social, economic, and psychological commitments are eroded or • broken and people become available for new patterns of socialization and behavior."
Social mobilization, while entailing the disruption of individuals, provides them opportunities to acquire new values and expectations. Nevertheless, it must be accompanied by appropriate economic development and political institutionalization in order to avoid social frustration and hence violent political behavior.
According to Deutsch's theory of modernization, increased contacts resulting from social mobilization would
7. This is different from the structural perspective of ethnic mobilization proposed by people such as Hector (1986). 42 encourage national assimilation8. He thus optimistically recommends a deliberate assimilation policy to eradicate ethnic consciousness: "As soon as we begin to make the [assimilation] process visible, we are beginning to change it." (1966: p. 164). It is noted that there are two processes concerned here: contact and assimilation. We would deal with the process of contact first. Taking an opposite stand, Connor (1977: p. 29) argues that contacts among ethnic groups could only reinforce their ethnic awareness, and in turn exacerbate ethnic conflict. He shows us the relationships between the Basques and the Castilians, between the Walloons and the Flemish, between the Welsh and the English, and between the French-speaking and the English-speaking Canadians, where increased contacts all led to increased ethnic conflict. Taking the case of Indonesia for example, Esman (1976: p. 416) notices that the urban areas tend to become battle fields for communally based migrants, where competition for scarce economic resources, such as employment, housing and education opportunities, and struggle for political influence would reinforce ethnic antagonism. On the other hand, the lack of ethnic strife in Switzerland is attributed to its homogeneous cantons, where inter-group contacts are limited,
8. Connor (1972: pp. 323-25) notices that Deutsch has swung between two opposite positions as to the causal effects of modernization on integration. For the sake of simplicity, we only consider this position here. in addition to its decentralized nature of decision-making (Connor, 1976: p. 124). In our view, intergroup contacts per se will not lead to ethnic strife. Rather, conflict may be due to competition for resources or due to cultural anomie or psychological alienation in industrialized society. In the past, ethnic groups would retain their separate lifestyles and cultural isolation. But nowadays, modernization threatens their lifestyles and deprives them of their unique culture. Alienation is especially prominent among the dislocated ethnic elite within an urbanized environment. This is confirmed by Urwin (1983: p. 235), who argues that political mobilization by the ethnic party was more successful among the middle class in Wales, Scotland, Belgium, and Spain.
The second facet of Deutsch's formulation of modernization is assimilation. Most political leaders and laymen seem convinced by the "law" that it is the assumed ill impacts of ethnic diversity that have led to violent political behavior. Accordingly, forced acculturation is deemed necessary for avoiding disruptive conflicts. Oftentimes, attempts at assimilating ethnic groups only magnifies the latter's ethnic consciousness and even draws negative responses. Connor (1973: p. 20) distinguishes between assimilation of overt characteristics and that of the psychological essence. Although objective cultural distinctions may be assimilated in the process of modernization, the affective essence, which is non- observable, cannot be assimilated. Thus, he goes on, any programmed assimilation policy is doomed to fail and arouse resistance. Even with generations of acculturation, he suggests that assimilation may be reversed (1972: p. 34). On both contact and assimilation, Deutsch and Connor sit on the opposite sides of the theoretical fence. To counter Connor's argument, Fenwick (1981) provides a sequential view of assimilation: primordial loyalties linger because of lack of intergroup contacts from segregation. Thus, ethnic conflict may arise at the earlier stage of modernization. But as soon as the process of modernization accelerates, intragroup ties will be broken up, and new types of loyalties, such as class or occupation, will take their place. He thus posits that ethnic conflict is actually the result of lack of contact, and that modernization will increase group contacts and lead to assimilation.
Mughan (1979) presents a different version of the modified theory of modernization: modernization may exacerbate ethnic conflict in the beginning, but will dissipate ethnic loyalties in the long run. Here, while insisting that contacts will eventually lead to assimilation, he argues that ethnic conflicts result because of insufficient inter-group contacts. Lijphart (1977: p. 45 48) also attempts to qualify the modernization theory: modernization may have an assimilative effect in the earlier stage, when local identities will be mobilized into ethnic consciousness. However, in the latter stage of development, attempts at assimilating ethnic groups will lead to conflict. Although taking a similar sequential perspective of modernization, Mughan (1979) and Lijphart (1977) have opposite opinions as to whether ethnic diversity would lead to ethnic conflict at the early or later stage of modernization. Up till now, we are unable to arrive at a testable hypothesis as to how the process of modernization would mediate the relationship between the degree of ethnic diversity and the level of violent political behavior. There is no marked consensus among the scholars with regard to the relationship between the process of modernization and the occurrence of violent political behavior because they have collected conflicting evidence in their analysis of ethnic conflict. Because of this theoretical ambiguity and empirical complexity, we must conclude that we have failed to delineate, in a propositional fashion, a relationship which could lend itself to hypothesis testing. Thus, we have decided to exclude the process of modernization from our model. 46
H. Favorable International Conditions (I8) Other facilitating conditions for the upsurge of ethnic nationalism in the post-war period may be broadly classified as favorable international conditions. Among these, the most important one is the change in the international political environment, in which there is no longer a clear relation between size of the state and military security and thus small and military weak states may survive (Birch, 1978: p. 335). As the formation of regional economic organizations, such as the Common Market, has contributed to the erosion of state sovereignty and the economic viability of minor political entities, ethnic groups do not foresee any benefits from membership in the larger state.
The second factor is the movement to democracy beginning in the 19th century. For instance, the electoral reforms that allowed for universal suffrage in the period of 1867-1885 in Great Britain had made possible the expression of ethnic sentiment and sped up the growth of nationalism in Ireland, Scotland and Wales (Berrington, 1985: p. 174). The third contextual variable is the diffusion of the idea of popular sovereignty and the principle of national self-determination as justification (Ryan, 1988: pp. 168- 69). Initially, the principle was applied only to the defeated multi-ethnic states after World War I; and it was finally endorsed by the United Nations after World War II. Armed with the doctrine of self-determination, ethnic groups 47 are increasingly unwilling to accept the "cast-iron grid," to borrow the term from Young (1976), imposed by the modern state system. Being more daring to challenge the existing state, ethnic elites are more resistant to compromise with the central government on ethnic issues. The last condition is the demonstration effect of the successful precedents of new states created in this century. The first wave of new states came with the nationalist movements in Eastern Europe, which ended with the break-up of the Austro-Hungarian and the Ottoman Empires and the multiplication of states in that region. The second wave of nationalist models came from the new independent states in the Third World immediately after World War II. Having been inspired by these successful cases, members of ethnic groups are convinced that nothing other than total independence, or at least autonomy, is acceptable (Connor, 1970: p. 96). At the same time, most of the leaders of those states, being besieged with ethnic phenomena, are reluctant to tolerate the appeals of ethnic groups (Coleman and Rosberg, 1964: p. 663). The ruling elite is determined to retain the contemporary state system and not to further break up the present territorial arrangement. Since the legitimacy of the principle of self-determination can not be refuted easily, they tend to refuse the existence of ethnic problems or outright reject the existence of ethnic diversity (Ryan, 1988: p. 169). Consequently, frustration is the result of unfulfilled aspiration of ethnic groups for autonomy or independence. Conflicts between competing nationalist sentiments seem inevitable. Ethnic diversity is invariably perceived by them as disruptive, if not pathological; and politicized ethnic sentiment (or nationalism) is generally perceived to be one of the most disruptive forces conducive to violent political behavior. We note that these favorable international conditions are constant rather than variables for all the states. That is, there are in the nature of contextual conditions. We thus decide to drop this intervening variable out of the model.
I. Historical Explanations (Ig) There are various strands of historical explanation of ethnic nationalism, with somewhat variant emphases. Connor (1977: pp. 25-27) traces the origin of ethnic nationalism in the Spanish Basque provinces, Catalonia, and Scotland, back to the time of the French Revolution, when popular sovereignty was for the first time connected with ethnicity to justify national self-determination. His view of the process of ethnic movement is sequential and evolutionary. And since the process of self-determination has never been completed so far, he asserts that their resurgence after World War II is natural and predictable. For him, any 49 country that "falls short of actual nation-statehood would become suspect to a segment or segments of its population" (p. 25). With only few countries being ethnically homogeneous in our days9, he thus predicts the inevitable plague of ethnic turbulence for most countries. His view is basically deterministic, since he claims that the other factors, including the process of modernization, are only catalytic. Any deterministic view would fail to explain the variations in degree of ethnic awareness and in that of mobilization across countries. Taking a similar historical perspective, Smith (1986: p. 51) further distinguishes between underlying structural conditions and immediate precipitating factors of ethnic nationalism. For him, any explanation that lays emphasis on the latter, such as economic factors, is short-sighted and misleading. Rather, he maintains, the inner dynamics lies in the intersection of the following two movements starting in the last century: the movement of social and political emancipation, and the quest for personal and collective identity. Thus, the post-war ethnic nationalism is nothing but the continuity of classic nationalism.
A version of historical determinism portrays the
9. According to Connor (1972: 320), only 10% of the states studied are homogeneous in terms of ethnic composition. A more recent study by Nielsson (1987: 20) reveals that only 28 out of the contemporary 161 states are composed of "monostate nation." 50 transformation of ethnic nationalism as a dialectic process. Da Silva (1977: p. 99) deems the Basque nationalism as the legacy of the attempt of the Spanish liberals to centralize the state in the last century. Adopting a similar view, Zolberg (1977: pp. 26-27) perceives recent separatist movements as a result of the dialectic interaction between two trends: the incomplete process of nation-building and the striving for ethnic regionalism. When the elites of the two processes encounter each other, tensions are inevitable. Thus, he views the rise of ethnic separatism as the result of the provocative efforts at nation-building from the center. Then, why did these separatist movements surge only after World War II? Zolberg (1977: p. 33) contends that the decline of tension between the Flemings and the Walloons in the inter-war period was due to the deception of the authoritarian corporative regime. He similarly sees French ethnic nationalism as the resumption of "normal politics," which was interrupted by World War II, rather than any reversal (p. 31). Once the interruptive element was removed, usually in the form of governmental coercion, the quarrels would resume immediately.
Nevertheless, being more or less deterministic, these seemingly plausible views do not explain much of the complicated picture of competing ethnic mobilization. At best, they may be subsumed by the variables we have 51 discussed earlier, such as economic discrepancies or governmental coercion. Consequently, no satisfactory hypothesis seems to be in order here. CHAPTER IV TOWARD A MODEL OF VIOLENT POLITICAL BEHAVIOR
A. Revised Model Now, we may present our revised model by way of synthesis in Figure 2. It starts with the existence of ethnic diversity. From the Wilsonian/Value perspective, ethnic diversity is expected to conduce to violent political behavior directly; and from the psychological perspective, however, it must be followed by a feeling of psychological deprivation/frustration perceived by those who share those objective manifestations of cultural distinctions, the mass and the elite as well.
At the structural level, cultural differences are capable of forming ethnic nationalism and developing into political conflict provided that they are reinforced by ethnic mobilization processes that solidify the groups internally and promote friction among them. Hence, ethnic diversity cannot be the sufficient cause for ethnic mobilization, much less for violent political behavior. In other words, there is no guarantee that distinct cultural attributes will lend themselves to deprivation/frustration. Without any form of psychological deprivation/frustration,
52 53
Intervening Variables
Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites
Ethnic Diversity
Intervening Variables
FIGURE 2. Revised Model 54 be it economic, political, or cultural, subjective ethnic awareness and the concern with the group can hardly emerge. Four primary intervening variables are observed to have the tendency to contribute to the reinforcement or alleviation of the level of deprivation/frustration: level of economic development (Ix), degree of economic and political discrepancies (I2), governmental coercion (I3), and role of ethnic elites (I4). While a higher level of economic development may alleviate the feeling of deprivation/ frustration, economic and political discrepancies could exacerbate it, and government coercion would suppress that feeling. On the one hand, ethnic elites in the peripheries, or of the subjugated ethnic groups, may feel deprived by the policy undertaken by the central government. On the other hand, they may create or reinforce the degree of deprivation felt by the mass.
Two secondary intervening factors are found to have useful explanatory utility: geographic concentration of ethnic groups (I5) and external links of ethnic groups (I6). While the former may facilitate ethnic mobilization in the ethnic homeland, the latter may provide the opportunity of cross-border ethnic solidarity, whether initiated by the ethnic group or by the neighboring country. Either would reinforce the possibility that ethnic diversity would lead to violent political behavior though neither causal link has to follow the deprivation/frustration route. 55 As for the process of modernization, we failed to arrive at any testable hypothesis since there is no agreement as to their impacts on the relation between ethnic diversity and violent political behavior. Favorable international conditions justifying the causes of the ethnic groups would create a certain degree of expectation. As a result of the gap between the real life and the expectation, a feeling of relative deprivation is formulated. As we have noted earlier, these conditions are shared by all human beings, provided that they enjoy the same level of information. Since they are constants rather than variables for every country in the world, it is more appropriate to treat them as the contextual conditions rather than any causal or facilitating factors. In reality, the explanatory factor is useful only if we undertake a historical comparison of changes in these conditions over time. We thus decided to omit this variable from the model.
As we have mentioned earlier, the various brands of historical explanation is either deterministic or can be subsumed by the intervening variables identified earlier, we decide to leave them out of the model.
B. Testable Hypotheses From the discussions of relevant theories and variables earlier, we may derive the following hypotheses: 56 Hypothesis 1: The degree of ethnic diversity and the level of violent political behavior is positively and linearly related.
Hypothesis•1-1: There exists a curvilinear (inverse U- shaped) relationship between the degree of ethnic diversity and the level of violent political behavior.
Hypothesis 2: In a country with a lower level of economic development, ethnic diversity is more conducive to violent political behavior than in one with a higher level of economic development.
Hypothesis 3: In a country with a higher degree of economic discrepancies, ethnic diversity is more conducive to violent political behavior than in one with a lower degree of economic discrepancies.
Hypothesis 4: In a country with a lower degree of governmental coercion, ethnic diversity is more conducive to violent political behavior than in one with a higher degree of governmental coercion.
Hypothesis 5: In a country where ethnic elites play a more active role in ethnic mobilization, ethnic diversity is more conducive to violent political behavior than in one where ethnic elites play a less active role in ethnic mobilization.
Hypothesis 6: In a country where there are more geographically concentrated ethnic groups, ethnic diversity is more conducive to violent political behavior than in one with fewer such ethnic groups.
Hypothesis 7: In a country where ethnic groups have more external ties, ethnic diversity is more conducive to violent political behavior than in one with fewer such external ties.
As we have pointed out earlier, the psychological perspective has its weakness when coming to cross-national empirical application. Since we are unable to aggregate individual deprivation/frustration into a group frustration measure to be applied for subsequent cross-national studies1. Therefore, this intermediary stage, though retained in the revised model, will not be shown in the working model (Figure 3).
*. Cooper (1974: p. 414) uses the measures of economic discrimination and of political discrimination provided by Gurr (1972: p. 191) to tap the concept of deprivation potential. See also Gurr (1968: pp. 1109-12) for various indirect measures of persistent deprivation. Still, the concept of realized deprivation is not directly measured. 58
Intervening Variables
Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites
Ethnic Diversity
Intervening Variables
FIGDRB 3. Working Hodol 59 G. Methods and Sample The primary methods of analysis employed in this study are cross-tabulation, and bivariate and multivariate regressions while Pearson correlation coefficients will also be evaluated for the preliminary probe. Although more sophisticated techniques, such as causal analysis2 or confirmatory factor analysis3, may also be attempted to estimate the whole model, we decide to use the simpler methods for they are adequate in assessing the interaction relationships embedded in our hypotheses among the independent variable (ethnic diversity), the dependent variable (violent political behavior), and the intervening variables. The other merit of these methods is that they permit simple and lucid interpretation of the test results, whereas results from the complex manipulations mandated by those complicated techniques tend to be obscure and remote from our original hypotheses.
Our unit of analysis is states rather than ethnic groups. Since this is intended to be a universally cross-
2. This technique involves the estimation of direct and indirect relationship between the variables in an one-way causation (recursive) system or in a reciprocal causation (non-recursive) system that is made up of simultaneous equations. See Blalock (1971), Asher (1983), Berry (1984), and Pedhazur (1982) for details. 3. The method is useful in analyzing the linear structural relationships (LISREL) among variables that cannot be measured directly through the construction of a covariance structural morel. See Long (1983a, and 1983b) and Stein (1976) for an introductory treatise on this method. 60 national study, we would like to include as many states as possible if indicators for the major variables are available, especially those pertaining to ethnic diversity and violent political behavior. As of October 1988, there were 170 countries in our universe (Europa World Year Book, 1989: pp. 3-4 )4. In all, the sample of the study consists of only 132 states. Our attempts at formulating a good indicator of ethnic diversity is limited by datum availability. While the data of the dependent variables are available for 156 polities, those of ethnic diversity are available for 136 polities. We decided to adopt the latter country list. Because we decided that colonies, dependencies, and overseas territories are not included, Hong Kong and Puerto Rico are to be excluded. Since both South Vietnam and North Vietnam have ceased to exist as separate polities since 1975, we thus decide to omit them even though the ethnic measure is available for them5. All counted, we have 132 countries in our sample6, which would warrant enough variations of the
4. Although enjoying a separate membership in United Nations, the Byelorussian SSR and the Ukrainian SSR are not considered independent countries. The unified Germany and the unified Yemen are treated as two separate political units here, while Namibia, although gaining its independence in 1990, is excluded because of lack of data. 5. These is no ethnic datum for Vietnam as a unified sovereign state. 6. See Appendix A for the lists of countries included and excluded in the study. 61 indicators of all variables. By so doing, we have virtually disqualified countries with one or more of the following characteristics: (1) mini- states with the population less than 100,000, such as St. Christopher and Nevis in the Caribbean, such as Nauru, Kiribati, and Tuvalu in the Pacific, and such as Seychelles in the Indian Ocean; (2) principalities and city states, such as Andorra, Liechtenstein, and Monaco; and (3) those countries gaining their independence in the 1970s or the 1980s, such as Bahamas, Bahrain, Belize, Djibouti, and Vietnam. Nevertheless, there are some exceptions: both Equatorial Guineas and Swaziland, although gaining their independence in 1968, are omitted because of lack of ethnic data. On the other hand, three countries gaining independence in the 1970s are included because all data are available for them as a political unit: Angola (1976), Mozambique (1975), and Papua New Guinea (1975). CHAPTER V OPERATIONALIZATION AND DATA SOURCES A. Defining Ethnic Groups Before we define the concept of ethnic diversity, detailed discussions of the concept of ethnicity1 are needed because vague descriptive concept has limited analytic and comparative use. In this section, we would like to make an excursion and find out what definitions of ethnic group have been offered by scholars in this field. Ethnicity is a relatively new term. While the 1931 edition of the Encyclopedia of the Social Sciences (Murdock, 1931) lists an entry of "ethnic communities," the term "ethnic group" is not defined until the 1968 edition of the International Encyclopedia of the Social Sciences (Morris, 1968). The usage of ethnic group has evolved over the years as interests in and research on ethnic relationships expanded. Its current usage has been restricted to social divisions resulting from culture, while, in contrast, racial group is used to represent biological differences, especially the Black-White one. This layman demarcation is too vague to be of use since cultural and biological
1. See Van den Berge (1967: p. 142) for the conceptualization and Grove (1973: p. 303) for the discussion of the distinction between ethnicity and race. 62 63 heritages are in essence inter-twined (van den Berghe, 1967: p. 10). We have identified four groups of definition of ethnic group: purist, objective, subjective, and objective- subjective:
Purist Definition Snyder (1983) is the lone proponent of a purist definition of ethnic group. Tracing the origin of ethnicity back to the Greek word "ethos," or nation, he insists that ethnic group be restricted to those resulting from racial differences. In its strictest form, therefore, the term may only be applied to the case of racial relationships such as those in the United States and South Africa. By defining it this way, Snyder has limited the utility of the term in terms of scope. His intention to avoid the contamination of the term by other uses than the original one is understandable. Language, however, is evolving so we can not avoid constantly assigning new meanings to the same term (see Riggs, 1986). His effort appears to gain few echoes (except van den Haag, 1973; and Wolf, 1986) judging by the fact that ethnicity is generally believed to result from cultural differences rather than racial ones.
Objective Definition Most scholars agree that there are some objective 64 criteria for classifying ethnic groups, such as culture, language, religion, history (or common experience), nationality, or even race. In this inclusive fashion, group division is based both on cultural and biological differences, but exclude economic ones. Nevertheless, there is no consensus as to which criteria are necessary for the identification of ethnic groups. We do not think this is a real issue, however, since not all characteristics of ethnic cleavage are equally identifiable or of the same importance in every country. In other words, objective criteria are contingent to countries under investigation. Even if any agreement can be reached, the presence of all objective criteria cannot guarantee sufficiently the existence of ethnic division. This type of definition is therefore criticized as "built on sand" (Young, 1976: p. 49).
Subjective Definition The opposite position to the above one is the emphasis of the subjective self-identification shared by people belonging to the same group. Weber (1968: p. 389) thus defines ethnic groups as:
"those human groups that entertain a subjective belief in their common descent because of similarities of physical type or customs or both or because of memories of colonization and migration" 65 Stack (1981: p. 18) similarly2 defines ethnicity as a "subjective identity that clearly distinguishes between group members and outsiders." This approach is adopted to redress the futility found in the former definition in identifying group differentiation. Still, the verstehn fashion is not devoid of any criticism. One flaw is the difficulty in defining ethnic group readily and objectively owing to its subjective nature. The strongest opposition to this way of defining ethnic group comes from those who attempt to quantify the phenomena of ethnicity. For them, it amounts to chaotic anarchy and is hence unacceptable.
Objective-Subjective Definition A compromising definition of ethnic group in the middle ground seems more feasible: a subjective belief based on objective criteria, be they real or putative (Schermerhorn, 1970: p. 12). Most scholars agree that both objective and subjective components of ethnic group are indispensable. As objective criteria cannot of themselves lead to self- consciousness among members of the group, so cannot the subjective "we-you" differentiation arise without being built on discernable objective criteria. A close examination of Weber's definition reveals that the subjective belief has to be built on certain objective
2. See also Sinnott and Davis (1981: p. 398) for a similar view. conditions. Put in another way, objective criteria of ethnic group are the bases of ethnic consciousness; however, it further needs certain catalyst to make group members perceive their distinctions to others. According to the presence of subjective consciousness among group members, van Haegendoren (1982: p. 49) differentiates between sleeping ethnic group and ethnic group, Jackson (1984: p. 207) does between ethnic category and ethnic group, and Yinger (1983: p. 395) does between ethnicity and full ethnicity. As Brass (1976: p. 226) rightly observes, these differences are in degree rather than in kind.
B. Ethnic Diversity3 In this study, we are inclined to define ethnic group in such a broad sense as to include linguistic, religious, and racial attributes since the subjective facet of ethnic consciousness is not obtainable. In this study, we use the measure of ethno-linguistic fractionaliation of Taylor and Jodice (1983, ICPSR codebook: p. 72) combined with a measure of religious fractionalization constructed by us. The task turns to the question of what the meaning of diversity is. By diversity, we would, at the first brush, presume that the larger the number of ethnic groups within a
3. See Appendix C for the detailed documentation of the derivation of the data used in this study. 67 country, the more ethnically diverse it is. However, the conceptualization of diversity by simply counting the number of ethnic group, although possessing some face validity, has one deficiency: it fails to tap other dimensions of diversity, such as the size of each ethnic group. By so doing, we would assume that countries with the same number of ethnic groups will have the same degree of ethnic diversity. Clearly, it is not the case in the real world. For example, a country composed of three ethnic groups with the same size is by intuition more diverse than one with one dominant and two minority ethnic groups.
Suppose there are three countries composed of three ethnic groups. The proportions of the three groups for them are .33, .33, and .33 for country A, .10, .10, and .80 for country B, and .05, .05, and .90 for. country C. We tentatively use the following formula to calculate the degree of ethnic diversity:
n n2 (1) F = 1 - S i=l-3 N2,
where n± is the size of the ith ethnic group and N is the total population of the country. Thus, we have .67 for country A, .34 for country B, and .185 for country C. From the above illustration, we thus decide that an indicator of ethnic diversity with higher construct validity should take into account both the number and the sizes4 of the ethnic groups. In other words, both the number and the sizes of ethnic groups jointly determine the degree of ethnic diversity5. Most measures concerning ethnic diversity are crude, incomplete or area specific6. For instance, the four indicators of ethnic diversity employed by Barrows (1975) are available for 32 Black African countries; and the 1-4 measure of cultural pluralism by Anderson et al. (1967: pp. 80-82) are available for 84 countries only. The measures of religious, racial, and linguistic homogeneity designed by Banks and Texter (1963), available for 100 countries or so, are dichotomous and hence fail to show relative degree of ethnic diversity. The 0-15 index of ethnic/racial pluralism by Grove (1974: p. 314), available for 113 countries, only reveals modal frequency, that is, the proportion of the largest ethnic group. An index of pluralism constructed by Haug (1967), which exceptionally takes language, race, religion and sectionalism into account, is handicapped by
4. See also Speed (1970: p. 37) for the importance to consider the group size. 5. See the discussion in Sani and Sartori (1983), Sartori (1976), Rae and Taylor (1970), Guseman et al. (1976), Lieberson and Hansan (1974) and Lieberson and O'Connor (1975) for general discussions; and consult Greenberg, 1956: p.109) for different ways of defining ethnicity, such as the "Monolingual nonweighted method." 6. See Appendix B for detailed comparison of some measures and Grove's (1974: p. 220) comparison of different measures. 69 the limited number of countries (n=114). On the other hand, Nielsson (1985: pp. 30-33) collapsed 161 sovereign states and 3 dependent territories into five groups according to the proportion of the three largest ethnic groups. Although an ordinal measure of ethnic fragmentation with the scale from 1 to 5 can be constructed from his categories, we would like to find an interval one. As a result, two more comprehensive measures are left to our choice: Rustow's (1967: pp. 284-86) measure of linguistic unity and Taylor and Hudson's (1972) and Taylor and Jodice's (1983) measures of ethnic-linguistic fractionalization7. Rustow's linguistic unity only shows the percentage of the two largest linguistic groups. Taking into account the percentage of all linguistic groups, the indicator of fractionalization provided by Taylor and Hudson (1972: p. 216) is found to possess the highest construct validity. The following formula is given (p. 216):
n (nj (nt - 1) (2) F1 = 1 - S * i=l (N) (N - 1),
where N is the country's total population and ni is the
7. See also Kurian (1979: pp. 44-46; 1984: pp. 47-49) for a related measure of ethnic and linguistic homogeneity in the 1960-65 period. 70 population of its ith ethnic or linguistic group8. According to the data provided in Atlas Narodov Mira, by Roberts (1962), and by Muller (1964), three highly correlated measures of ethnic and linguistic fractionalization with slightly different observations are derived. Accordingly, we decide to the use as part of our measure ethnic diversity Taylor and Jodice's ethno- linguistic fractionalization (1983, ICPSR codebook: p. 72), which is mainly based on Atlas Narodov Mira augmented by the data provided by Roberts (1962) and by Muller (1964). As noted by Taylor and Hudson (1972: p. 215), Atlas Narodov Mira assumes that language and ethnic origin are closely related and thus does not distinguish one from the other.
The second component of our measure of ethnic diversity is religious fractionalization. One deficiency of the above measure is that it fails to consider groups with different religious allegiance, national origins or races9. To
8. A similar measure may be constructed based on the most recent and much more detailed data provided by Decsy (1988). One potential threat to reliability of the measure is the double count of the linguistic groups in some countries: his grouping is not always mutually exclusive for the data are compiled from disparate sources in different years. The list of second-language speakers is another problem. It is also doubtful whether to include so many linguistics groups recognized by linguists. 9. Racial differences are more important for the Latin American countries. We may calculate racial diversity for the 20 latin American countries based on the data provided by Anderson et al. (1967: p. 46) It is noted that Taylor and Hudson's measure of ethnic-linguistic fractionalization 71 augment that measure, we need to construct a measure of religious fractionalization based on the proportions of the Christian community and of Muslim community provided by Taylor and Hudson (1972: pp. 217-18, and pp. 275-81). We decide to use the following simple formula:
n n.2 (3). Fr = 1 - S --- i=l-3 N2,
where N is the country's total population and nL (A = 1, 2, or 3) is the proportion of the population of its ith religious groups: Christians, Muslims, and others. Taylor and Hudson (1972: pp. 275-81) provide the proportion of Christians in 136 political units around 1964 and that of the Muslims in 135 political units around I96010. Additional information regarding the percentage of the Muslims is derived from Kurian (1979: p. 48; and 1984: pp. 51-53), Demographic Yearbook (various years), Weigert et al. (1957: pp. 412-15), Weekes (1984), World Almanac and Book of Facts (various years), and World Christian Encyclopedia (Barrett, 1982). only adjusts that for the United States. 10. Kurian (1979: pp. 46-49) provides the percentage of the Christians in 135 countries for the mid-1970s and the percentage of the Muslims in 67 countries for the early 1970s, which are similar to those in Taylor and Hudson. 72 By deducing the percentages of the Christians and of the Muslims from 1, we arrive at the percentage of the adherents of other religions, such as Buddhism, Judaism, Hinduism, Animism, and traditional beliefs. Except Judaism11, we are unable to disaggregate the compositions of these religions that are amenable to cross-national comparison12. To make sure we have not lumped together two or more religions under this category, we examine the composition of those countries where the percentage constitutes more than half of the population by consulting Demographic Yearbook (various years), Weigert et al. (1957: pp. 412-15), Weekes (1984), World Almanac and Book of Facts (various years), and World Christian Encyclopedia (Barrett, 1982). We find that Eastern Asian and Southeast Asian countries fall under this category, where the size of Buddhists makes up of more than three quarters of the total population, for instance, Burma (92%), Japan (99%), Laos (99%). No further breakdown within the Buddhism seems warranted. Similarly, Hindus constitute 87% of the Indian population and over 50% in Nepal, and are
11. Kurian (1979:p. 49; and 1984: pp. 53-54) has provided the distribution of the Jewish population in 87 and 85 countries respectively, ranging from 30 in Cyprus to 5,800,00 in the US in 1977, and from 20 in Libya to 5,920,890 in the US as of 1980. 12. Although detailed religious compositions in mid- 1970, mid-1975, and mid-1980 are available in World Christian Encyclopedia (Barrett, 1982), the data for the 1960s are not reported there. 73 the largest religious group in Guyana, Fiji, and Mauritius (Kurian: 1979: p. 44). And Jews make up 89% of the Israeli population. In African countries, it is usually the case that around half of the population practice traditional beliefs, such as Gambia (50%), Ivory Coast (60%), and Zimbabwe (64%). Consequently, our measure largely taps the three dimensions of religious fragmentation. The only country that defies our three-fold classification is Sri Lanka, where the religious cleavages cut between the Buddhists and Hindus. For the sake of consistence, we decide not to adjust for its religious composition.
Therefore, the final measure of ethnic diversity is the mathematical average of Taylor and Jodice's ethno-linguistic fractionalization and our own religious fractionalization13. Hence.
(4) F = (F1 + Fr)/2.
We note that this measure, as with most similar measures of diversity, does not take into account the degree of difference14 among groups, which is, however, generally considered inadequate to account for the emergence of ethnic consciousness or the intensity of ethnic conflict. The lone
13. The Pearson correlation coefficient between the two measures is 0.52480. 14. See Bonacich (1980: p. 11) and Duvall and Welfling (1973) for this position. 74 exceptional view is suggested by Enloe (1986: pp. 20-21), who argues that ethnic mobilization is difficult for groups lacking strong cultural differentiations:
"Formosan nationalists on Taiwan, for example, have asserted that indigenous Formosans are distinct from mainland Chinese, whether those mainlanders are backers of Mao Tse-tung or Chiang Kai-shek. Perhaps one explanation for the movement's apparent inability to gain widespread support on Taiwan lies in the inseparability of mainland Chinese and Formosan ethnicity."
Whether her observation is accurate or not, we decide that we are not in a position to come up with such a measure.
C. Violent Political Behavior In this study, we would measure two dimensions of violent political behavior: (1) its level, and (2) its magnitude (or severity or intensity). The level of violent political behavior is represented by the frequencies of riots, armed attacks, and assassination respectively, while the magnitude of violent political behavior is determined by the aggregation of the death from these incidents and the incidents of political execution. We thus have four measures of violent political behavior.
The concept of violent political behavior is relatively straightforward: it denotes political behavior that turns violent. Since not all political conflicts would lead to violence, violent political behavior is not equal to 75 political conflict, which, for instance, may be in the form of strife in the parliament. According to the studies of destabilizing political behaviors by Wilkenfeld (1980: pp. 66-67), Morrison and Stevenson (1974: pp. 256-59), Eichenberg et al. (1984: pp. 14-16), Feierabend and Feierabend (1966: pp. 255-56), Gurr (1968), and Tanter (1966: pp. 49-50), we may find such conflict behavior as riots, demonstrations, coups, social revolutions, guerrilla warfare, assassinations, general strikes, and the like, which may be broadly collapsed into two categories: political protest and internal war. However, not all forms of political protest are necessarily violent: while strikes may be confined to the economic sphere only, neither would protest demonstrations nor political strikes necessarily lead to violence. Similarly, not all types of internal war are bloody. Coups, for instance, may be carried out without incurring any violence.
The event data for these four indicators are provided by Taylor and Jodice (1983): The original data were originally coded from the period of 1948-1977, mainly from the New York Times Index (Taylor and Jodice, 1983: pp. 19- 47). And the data in the ICPSR version have been expanded to 1982 (see ICPSR 7761, Code Book II). To insure sufficient variations of the two measures, we are tempted to choose those data for the whole period of 1948-1982. Still, it is noted that not all countries have been independent 76 since 1948. One solution is to decide how many years the country has existed in the period and divide the frequency and the magnitude of violent political behavior by it. The other strategy is to select a shorter time period in which as many countries as possible would have the same time frame, which would certainly contradict to our earlier criterion of having as many variation as possible. To strike a balance between the two criteria, we decide that the 1963-1982 period is our basic time frame15, so that we may have as many variations as possible in the dependent variables while as many countries are included.
D. Intervening Variables
Economic Development The level of economic development is measured by per capita Gross National Product16: the GNP in U.S. dollars in 1978 divided by the mid-year population in 1975 (ICPSR 7761, Code BooK: p. 73).
15. Initially, we intended to examine them during the 1963-1972 and the 1973-1982 periods in order to discern whether there were any variations in the dependent variables across-time. Accordingly, we would have three sets of data for the dependent variables. So far, we have only test the hypotheses against the data in the whole 1963-1982 period. See Appendix D for the data in the other two periods. 16. Other indicators are also popular, such as energy consumption in coal units (Eichenberg et al., 1984: p. 17). 77 Economic and Political Discrepancies Economic discrepancy is how equally economic wealth is distributed among the ethnic groups within a polity; and political discrepancy is similarly defined. Although there have been some attempts to measure such a distribution pertaining to individuals (Gini Index, for instance), few such group measures have been designed as far as we know. In this study, our measure of discrepancies include an economic and a political component based on the data of economic discrimination and of political discrimination provided by Gurr and Gurr (1983).
Applying the concept of Gini index, Grove (1973) has designed three indicators to measure socio-economic disparities for 35 ethnic groups: ethnic income distribution, ethnic groups in higher education, and proportion of ethnic in professional occupations. More recently, Grove (1991), constructs two measures of between- group occupational and educational differentials to tap the concept of ethnic division of labor for seven countries. As he rightly recognizes, the major difficulty students of ethnic relations meet is the reluctance of government to conduct census or to release any census data based on group differentiations.
Nonetheless, we are not helpless in this regard. Gastil (1975: p. 8; 1976: p. 19; and 1977: p. 14) reports a low-medium-high measure of political equality for 92 ethnic 78 groups with a population larger than 1,000,000 in 35 countries in the 1975-1977 period. A similar measure of political equality is reported for 91 ethnic groups by him in 1978 (pp. 14-15). The major deficiency of these two measures reported by Gastil is its limited observations. Gurr (1966) provides a measure of group discrimination for 119 political units, showing only the proportion of the ethnic groups discriminated against. Gurr and Gurr (1983) later provide a 0-4 ordinal indicator of economic discrimination and one of political discrimination against ethnic groups in 84 countries17 in 1960 and 1975. We thus decide to compute an interval measure of economic discrepancies by multiplying it with the percentage of the group size. A similar measure of political discrepancies is constructed likewise also based upon their data. In the actual analyses later on, the measure of economic and political discrepancies is the average of these two measures18.
Governmental Coercion Governmental coercion is the use of force by the governmental to restrict political behavior of individual or
17. The measure for Asian communist countries and some Third World countries are missing because of insufficient information. 18. The Pearson correlation coefficient between the two measures is 0.85967. 79 groups. While two dimensions of governmental coercion are generally distinguished: potential and realized, we judge that the latter possess higher construct validity. Thus our measure of governmental coercion will focus on this dimension, based on Taylor and Jodice's (1983: pp. 62-63) measure of imposition of political sanction19, defined as "an action taken by the authorities to neutralize, suppress, or eliminate a perceived threat to the security of the government, the regime, or the state." (p. 62) Specifically, these coercive actions include censorship and restrctions on political behavior, such as the declaration of martial law, the imposition of curfew, and restrictive measures against individual political figures or political organizations, (pp. 62-63)
Alternatively, it may be inferred that the more rights people enjoy, the lower level of governmental coercion is; conversely, the less rights they enjoy, the higher level of governmental coercion is. Taylor and Jodice (1983) report two indexes of political right and civil right, which are mean scores varying from 1 to 7 for the period of 1973-1979. We may derive an indirect measure of governmental coercion by firstly adding up the two mean scores and then dividing it by two.
19. Actually, Taylor and Jodice (1983: pp. 62-63) supply two indexes of state coercive behaviors: imposition of political sanction and of political execution. The latter aspect has been incorporated into the measure of death from violence. 80 While the former measure of governmental coercion is interval, which is more attuned to regression analysis, the latter is ordinal in essence although the two-step manipulation process makes it appear interval. We decide to adopt the former measure20.
Role of Ethnic Elites The role of ethnic elites in mobilizing ethnic consciousness may be manifold: political, economic, and cultural. Oftentimes, they are so intertwined that we may not be able to disentangle them. In addition, it is hardly possible to assign weights to those activities and aggregate them, not to mention how to compare them cross-nationally. One way to tap the concept is through the observation of the existence of ethnic parties or insurgent movements.
Nielsson and Jones (1988) provide a list of ethnic groups according to the degree of mobilization: unmobilized, latent, early phase of political mobilization, mobilized as insurgent movements, and mobilized as political movements and parties. We have no way to ascertain whether these ethnic activities are spontaneous or the consequence of elites' mobilization. The other problem would be how to aggregate them and come up with a single measure for each country.
20. Statistical results using the latter indirect measure of governmental coercion are reported in Appendix G. 81 Based largely on the data from Gurr (1966), Gurr and Gurr (1983: pp. 55-57 and pp. 72-75) provide a 0-4 measure of separatist potential in 84 countries. Additionally, they show the proportion of the ethnic groups concerned in terms of the total population. We thus design an interval measure by multiplying the proportion of the group size and the intensity of potential separatism. As constructed and transformed, this measure of role of ethnic elites would be interval. However, we must point out that we do not know how they have obtained the value for each group within the polity and how the national value is derived if there are more than one ethnic group. Another deficiency is that the separatist movement is not necessarily led by the elite, which is, however, highly improbable.
Geographic Concentration of Ethnic Groups As for geographic concentration of ethnic groups, we may design a dummy variable by visual observation of ethnic distribution of each country. The difficulty lies in the fact that we have yet to combine a group-centric perspective within a state system. If there is any territorially based group identity, we would assign value one to the country. However, the measure of geographical distribution of ethnic groups within a country is rarely available, given the fact that most governments are reluctant to recognized the existence of ethnic minorities. Some may be found in the 82 Area Handbook or the Country Study series published by the American University. Additionally, the geographical distributions of linguistic minorities in Western Europe are provided by Stephens (1976). Alternatively, we may infer indirectly the existence of territorially based ethnic groups from Gastil*s lists of major subordinated peoples (1975: p. 8; 1976: p. 19; 1977: p. 14) and the list of major peoples without a nation-state (1978: pp. 14-15). Since only territorial peoples are considered by him (1975: p. 8; 1978: p. 13), we would not have included groups that are dispersed. Gurr and Gurr (1983: pp. 55-57) and Gurr (1966) also list major ethnic groups with separatist potentials, which serves as complimentary lists. Anderson et al. (1967: pp. 68-69) also list important secessionist movements.
External Links of Ethnic Groups Still fewer distributions of ethnic groups residing across states are available. Two plausible ways are considered. A dummy variable for external links of ethnic groups may be similarly derived from Gastil's (1978: p. 18) list of separated peoples and major peoples without a nation-state (pp. 14-15) and Anderson et al.'s list of irredentism (1967: pp. 68-69), which both show the locations of these groups but are handicapped by limited observations.
Nielsson (1985: pp. 39-41), using an ethnic-centric 83 perspective, classifies 614 political units into 5 categories according to how ethnic groups are dispersed: single-state nation-groups, single-state majority nation- groups, single-and-multistate minority nation-groups, multistate majority and minority nation-groups, and multistate nation-groups. We accordingly arrive at a 0-4 ordinal measure of external links.
E. A Word on Aggregated Data Since the present study is a cross-national one that is highly dependent on aggregate data as evidence, it is imperative that we examine the advantages and limitations of this type of data. We shall start with a brief recognition of their merits, and focus on their limitations and drawbacks that we should be aware of. Finally, we shall conclude with our attitudes in employing them. Two advantages of aggregate data are appreciated here. The first one is concerning the way they are collected. Compared with such a measure as interview, aggregate data are relatively nonobtrusive. Secondly, they can be more readily used in cross-national studies in order to examine the relationship between various attributes of the political systems. More specifically, they can be applied for univariate, bivariate, multivariate, and factor analyses. Thus, not only can we utilize them for descriptive purposes, we can also look into possible correlation relations between 84 the variables, uncover potential third variables, and reduce data into dimensions of the phenomenon under investigation. The first problem involved is lack of data. A related but distinctive issue is the sampling problem. For some third countries, there may be no data available for comparison. We thus must be very careful when drawing generalization from the data. Further, we may have too many missing values in our observations, which in turn lead to the skewedness of the distribution, especially for event data. One way to redress the problem is to transform the original data logrithmically. Nevertheless, it is not always useful.
Since aggregate date are usually used to tap systemic attributes, they are parameters for political systems. Thus, there is usually no variation within a state. In other words, there are few intra-systemic data, except for such a country as the US. A related issue called "ecological fallacy" may arise if we attempt to apply the generalization from aggregate data to the constituents of the system. In other words, what hold for the group is not necessarily hold for the individuals of that group.
The third problem is data comparability of aggregate data. Two issues may come up if we use data from different systems. Firstly, there may be different procedures of gathering data. For instance, the accounting systems of different countries may not be the same. Take GNP, the most 85 used measure of economic development, for example. In the socialist systems, where barter system is used among themselves and the service sector is not included. We must be aware of what method is adopted to calculate GNP. Furthermore, we are concerned with the problem of functional equivalence. We must ask ourselves: Do the indicators tap the same underlying dimensions in each country? In other words, the same indicators, such as race, may mean different things in different countries. Therefore, we may need to adopt comparable indicators for the same phenomenon.
The fourth problem is the validity issue: whether the indicator captures what it is supposed to tap. At times, a phenomenon may possess different dimensions. In such a case, a single all-encompassing indicator may be inadequate. We thus need multiple indicators for the same phenomenon. If necessary, we may want to factor analyze these indicators and design a composite indicator. This issue can also be addressed with a more thorough theorizing of the relationship we want to investigate.
The last, and most complicated, problem is reliability issue. Many errors may contribute to low reliability of aggregate data. Random human errors are less series since they could cancel each other out. The most serious problem is systematic error resulting from the inadequate reporting system. It could be due to insufficient training of the reporter, expansion of the statistic agency over time, varying definitions, ways of counting, and units used, and other social conditions. Systematic errors are more grave in the case of event data, for which prestigious newspapers, such as New York Times or Washington Post, are the main sources. Their report may be uneven owing to their own professional criteria. Therefore, some events may be neglected because of 'lack of news value. The situation may also be due to censorship found in most the Third World countries (Merrit, 1970: p. 40). Similarly, the national statistic may purposefully attempt to deflate or inflate statistics for any reason. For instance, the government may decide to the existence of ethnic minorities. The last threat to reliability comes from secondary data. As Banks and Texter (1963) point out, two thirds of the data in their A Cross-Polity Survey are secondary and only one third primary. Judgement variables are thus indispensable. This is especially the case when insufficient data are available. When we thus have to rely on estimation, discrepancies are expected.
Three procedures are suggested by Gillespie (1971: p. 19): to consult alternative data sources, to refer to other studies on the same phenomena, and look into data-collection procedures. After examining the above, should we come to the 87 conclusion that we stop using aggregate data? The answer is negative. Every data set has its problems, which we must learn to live with. We therefore must be aware of them when doing research. Furthermore, every data set has its purpose and procedure. We thus have to ask by whom, how, and why they are collected. In a word, we should never take any data at their face values.
F. Univariate Statistics21 Before proceeding with our hypothesis tests, it will be useful to note some univariate statistics on the independent and dependent variables in our data. For the complete set of 132 states, the average value of ethnic diversity is 0.33 with an associated standard deviation of 0.2135, varying from 0.02 to 0.78; the average level of economic development is 1709, with a standard deviation of 2376, ranging from 90 to 15190; the mean level of governmental coercion22 is 0.03, with a standard deviation of 0.09, varying from 0 to 0.75; the average degree of geographical concentration is 0.51, with a standard deviation of 0.05, varying from 0 to 1 the average level of external link is 1.95 with a standard deviation of 1.42, varying from 0 to 4.
21. See Appendixes D and E for the original data before and after any manipulations and transofmormations. 22. The data have been standardized and logarithmically transformed since the variable is aggregated from frequencies. 88 Before any transformation, the frequency of riot in the 1963-1982 period varies from 0 to 780, that of attack varies from 0 to 12405, and that of assassination varies from 0 to 67; and the range of death from violence is quite wide, varying from 0 to 1995235, with a mean of 102 and a standard deviation of 182103. In the actual statistical manipulations, however, the frequency of riot, armed attack, assassination are firstly standardized for cross-tabulations, and then logarithmically transformed for regression analyses. The rationale is that these three event data tend to vary with the population of the country. Thus, the frequency for each observation of the three indicators is divided by the population of each country in 1970, which is the middle of the 1963-1982 period. Since the standardized frequencies of three indicators are still too skewed for regression analysis, we then undertake a logarithmic transformation. The transformed riot has an average of 0.014 and a standard deviation of 0.038, ranging from 0 to 0.389; the transformed armed attack has a mean of 0.046, a standard deviation of 0.178, varying from 0 to 1.649; and the transformed assassination has an average of 0.002, a standard deviation of 0.012, ranging from 0 to 0.135.
For the reduced sample of 82 countries, the average degree of economic and political discrepancies is 0.27, with a standard deviation of 0.52, the minimum and maximum being 89 0 and 2.89 respectively; the average level of the role of elites is 0.38, with a standard deviation of 0.65, varying from 0 to 3.76. These statistics are presented in Table 1. TABLE 1. Univariate Statistics of All Variables
Variable Mean Median Standard Minimum Maximum N Deviation DIV 0.33 0.31 0.21 0.02 0.78 132 ECO 1709 600 2376 90 15190 132 DISQ 0.27 0.07 0.52 0 2.89 82 COER 0.03 0.01 0.09 0 0.75 132 ELT 0.38 0 0.65 0 3.75 82 GEO 0.51 1 0.51 0 1 132 EXT 1.95 2 1.42 0 4 132 RIOT 45 14 93 0 780 132 ATT 258 32 1118 0 12405 132 ASS 4.64 1 10 0 67 132 DEATH 25010 102 182103 0 1995235 132 RIOT" 0.01 0 0.05 0 0.48 132 ATTa 0.07 0.01 0.40 0 4.20 132 ASSa 0 0 0.01 0 1.14 132 DEATH" 0.03 0.02 0.09 0 0.75 132 RIOT" 0.01 0 0.04 0 0.39 132 ATTb 0.05 0.01 0.18 0 1.65 132 ASSb 0.01 0 0.01 0 0.13 132 DEATH" 0.27 0.02 0.64 0 3.62 132 a: Y' = Y/POP, where Y = RIOT, ATT, ASS, or DEATH. b: Y" = In (Y/POP + 1), where Y = RIOT, ATT, ASS, or DEATH. CHAPTER VI EMPIRICAL TESTS AND ANALYSES
A. Correlation Analysis To provide quick evidence on the accuracy of Hypothesis 1, we compute the Pearson (also termed product-moment) correlations between the four indicators of violent political behavior and ethnic diversity. The results of those computations are presented in Table 2. Given the weak negative correlation coefficients between ethnic diversity and riot, armed attack, and assassination, Hypothesis 1 is explicitly not supported. Nonetheless, the relationship between ethnic diversity and death from violence, while in the hypothesized direction, is minimally supported.
B. Bivariate Cross-tabulation We need to employ more sophisticated methods to test for the interaction effects of ethnic diversity together with the other independent variables suggested in Hypothesis 2 through Hypothesis 7. A 2 by 2 cross-tabulation (also termed cross-classification) is appropriate for the
91 TABLE 2. Pearson Correlation Coefficients Among All Variables"
DIV ECO DISQ COER ELT GEO EXT RIOT ATT ASS DEATH 1.00000 -0.35569 0.19795 -0.12350 0.58900 0.27170 0.23777 -0.12078 -0.06824 -0.11100 0.18352 DIV 0.0 0.0001 0.0746 0.1583 0.0001 0.0016 0.0060 0.1678 0.4369 0.20S1 0.0352 132 132 82 132 82 132 132 132 132 132 132 -0.35569 1.00000 -0.20782 0.09176 -0.27981 0.04911 -0.09639 0.06128 0.07728 0.03791 -0.08164 ECO 0.0001 0.0 0.0610 0.2954 0.0109 0.5760 0.2715 0.4851 0.3785 0.6661 0.3520 132 132 82 132 82 132 132 132 132 132 132 DISQ 0.19795 -0.20782 1.00000 -0.06293 0.00222 0.05722 -0.03142 -0.06151 -0.05442 -0.06184 -0.07345 0.0746 0.0610 0.0 0.5744 0.9842 0.6096 0.7793 0.5830 0.6273 0.5810 0.5119 82 82 82 82 82 82 32 82 82 82 82 -0.12350 0.09176 -0.06293 1.00000 -0.05330 0.02409 0.17285 0.85507 0.59951 0.64092 -0.03360 COER 0.1S83 0.2954 0.5744 0.0 0.6344 0.7840 0.0475 0.0001 0.0001 0.0001 0.7021 132 132 82 132 82 132 132 132 132 132 132 ELT 0.58900 -0.27981 0.00222 -0.05330 1.00000 0.40485 0.28176 -0.06342 0.04631 -0.06260 0.61064 0.0001 0.0109 0.9842 0.6344 0.0 0.0002 0.0103 0.5714 0.6795 0.5763 0.0001 82 82 02 82 92 82 82 82 82 82 82 GEO 0.27170 0.04911 0.05722 0.02409 0.40485 1.00000 0.08625 0.05796 0.13809 0.07872 0.12426 0.0016 0.5760 0.6096 0.7840 0.0002 0.0 0.3254 0.5092 0.1143 0.3696 0.1557 132 132 82 132 82 132 132 132 132 132 132 EXT 0.23777 -0.09639 -0.03142 0.17285 0.28176 0.08625 1.00000 0.14229 0.10393 0.12992 0.00877 0.0060 0.2715 0.7793 0.0475 0.0103 0.3254 0.0 0.1036 0.2357 0.1376 0.9205 132 132 82 132 82 132 132 132 132 132 132 RIOT -0.12076 0.06128 -0.06151 0.85507 -0.06342 0.05796 0.14229 1.00000 0.69596 0.91321 -0.02701 0.1678 0.4851 0.5830 0.0001 0.5714 0.5092 0.1036 0.0 0.0001 0.0001 0.7585 132 132 82 132 82 132 132 132 132 132 132 ATT -0.06824 0.07728 . -0.05442 0.59951 0.04631 0.13809 0.10393 0.69596 1.00000 0.58513 -0.01637 0.4369 0.3785 0.6273 0.0001 0.6795 0.1143 0.2357 0.0001 0.0 0.0001 0.8522 132 132 82 132 82 132 132 132 132 132 132 ASS -0.11100 0.03791 -0.06184 0.64092 -0.06260 0.07872 0.12992 0.91321 0.58513 1.00000 -0.01782 0.20S1 0.6661 0.5810 0.0001 0.5763 0.3696 0.1376 0.0001 0.0001 0.0 0.8392 132 132 82 132 82 132 132 132 132 132 132 DEATH 0.18352 -0.08164 -0.07345 -0.03360 0.61064 0.12426 0.00877 -0.02701 -0.01637 -0.01782 1.00000 0.0352 0.3520 0.5119 0.7021 0.0001 0.1557 0.9205 0.7585 0.8522 0.8392 0.0 132 132 132 132 132 132 82 132 82 132 132 a: Pearson Correlation Coefficients / Prob > |R| under HOJ Rho>0 / ttuaber of Observations
to 93 preliminary probe before we conduct regression analyses1. The cross-tabulation table is one of the most useful tolls for the bivariate or multivariate analysis of nominal or ordinal data. But it is often used as a first step in detecting relationships between interval variables. Again, the measures of violent political behavior, riot, armed attack, and assassination, have been standardized transformed before the construction of cross- tabulations. Each indicator is then classified into two groups, "Low" and "High," according to its values, thus generating roughly equal percentages of frequencies in each category2. The independent and all intervening variables except geographical concentration3 are also divided in a similar way. We check the expected frequencies in all cells to make sure none of them is smaller than 5. All of our 2 by 2 classifications conform to this requirement.
As we can see from Table 3 through Table 6, the large P values for the %2 statistics, with the lxl degree of
1m We might as well construct 3 by 3 cross-tabulations. However, with the limited cases in hand (N = 132 for most of the variables and N = 82 for two intervening variables; see Table 1), the expected frequency for certain may be less than 5, which will render the statistical results less reliable. See the discussions by Agresti and Agresti (1979: p. 212). 2. The median of each variable is listed in Table 1. 3. Geographical concentration is a dichotomous variable, with 65 observations having the value of 0, and 67 observations having the value of 1. It is thus convenient to have the two categories demarcated according to the two groupings. 94 freedom, show that there is no statistical evidence beyond the 0.1 level to reject the null hypothesis that ethnic diversity is independent from riot, armed attack, assassination, or death from violence. Accordingly, Hypothesis 1 that the degree of ethnic diversity and the level of violent political behavior is positively related is not supported by our data. Table 7 through Table 30 are constructed to evaluate the relationships between violent political behavior and the six intervening variables. Although these variable may have interaction impacts on the relationship between ethnic diversity and violent political behavior, they may also exert direct influence upon the violent political behavior. In this sense, they are considered independent variables in their own right and thus deserve our separate treatment before we can evaluate the adequacy of our model. From Table 7 through Table 104, we find that economic development is interdependent with death from violence at the level of 0.1, but not with riot, armed attack, or assassination. As shown in Table 11 through Table 14 , the null hypothesis that economic and political discrepancies and violent political behavior are independent is moderately rejected at the level of 0.05 in so far as riot, armed attack, or assassination is used as indicator. But when
4. Table 7 through Table 30, Table 32 through Table 55, and Table 57 through Table 70 are presented in Appendix F. TABLE 3. Cross-Tabulation of Ethnic Diversity with Riot
Ethnic Diversity Riot
Low High JPAN FNLD IRAQ DMNR ITLY ARGN YMNS LXBG TWAN FRG GRCE SPAN NRWY BRBD URGY DNMK CLMB TRKY IRLD PRGY BLGR PRTG VNZL CRCA SDAR RMNA MXCO MLTA CHLE TNSA Low BRZL KWAT BRND YMNA LBYA FRNC SWDN CUBA AUSL SMLA PNMA JMCA KORN EGPT MNGL PLND NCRG MRTN GDR KMPC HNDS KORS JRDN LSTO ICLD HNGR MDGS MLDV ISRL ELSL HATI RWND CHNA NTHL BRMA ' AUST N=37 N=29
NZLD PPNG GNEA LAOS TRNT USSR SRLK YGSL MALI BLGM ZIMB GHNA ALBN CZCH GMBA GTML IRAN SRLE ALGR PHLP ETHP USA MRTS ZMBA High SWTZ INDS ZAIR BOLV BNIN SDAN SNGP MZBQ LBRA NPAL GBON MLYS TLND CNDA IVCT SYRA CAFR ANGL AFGN NGER CMRN PERU SNGL KNYA BTSN INDA TNZN ECDR CNGO SAFR MRCO UPVL CYPR GYNA CHAD UK TOGO UGND LBNN MLWI NGRA PKST N=29 N=37 Total N=66 N=66 X2 = 1.94 P = 0.16 * = 0.12 96 TABLE 4. Cross-Tabulation of Ethnic Diversity with Armed Attack
Ethnic Diversity Armed Attack Low High JPAN KORS NTHL YMNS CLMB ARGN NRWY CHNA TWAN IRLD VNZL SPAN SDAR AUST URGY DMNR CUBA IRAQ BRZL FNLD BLGR FRG KMPC CRCA SWDN LXBG TNSA DNMK HATI BRND Low PRTG BRBD MXCO YMNA LBYA FRNC MLTA PRGY AUSL ELSL PNMA JMCA KORN RMNA MNGL MLDV NCRG HNDS GDR EGPT MRTN ITLY JRDN LSTO ICLD CHLE MDGS GRCE ISRL BRMA SMLA HNGR RWND KWAT PLND TRKY N=35 N=31
NZLD YGSL TOGO LAOS AFGN ZMBA USA CZCH GHNA BLGM PKST SDAN SRLK INDS SRLE GTML ZIMB MLYS ALBN CNDA GNEA BOLV PHLP ANGL NPAL BNIN MALI ALGR IRAN ETHP High SWTZ NGER GMBA SYRA MRTS KNYA SNGP CAFR LBRA PERU MZBQ SAFR BTSN SNGL IVCT ECDR GBON CHAD MRCO INDA CMRN CYPR CNGO ZAIR PPNG UPVL TNZN UK GYNA UGND TRNT LBNN MLWI NGRA TLND USSR
N=31 N=35 Total N=66 N=66 97 TABLE 5. Cross-Tabulation of Ethnic Diversity with Assassination
Ethnic Diversity Assassination Low High JPAN CHNA TWAN YMNS PRGY SPAN NRWY MLDV BLGR IRLD KWAT TRKY BRZL FNLD CRCA SDAR CLMB NTHL DNMK LXBG MXCO DMNR VNZL IRAQ PRTG BRBD AUSL SWDN CUBA URGY Low MLTA RMNA MNGL FRG KMPC TNSA KORN EGPT MRTN YMNA CHLE BRND GDR HNGR HNDS SMLA HATI FRNC ICLD LBYA RWND ELSL DNMA JMCA PLND JRDN BRMA AUST NCRG LSTO KORS ITLY ISRL MDGS GRCE ARGN N=31 N=35
BLGM PPNG SRLE LAOS YGSL MLWI NZLD TRNT SDAN GTML ZIMB USSR USA CZCH MALI BOLV PHLP ZMBA SLRK INDS GMBA ALGR IRAN GNEA ALBN MRTS ANGL SYRA BNIN MLYS High NPAL MZBQ SAFR ECDR NGER ETHP PERU CNDA CHAD CYPR SNGL KNYA SWTZ GBON ZAIR UK CNGO UGND SNGP CAFR LBRA LBNN GYNA NGRA TLND CNDA IVCT AFGN TOGO TNZN BTSN UPVL CMRN PKST MRCO GHNA N=35 N=31
Total N=66 N=66 98 TABLE 6. Cross-Tabulation of Ethnic Diversity with Death from Violence
Ethnic Diversity Death from Violence
Low High JPAN PLND PNMA YMNS VNZL SPAN NRWY MLDV NTHL SDAR CUBA TRKY IRLD AUST URGY NRZL EGPT IRAQ SWDN FNLD BLGR DMNR KMTC TWAN FRG GRCE CRCA YMNA CHLE MXCO DNMK LXBG TNSA KORS RATI BRND PRTG BRBD FRNC ELSL NCRG JMCA MLTA PRGY AUSL CHNA JRDN MDGS KORN RMNA MNGL ITLY ISRL RWND GDR KWAT MRTN • CLMB ARGN BRMA ICLD HNGR HNDS SMLA LBYA LSTO N=36 N=30
BLGM CZCH MLWI LAOS AFGN USSR NZLD MRTS GHNA GTML MRCO ZMBA ALBN CNDA SRLE USA PKST SDAN NPAL BNIN GNEA SRLK YGSL MLYS High ECDR GBON MALI BOLV ZIMB ANGL SWTZ NGER GMBA ALGR PHLP ETHP SNGN CAFR LBRA SYRA IRAN KNYA BTSN SNGL IVCT PERU INDS SAFR PPNG UPVL CMRN CYPR MZBQ CHAD TRNT TOGO TNZN UK INDA ZAIR LBNN CNGO UGND TLND GYNA NGRA N=30 N=36
Total N=66 N=66 X2 - 1.09 P = 0.30 0 = 0.09 99 death from violence is used, the relationship disappers. Nevertheless, it is noted that there are only 82 cases in these four cross-tabulations, since 50 states lack the value of economic and political discrepancies. In Table 15 through Table 18, the %2 statistics give the strongest statistical and substantial support (P < 0.001) to the hypothesis that governmental coercion is positively related with violent political behavior, yielding
= 0.27). From Table 23 through Table 26, statistical evidence suggest that geographical concentration is positively related with armed attack and with death from violence at the significant level of 0.1 ( Dependent Variable Inde pendent Armed Assassi- Death from Variable Riot Attack nation Violence DIV ECO DISQ ** ** ** COER *** *** *** ELT *** GEO ** *** EXT ** a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01.that death from violence, one indicator of 101 violent political behavior no matter which of its measures is used. Only economic and political discrepancies and governmental coercion possess statistical association with three of the four measures of violent political behavior, riot, armed attack, and assassination. The other five intervening variables are interdependent with only one or two measures of the dependent variable in varying degrees: economic development with death from violence, role of ethnic elites with death from violence, geographical concentration with armed attack and death from violence, and external links with riot and death from violence. Thus, if the intervening variables do covary with violent political behavior, they differ in its association with its different indicators. Judging from a different angle, we discover that death from violence, one indicator of violent political behavior, is related with four of the independent variables we have selected except economic and political discrepancies and governmental coercion. Riot and armed attack are associated with three intervening variables respectively, and assassination is found to covary with economic and political discrepancies and governmental coercion only. C. Multivariate Cross-tabulation In the previous sections, we have used 2 by 2 cross- tabulations to analyze the relationship between the dependent variable and the seven independent variables. The 102 following multivariate inquires will investigate whether the direct relationship, if any, between ethnic diversity and violent political behavior still holds true within different levels of the six intervening variables. In other words, we need to determine whether ethnic diversity is still a factor affecting difference in level of violent political behavior when we control for differences in one of the intervening variables, say, economic development. Three-dimensional 2 by 2 by 2 cross-tabulations, as shown in Table 32 through Table 35, are appropriate for this controlling purpose. In Table 32, we construct a partial table relating riot to ethnic diversity for those states in the sample that have "Low" level of economic development. A similar table for those countries that have "High" level of economic development is also developed. Thus, we have two partial tables that may reveal the relationship between ethnic diversity and riot within the two levels of economic development, the control variable. We find that countries with higher levels of economic development show higher partial relationship ( Similar partial tables are made for ethnic diversity 103 and armed attack controlling for economic development. The results are reported in Table 33. According to the values of As Table 38 demonstrates, in states with a higher level of economic and political discrepancies, ethnic diversity and assassination is negatively associated at the significance level of 0.1, which is opposite to what we have expected in Hypothesis 3. We thus conclude that there is an interaction relationship among ethnic diversity, economic and political discrepancies, and assassination. More specifically, in states with higher economic and political discrepancies, ethnic diversity is less conducive to assassination. It is noted that the result is opposite to what we have hypothesized in Hypothesis 3. Hypothesis 4 is statistically tested in Table 40 through Table 43. Table 40 demonstrates whether the association between ethnic diversity and riot is intervened by governmental coercion. The overall (1) As the related %2 statistic is significant at the 0.1 level, we are able to conclude that in countries with higher level of governmental coercion, the association between riot and governmental coercion seems stronger after we control for governmental coercion ( As shown in Table 42, we find that there is no association between ethnic diversity and assassination after we control for governmental coercion. In other words, there is no statistical interaction among the three variables. According to the results in Table 43, it is discovered that in countries with low level of governmental coercion, ethnic diversity is not associated with death from violence at the 0.1 level, and that in countries with high level of governmental coercion, ethnic diversity and death from violence are not statistically interdependent on each other. Accordingly, we conclude that there is no interaction relationship among ethnic diversity, governmental coercion, and death from violence as suggested by Hypothesis 4. In conclusion, Hypothesis 4 is supported inasmuch riot, or 107 armed attack is used as indicator of violent political behavior. Table 44 through Table 47 are constructed to test for Hypothesis 5: in countries where ethnic elites play a more active role in ethnic mobilization, ethnic diversity is more conducive to violent political behavior than in ones where ethnic elites play a less active role in ethnic mobilization. Partial tables show that there is no relationship between ethnic diversity and riot, armed attack, assassination, or death from violence after we control for role of ethnic elites. Since the expected frequency of one or two cells in the four tables is smaller than 5 (for instance, /e = 3 in Table 44), the Fihser's exact test is preferable to the %2 test (Agresti and Agresti, 1979: p. 209; Blalock, 1979: pp. 292-97). Still, the probability of getting the association is smaller than 0.1. Still, none of them is significant at the 0.1 level. It is recalled that ethnic diversity and death from violence are marginally associated (0 = 0.15) at the 0.1 level in the original 2 by 2 cross-tabulation. When role of ethnic elites is hold constant, the original relationship is explained away. Thus, the original association between ethnic diversity and death from violence is spurious (Babbie, 1979: pp. 452-54). To conclude, Hypothesis 5 is not supported by the data. 108 Hypothesis 6 is tested in Table 48 through 51: in a country where there are more geographically concentrated ethnic groups, ethnic diversity is more conducive to violent political behavior than in one with fewer such ethnic groups. According to Table 48, in countries in the category of "High" geographical concentration, ethnic diversity and riot are positively associated; in countries in the "Low" geographical concentration, there is no statistical interdependence between ethnic diversity and riot. Thus, there is an interaction relationship among ethnic diversity, geographical concentration, and riot. Since there is no statistical relation between ethnic diversity and riot in the original 2 by 2 cross-tabulation, we conclude that their relationship is actually suppressed by geographical concentration. As shown in Table 49, in countries with high level of geographical concentration, ethnic diversity is positively associated with armed attack; on the other hand, in countries with low level of geographical concentration, ethnic diversity is found to be negatively related to armed attack. Both the %2 statistics are significant at the 0.1 level. Accordingly we determine that there is an interaction relationship among ethnic diversity, geographical concentration, and armed attack. Since the association patters are reverse within different levels of geographical concentration, no meaningful summary measure of 109 association is available. As with riot, the relationship between ethnic diversity and armed attack is suppressed by geographical concentration. According to the results in Table 50, we determine that there is no association between ethnic diversity and assassination even if we held the value of geographical concentration constant. Thus, there is no interaction among ethnic diversity, assassination, and geographical concentration. Table 51 reveals that in countries with high level of geographical concentration, ethnic diversity is positively associated with death from violence ( Table 52 through Table 55 are constructed to test for Hypothesis 6: in a country where ethnic groups have more external ties, ethnic diversity is more conducive to violent political behavior than in one with fewer such external ties. Partial tables in all the four tables reveal that there is no association between ethnic diversity and riot, armed attack, assassination, or death from violence after the value of external links is hold constant. We infer that 110 there is no interaction relationship among ethnic diversity, external links, and violent political behavior and that Hypothesis 6 is not substantiated by the data. Since the original 2 by 2 cross-tabulation shows that ethnic diversity and death from violence are positively related, this association is found spurious. To conclude, Hypothesis 2, regarding the interaction relation among ethnic diversity, violent political behavior, and economic development, is marginally supported only in the case of death from violence. Although economic and political discrepancies are found to interact with ethnic diversity and assassination, the direction is opposite to what Hypothesis 3 suggests. Hypothesis 4, regarding the interaction relation among ethnic diversity, violent political behavior, and governmental coercion, is supported in the cases of riot and armed attack. The interaction relationship among ethnic diversity, violent political behavior, and role of ethnic elites, as suggested in Hypothesis 5, is not substantiated by our data. Hypothesis 6, regarding the interaction relationship among ethnic diversity, violent political behavior, and geographic concentration of ethnic groups, is favorably supported in the cases of riot, armed attack, and death from violence, but not in the case of assassination. Finally, the interaction relationship among ethnic diversity, violent political behavior, and external links of ethnic groups, as Ill suggested in Hypothesis 7, is not rejected by our data. These results from cross-tabulations are summarized in Table 56 and visually presented in Figure 4 through Figure 7. In summary, there is no relationship between ethnic diversity and violent political behavior. While economic development is negatively related to death from violence as expected, the interaction effect is positive, opposite to what we have hypothesized. The degree of economic and political discrepancies is positively related to violent political behavior as suggested, but one interaction effect is positive, which is contrary to what we have hypothesized. Governmental coercion is found to be positively related to violent to behavior, which is opposite to what the common sense tells us. While governmental coercion reinforce riots that ethnic diversity may lead to, it relieves armed attack that ethnic diversity may cause. The results confirm our earlier concern that ethnic diversity and the intervening variables may have different impacts on disparate aspects of violent political behavior. While the role of ethnic elites is found to be positively related to death from violence, there is no interaction effect as hypothesized. Geographical concentration of ethnic groups is related to violent political behavior, and the positive interaction effect is also confirmed. Finally, while external link of ethnic links is positively related to violent political behavior as expected, the interaction effect is not found. 112 TABLE 56. Summary of the Cross-Tabulations" Dependent Variable Controlled Armed Assassi Death from for Riot Attack nation Violence ECO Low - High .. _ * DISQ Low - High - - * COER Low *** High * * - - ELT Low - High - - - GEO Low - *** High * * ** EXT Low High - - a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 113 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Intervening Variables FIGURE 4. Final Hodel of Riot from Cross-tabulations 114 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Intervening Variables FIGURE 5. Final Hodel of Armed Attack from Cross-tabulations 115 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Geographic External Concentration Links Intervening Variables FIGURE 6. Final Hodel of Assassination from Corss-tabulations 116 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Intervening Variablas FIGURE 7. Final Hodel of Death from Violence from Cross-tabulation 117 D. Bivariate and Multiple Regressions Perhaps the dichotomous nature of the preceding cross- tabulation analyses is masking some underlying relationships we have hypothesized. Since our variables are either ordinal or interval, stronger statistical procedures that use the additional information given by the higher level of measurement are sought. Therefore, our next step is to construct bivariate and multivariate regression lines to investigate the hypothesized linear relationship between ethnic diversity and violent political behavior and their interaction relationships with economic development, economic and political discrepancies, governmental sanction, role of ethnic elites, geographical concentration, and external links. Before undertaking the bivariate regression analyses, four scatter plots are constructed in order to inspect the aptness of the linear model (Neter and Wasserman, 1974: p. 99). We construct median traces in the plots to evaluate the linearity and homoscedasticity of the relationship between ethnic diversity and the four measures of violent political behavior. Except the scatter plot of death from violence on ethnic diversity, the other three distributions are all near the X axis. In other words, most of the observations of riot, armed attack, and assassination are around the value of zero, which makes impossible the construction of median traces. As we have mentioned earlier 118 in the stage of operationalization, a logarithmic transformation of the three former indicators of violence of political behavior is necessary. By so doing, we draw those observations with extreme large values to median values and spread the values of those observations with small values (Tufte, 1974: p. 109). In addition, the procedure helps to normalize the residuals (Neter and Wasserman, 1974: p. 123) That violent political behavior is positively influenced by ethnic diversity as suggested in Hypothesis 1 is tested against Equation 1, where we try to explain cross- national difference in the aggregated frequency of violent political behavior in the period of 1963-1982. The general form of the equation is expressed as: (1) Y = a + (3 DIV + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, e = stochastic disturbance. Since we have four indicators of violent political behavior, the above equation can be rewritten as the following four companion equations: (2) In (RIOT/POP + 1) = a + p DIV + e, (3) In (ATT/POP + 1) = a + p DIV + e, 119 (4) In (ASS/POP +1) = a + p DIV + 6, (5) DEATH = a + P DIV + 6, where RIOT = Riot 1963-1982, ATT = Armed Attack 1963-1982, ASS = Assassination 1963-1982, DEATH = Death from violence 1963-1982, DIV = Ethnic Diversity, e = stochastic disturbance. The results of our analysis are presented in Table 57, including the estimated coefficient for the independent variable, ethnic diversity, the related t statistic, the R2, and the sample size (N). Since our sample size is relatively large (N = 132), we relax the criterion for statistical significance to the level of 0.1. Thus, if the attained significance level for the F test, P value, is equal or smaller than 0.1, it is marked "*;" if P is equal or smaller than 0.5, it is marked "**," and if P is equal or smaller than 0.01, it is marked "***." The results for Equation 2 through Equation 4, presented in Table 57, are hardly robust. None of the estimates for ethnic diversity are significant at the 0.1 level. Nor are the attained significance levels of the F test for the fitness of the regression lines equal or smaller than 0.1. On the other hand, the regression for 120 Equation 5 is able to account for 3.4% of the cross-national difference in death from violence in the period of 1963-1982 at the 0.5 level. As expected, the impact of ethnic diversity on death from violence is positive with significant t value for the coefficient estimate. These results suggest that ethnic diversity and violent political behavior are linearly and positively only when death from violence is used as indicator. Hypothesis 1 is thus partially supported. Hypothesis 1.1 suggests that there is a curvilinear (inverted U-shape) relationship between the degree of ethnic diversity and the level of violent political behavior. The subsequent examination of the residual plots after the regressions reveals that there is no identifiable nonlinear relation between ethnic diversity and violent political behavior. To confirm this preliminary finding, we add the quadratic term of ethnic diversity into the regression model. We thus have Equation 6: 2 (6) Y = a + P1 DIV + p2 DIV + 6, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, 6 = stochastic disturbance. The above general equation can be rewritten as the following four specific companion equations: 121 2 (7) In (RIOT/POP +1) = a + px DIV + p2 DIV + e, 2 (8) In (ATT/POP + 1) = a + P1 DIV + P2 DIV + e, 2 (9) In (ASS/POP + 1) = a + pi DIV + p2 DIV + e, 2 (10) DEATH = a + px DIV + P2 DIV + e, where all terms are as in Equation 1 through Equation 5. The regression results are reported in Table 58. For riot, armed attack, and assassination (Equation 7 through Equation 9), none of the F test value are significant enough to allow us to reject the null hypothesis that the constant, and the regression coefficients for ethnic diversity and for the square of ethnic diversity are equal to zero. We further examine the incremental variation explained between Equation 5 and Equation 1, that is, R2 - R2, and look into their related F tests1. The F test for the incremental R2 1. The calculation of the F statistic takes the following formula: F = F k-g,n-(k+l) (R2 - R2)/k-g (1 - R2)/n-(k+l) , where R2 = the R2 of the complete model (here, Equation 5), R^ = the R2 of the reduced model (here, Equation 1), k = the number of the independent variables in the complete model (k = 2 in Equation 5), g = the number of the independent variables in the reduced model (g = 1 in Equation 1), 122 is equivalent to t test for the partial correlation coefficient (Agresti and Agresti, 1979: pp. 383-85). Nonetheless, none of them suggest that the quadratic form of ethnic diversity is necessary. However, the regression results for Equation 10 show different patterns. This regression model with the quadratic term of ethnic diversity is able to explain 5.6% of the variance at the 0.5 significance level. Further, the squares of the partial correlation between the quadratic term of ethnic diversity and death from violence (that is, R2 - R2) is significant at 0.1 level. While the test for the estimated coefficient of squared ethnic diversity is significant at the 0.1 level, the t test for the estimated coefficient of ethnic diversity is no longer significant. Based on the results from Equation 7 through Equation 10, we conclude that the curvilinear relationship between ethnic diversity and violent political behavior suggested in Hypothesis 1.1 is found to exist only when death from violence is used as indicator of the dependent variable. It is noted that the bivariate results from Equation 1 through Equation 5 provide the evidence for Hypothesis 1 without reference to the intervening variables. As we have recognized, ethnic diversity alone is not expected to be overwhelmingly associated with violent political behavior. n = the number of the observations (here n = 132). 123 Thus, the addition of another variable into the regression model may allow us to account for more variation in our dependent variable. Of more important is the fact that we are able to examine the independent impact of ethnic diversity controlling for the influence of the intervening variables. Thus, our work in this section should result in more accurate estimates of the relative importance of the each independent variable. The regression line with the intervening variable of economic development takes the following form: (11) Y = a + pi DIV + p2 ECO + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, ECO = Economic Development, e = stochastic disturbance. As with Equation (1), the above general equation can be rewritten as the following four specific equations: (12) In (RIOT/POP + 1) = a + p1 DIV + P2 ECO + e, (13) In (ATT/POP + 1) = a + Px DIV + P2 ECO + e, (14) In (ASS/POP + 1) = o + Pj DIV + p2 ECO + e, 124 (15) DEATH = a + P1 DIV + P2 ECO + 6, where ECO = Economic Development, and all the other terms are as in Equation 2 through Equation 5. The regression results are shown in Table 59. Only in the case of death from violence (Equation 15) does the F test reject the null hypothesis HQ: Pa = P2 = 0. However, further investigation of the t test for the coefficients reveals that the estimated coefficient of economic development is not significant at the 0.1 level. We further examine the increase in the proportion of explained variation when economic development is added to Equation 1, that is, R2 - R2, and look into their related F tests2. * c r * None of them suggest that the addition of economic development into the regression is necessary. To test Hypothesis 2, we then add the interaction term of ethnic diversity and economic development into Equation 11 and have the following: (16) Y = a + Px DIV + p2 ECO + p3 IECO + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, 2. As we have mentioned earlier, we may also examine the squares of the partial correlation between ethnic diversity and violent political behavior. In such cases, t tests, rather than F tests, are to be conducted. 125 ECO = Economic Development, IECO = (DIV - 0.3342106) * (ECO - 1709.09) e = stochastic disturbance. As with Equation (1), the above general equation can be rewritten as the following four specific equations: (17) In (RIOT/POP +1) = a + P1 DIV + p2 ECO + P3 IECO + e, (18) In (ATT/POP + 1) = a + P1 DIV + P2 ECO + P3 IECO + e, (19) In (ASS/POP + 1) = a + P1 DIV + P2 ECO + P3 IECO + 6, (20) DEATH = a + px DIV + p2 ECO + p3 IECO + e, where the terms are the same as in Equation 2 through Equation 5 and Equation 16. The results are reported in Table 60. The F tests of the overall significance of the regression fail to reject the null hypothesis HQ: P1 - P2 = P3 = 0. None of the F statistics for the incremental R2 owing to the interaction terms is significant at the 0.1 level. Furthermore, none of the t tests for the interaction term is significant at the 0.1 level. Thus, the results do not support Hypothesis 2 that ethnic diversity and economic development have a joint influence on violent political behavior. As with economic development, economic and political 126 discrepancies is added to Equation 1 to analyze whether the linear relationship, if any, between ethnic diversity and violent political behavior is still the same after we control for the value of economic and political discrepancies. The new model takes the following form: (21) Y = a + P1 DIV + p2 DIS + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, DIS = Economic and Political Discrepancies, e = stochastic disturbance. As with Equation (1), the above general equation can be rewritten as the following four specific equations: (22) In (RIOT/POP + 1) = a + P1 DIV + P2 DIS + e, (23) In (ATT/POP + 1) = a + P1 DIV + P2 DIS + 6, (24) In (ASS/POP +1) = a + Px DIV + P2 DIS + e, (25) DEATH = a + px DIV + p2 DIS + 6, where DIS = Economic and Political Development, and all the other terms are as in Equation 2 through Equation 5. 127 As shown in Table 61, only the F test for Equation 25 rejects the null hypothesis H0: P1 = P2 = 0 at the 0.05 level. The F tests for Equation 22 through Equation 24 show that ethnic diversity and economic and political discrepancies do not have significant impacts on riot, armed attack, or assassination at the level of 0.1. Further investigation of the t test for the estimate of economic and political discrepancies indicates that it has no linear relationship with death from violence. We further examine the increased proportion of the variation explained when economic and political discrepancies is added to Equation 1, that is, R2 - R2, and look into their related F tests. While the F statistics for Equation 22 through Equation 24 suggest that the economic and political discrepancies has not significantly improved the explanation in the cross-national variation of riot, armed attack, or assassination, the F test for Equation 25 shows that the degree of economic and political discrepancies has significantly increased the proportion of the variation in death from violence explained in Equation 5 at the 0.05 level. To test Hypothesis 3, we investigate whether the interaction term of ethnic diversity and economic and political discrepancies is helpful in specifying the regression model. The equation is expressed as: 128 (26) Y = a + p1 DIV + p2 DIS + P3 IDIS + 6, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, DIS = Economic and Political Discrepancies, IDIS = (DIV - 0.3342106) * (DIS - 0.2715854) 6 = stochastic disturbance. The above general equation can be rewritten as the following four specific equations: (27) In (RIOT/POP + 1) = a + P1 DIV + P2 DIS + P3 IDIS + 6, (28) In (ATT/POP + 1) = a + Px DIV + P2 DIS + p3 IDIS + e, (29) In (ASS/POP +1) = a + P^^ DIV + p2 DIS + P3 IDIS + e, (30) DEATH = a + p1 DIV + P2 DIS + P3 IDIS + e, where the terms are the same as in Equation 2 through Equation 5 and Equation 26. The test results are shown in Table 62. None of the F statistics for Equation 27 through Equation 29 support the null hypothesis H0: P1 = P2 = P3 = 0. For Equation 30, the F test for the overall fitness of the model is significant at the 0.05 level, and the F statistics for the incremental R2 owing to the interaction terms is significant at the 0.1 129 level. However, none of the t tests suggest that the estimates of the interaction terms are significant. Thus, the results do not support Hypothesis 3 that ethnic diversity has different influence on violent political behavior within different degrees of economic and political discrepancies. The next step is to assess the impacts of the intervening variable governmental coercion on the relationship between ethnic diversity and violent political behavior. The regression is estimated in the following general and four specific forms: (31) Y = a + px DIV + p2 COER + e, (32) In (RIOT/POP + 1) = a + Pa DIV + P2 COER + e, (33) In (ATT/POP + 1) = a + px DIV + P2 COER+ e, (34) In (ASS/POP + 1) - a + Px DIV + P2 COER + 6, (35) DEATH = a + P1 DIV + P2 COER + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, COER = Governmental Coercion, e = stochastic disturbance, 130 and all the other terms are as in Equation 2 through Equation 5. The regression results are indicated in Table 63. In the cases of riot, armed attack, and assassination, the overall fitness of the model is significant at the 0.01 level; and in the case of death from violence, the overall significance of regression is marginal (P = 0.109). Except in Equation 35, the F tests for the incremental proportion of the variation explained are significant at the 0.01 level in Equation 32 through Equation 34 where the t tests for the estimates of governmental coercion are also significant at the level of 0.01. Hypothesis 4 is tested against the data in the following regression models: (36) Y = a + px DIV + p2 COER + p3 ICOER + 6, (37) In (RIOT/POP + 1) = a + Px DIV + p2 COER + p3 ICOER + e, (38) In (ATT/POP + 1) = a + Px DIV + P2 COER + p3 ICOER + 6, (39) In (ASS/POP + 1) = a + px DIV + P2 COER + p3 ICOER + e, (40) DEATH = a + px DIV + p2 COER + P3 ICOER + e, where Y = Violent Political Behavior 1963-1982, 131 DIV = Ethnic Diversity, COER = Governmental Coercion, ICOER = (DIV - 0.3342106) * (COER - 0.0321793) e = stochastic disturbance, and the other terms are the same as in Equation 2 through Equation 5. The results are shown in Table 64. For Equation 37 and Equation 40, the t tests for the interaction term of ethnic diversity and governmental coercion are not significant, suggesting that neither the relationship between ethnic diversity and riot nor that between ethnic diversity and death from violence is dissimilar within different levels of governmental coercion. On the other hand, the t statistics for the estimates of the interaction term in both Equation 38 and Equation 39 are significant at the 0.01 and 0.05 respectively. Further investigation of the direction of the regression coefficients indicates that the interaction term has different impacts on different forms of violent behavior. The positive sign of the estimate suggests that in states with a higher level of governmental coercion, ethnic diversity is more conducive to armed attack; on the other hand, the negative sign of the estimate reveals that in states with a lower level of governmental coercion, ethnic diversity is more conducive to assassination. In conclusion, Hypothesis 4 is supported only when 132 assassination is used as indicator of violent political behavior. If armed attack is observed, the interaction pattern is opposite to what the hypothesis predicts; and if riot or death from violence is used as indicator of violent political behavior, the hypothesis is not supported. A set of five regression equations are constructed to evaluate whether the addition of the role of ethnic elites into the model will increase the explanation of cross- national variation in violent behavior. (41) Y = a + px DIV + p2 ELT + e, (42) In (RIOT/POP +1) = a + Pa DIV + P2 ELT + e, (43) In (ATT/POP +1) = a + px DIV + P2 ELT+ e, (44) In (ASS/POP + 1) = a + Px DIV + p2 ELT + e, (45) DEATH = a + px DIV + P2 ELT + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, ELT = Role of Ethnic Elites, e = stochastic disturbance, and all the other terms are the same as in Equation 2 through Equation 5. 133 As shown in Table 65, only the F test for Equation 45 rejects the null hypothesis HQ: Pa = P2 = 0 (P < 0.01). The F tests for Equation 42 through Equation 44 show that ethnic diversity and role of ethnic elites do not have any significant impact on violent political behavior at the level of 0.1. The t tests for the estimates of coefficients reveals that only death from violence has a linear relationship with role of ethnic elites. We then examine the increased proportion of the variation explained when role of ethnic elites is added into the model, that is, R2 - R2, and look into their related F tests. While the F statistics for Equation 42 through Equation 44 suggest that the role of ethnic elites has not significantly improved the explanation in the cross-national variation of riot, armed attack, or assassination, the F test for Equation 45 shows that the role of ethnic elites has significantly increased the proportion of the variation in death from violence explained in Equation 5. To test Hypothesis 5, we add the interaction term of ethnic diversity and role of ethnic elites into Equation 41 and have the following: (46) Y = a + px DIV + P2 ELT + p3 IELT + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, 134 ELT = Role of Ethnic Elites, IELT = (DIV - 0.3342106) * (ELT - 0. 3756098) e = stochastic disturbance. The above general equation can be rewritten as the following four specific equations: (47) In (RIOT/POP + 1) = a + Pa DIV + p2 ELT + P3 IELT + e, (48) In (ATT/POP + 1) = a + P1 DIV + p2 ELT + P3 IELT + e, (49) In (ASS/POP + 1) = a + px DIV + P2 ELT + P3 IELT + e, (50) DEATH = a + Px DIV + p2 ELT + P3 IELT + 6, where the terms are the same as in Equation 2 through Equation 5 and Equation 46. The statistic results are presented in Table 66. The regression results for Equation 47 through Equation 49 are not roust at all: the F tests of the overall significance of the regression fail to reject the null hypothesis HQ: P1 = P2 = P3 = 0; and none of the F statistics for the incremental R2 owing to the interaction terms is significant at the 0.1 level. None of the three independent variables feature significant t test values. In particular, the interaction term has had little direct influence on riot, armed attack, or assassination. 135 On the other hand, the regression for Equation 50 is ale to account for 57.4% of the cross-national difference in death from violence. Moreover, the t statistic for the estimate of the interaction term is significant at the 0.01 level, suggesting that the influence of ethnic diversity on death from violence tends to increase when ethnic elites play a more active role in ethnic mobilization. We conclude that Hypothesis 5 is supported only when death from violence is used as indicator of violent political behavior. The next step is to ascertain whether the inclusion of the intervening variable geographical concentration will have any impact on the relationship between ethnic diversity and violent political behavior. The regression is estimated in the following general and four specific forms: (51) Y = a + Px DIV + p2 GEO + 6, (52) In (RIOT/POP +1) = a + px DIV + P2 GEO + e, (53) In (ATT/POP + 1) = a + pa DIV + P2 GEO + e, (54) In (ASS/POP + 1) = a + Px DIV + p2 GEO + e, (55) DEATH = a + Px DIV + P2 GEO + e, where Y = Violent Political Behavior 1963-1982, 136 DIV = Ethnic Diversity, GEO = Geographical Concentration, e = stochastic disturbance, and all the other terms are as in Equation 2 through Equation 5. As can be seen in Table 67, the overall fitness of the model is not significant at the 0.1 level in the cases of riot, armed attack, and assassination, although the F tests for the incremental proportion of the variation explained are significant at the 0.01 level in Equation 53, where the t test for the estimates of geographical concentration is also significant at the level of 0.1. In the case of death from violence, although the regression is significant at the 0,1 level, the t test for the estimate of geographical concentration is not significant. We test Hypothesis 6 against the data by adding the interaction term of ethnic diversity and geographical concentration in the above regression lines. Five equations are constructed: (56) Y = a + px DIV + p2 GEO + P3 IGEO + 6, (57) In (RIOT/POP + 1) = a + Px DIV + p2 GEO + P3 IGEO + e, (58) In (ATT/POP +1) = a + px DIV + P2 GEO + P3 IGEO + 6, 137 (59) In (ASS/POP +1) = a + Px DIV + P2 GEO + P3 IGEO + e, (60) DEATH = a + px DIV + p2 GEO + p3 IGEO + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, GEO = Geographical Concentration, IGEO = (DIV - 0.3342106) * (GEO - 0.5075758) e = stochastic disturbance, and the other terms are the same as in Equation 2 through Equation 5. The results are shown in Table 68. Only in Equation 60 is the t test for the interaction term of ethnic diversity and geographical concentration significant, suggesting that the relationship between ethnic diversity and death from violence is dissimilar within different levels of geographical concentration. As we have hypothesized, the positive sign of the estimate of the interaction term suggests that in states with a higher level of geographical concentration, ethnic diversity is more conducive to death from violence. We conclude that Hypothesis 6 is supported in so far as death from violence is used as indicator of violent political behavior. If the other measures are used, the hypothesis is not substantiated. A set of five regression equations are tested against the data to evaluate whether the addition of the role of 138 external links into the original bivariate model will increase the explanation of cross-national variation in violent political behavior. (61) Y = a + px DIV + p2 EXT + e, (62) In (RIOT/POP + 1) = a + Px DIV + P2 EXT + 6, (63) In (ATT/POP + 1) = a + Pa DIV + P2 EXT+ 6, (64) In (ASS/POP + 1) = a + Pa DIV + P2 EXT + e, (65) DEATH = a + px DIV + P2 EXT + e, where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, EXT = External Links e = stochastic disturbance, and all the other terms are the same as in Equation 2 through Equation 5. The regression results are presented in Table 69. Except in Equation 63, the F tests reject the null hypothesis H0: Px = P2 - 0, showing that ethnic diversity and external links together do have statistically significant impacts on riot, assassination and death from violence but not on armed attack. Close examination of the t tests for 139 the estimates of coefficients for external links reveals that only riot and assassination have a linear relationship with external links respectively. We further examine the increased proportion of the variation explained when the intervening variable external links is added to Equation 1, that is, R2 - R2, and look into their related F tests. While the F statistics for Equation 62 and Equation 64 suggest that external links have improved the explanation in the cross-national variation of riot and assassination, the F test for Equation 63 and Equation 65 shows that external links fail to increase the proportion of the variation in armed attack and in death from violence explained. To test Hypothesis 7, we finally add the interaction term of ethnic diversity and external links into Equation 61 and have the following: (66) Y = a + p2 DIV + p2 ExT + p3 IEXT + e, (67) In (RIOT/POP + 1) = a + Px DIV + p2 EXT + p3 IEXT + e, (68) In (ATT/POP + 1) = a + Px DIV + P2 EXT + P3 IEXT + e, (69) In (ASS/POP + 1) = a + px DIV + P2 EXT + p3 IEXT + e, (70) DEATH = a + Px DIV + P2 EXT + p3 IEXT + 6, 140 where Y = Violent Political Behavior 1963-1982, DIV = Ethnic Diversity, EXT = External Links, IEXT = (DIV - 0.3342106) * (ELT - 1.9545455) e = stochastic disturbance, and the other terms are the same as in Equation 2 through Equation 5. The results are presented in Table 70. Only the F test of the overall significance of the regression for Equation = 67 rejects the null hypothesis EQ: Px = P2 = P3 0/ with 6.1% of the variance in riot is accounted for. None of the F statistics for the incremental R2 owing to the interaction terms, including the one for Equation 67, are significant at the 0.1 level. Further investigation of the t statistics for the estimates of the interaction term indicates that none of them are significant at the 0.01 level, which suggests that the impact of ethnic diversity on violent political behavior is not dissimilar within different levels of external links. In conclusion, our regression tests do not provide significant statistical evidence to support Hypothesis 7. The above regression results are summarized in Table 71, which, it is noted, only gives us how well the independent and intervening variables in explaining the variance of the dependent variables. Unequivocal demonstrations of the performance of the interaction terms 141 TABLE 71. Summary of the Regression Results8 Dependent Variable Inde pendent Armed Assassi- Death from Variable Riot Attack nation Violence DIV ** DIV2 DIV, ECO DIV, ECO, ECO2 DIV, ECO, IECO DIV, DISQ ** DIV, DISQ, IDISQ - ** DIV, COER *** *** *** DIV, COER, COER2 *** *** *** DIV, COER, ICOER *** *** *** DIV, ELT *** DIV, ELT, IELT *** DIV, GEO DIV, GEO, IGEO ** DIV, EXT ** DIV, EXT, IEXT ** a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 142 are shown in Figure 8 through Figure 11. Hypothesis 1 that there is a linear and positive relationship between ethnic diversity and violent political behavior is moderately supported only when death from violence is used as indicator of the dependent variable. Although the estimate for the quadratic form of ethnic diversity is marginally significant in accounting for the cross-national variation in death from violence, the positive sign of the regression coefficient, suggesting a U- shape curvilinear relationship, fails to support Hypothesis 1-1 that there exists an inverse U-shaped curvilinear relationship between ethnic diversity and violent political behavior. Hypothesis 2 that economic development interacts with the relationship between ethnic diversity and violent political behavior is not supported as none of the estimates of the four interaction terms is statistically significant. Although the regression lines containing economic and political discrepancies are statistically significant when death from violence is used, neither the estimate of this intervening variable nor that of its interaction term is found to have any explanatory utility. Rather, the variance in death from violence is explained by ethnic diversity. Accordingly, Hypothesis 3 is rejected. Earlier, we have suggested in Hypothesis 4 that governmental coercion would suppress the relationship 143 Intervening Variables Economic Economic and Development Political Discrepancies Ethnic Diversity Geographic Concentration Intervening Variables FIGURE 8. Final Hodel of Riot from Regressions 144 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Intervening Variables FIGURE 9. Final Hodel of Armed Attack from Regressions 145 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Geographic External Concentration Links Intervening Variables FIGURE 10. Final Hodel of Assassination from Regressions 146 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Geographic External Concentration Links Intervening Variables FIGURE 11. Final Hodel of Death from Violence from Regressions 147 between ethnic diversity and violent political behavior. Our test results give disparate messages when different indicators of violent political behavior are employed. For riot and death from violence, the interaction impact of governmental coercion is not found either substantially or statistically. As we have expected, the negative sign of the estimate of the interaction term confirms that governmental coercion would weaken the relationship between ethnic diversity and assassination. On the contrary, the positive sign of the estimate signifies that governmental coercion would reinforce the positive and linear relationship between ethnic diversity and armed attack. Although role of ethnic elites is positively related to death from violence, it is found to interact with the relationship between ethnic diversity and death from violence. Thus, Hypothesis 5 is partialy supported. While geographic concentration of ethnic groups in itself is positively associated with armed attack, it also interact with the relationship between ethnic diversity and death from violence, given the positive sign of the estimate of its interaction term. Hypothesis 6 is thus partially supported in so far as death from violence is used as indicator of violent political behavior. While the variable external links of ethnic groups has direct impacts of the occurrence of riot and assassination, it is not found to influence the relationship between ethnic 148 diversity and any of the four indicators of violent political behavior. Thus Hypothesis 7 is not substantiated. In summary, while ethnic diversity is positively related to death from violence, the quadratic relationship is U-shaped rather than the inverse U-shaped one as we have hypothesized. There is no relationship between ethnic diversity and violent political behavior, and the interaction effect is not found. Governmental coercion is positively related to violent political behavior, but the interaction effects are not uniform: while governmental coercion will inhibits ethnic diversity to cause assassination, it will intensify armed attack induced by ethnic diversity. The role of ethnic elites is positively related to death from violence, and it also reinforce ethnic diversity to cause death from violence as hypothesized. Geographical concentration of ethnic groups is positively related to death from violence, and it also intensify the tendency of ethnic diversity to cause armed attack as we have hypothesized. Finally, while external link of ethnic groups is positively related to riot and assassination, the interaction effect is not founded. The above discussions of the results from cross- tabulations and from regressions are summarized in Figure 12 and Figure 13. For the purpose of information, we have also included all variables in the regressions (without and with the interaction terms). As shown in Table 84 and Table 85, 149 Intervening Variables Economic Economic and Governmental Role of Development Political Coercion Ethnic Discrepancies Elites Intervening Variables FIGURE 12. Final Hodel From Cross-tabulations 150 Intervening Variables Economic Economic and Governmental Development Political Coercion Discrepancies Ethnic Diversity Intervening Variables FIGURE 13. Final Hodel From Regression 151 TABLE 84. Regression Results of Violent Political Behavior on Ethnic Diversity and All Intervening Variables0 Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.004 -0.015 2.6xl0"4 26352 DIV -0.016 -0.052 -0.009 -38202 (-.81) (-0.36) (-0.91) (-0.27) ECO 6x10-7 4xl0-6 5xl0"7 4.325 (-0.43) (0.41) (-0.76) (0.41) DISQ -5.6xl0"4 -0.003 -6.8xl0-4 -24513 (-0.10) (-0.07) (-0.25) (-0.59) COER 0.375 1.232 0.087 11263 (13.60)*** (6.19)*** (6.69)*** (0.06) ELT -0.002 0.027 -0.001 257480 (-0.41) (0.66) (-0.45) (6.38)*** GEO 0.011 0.060 0.005 -64745 (1.45) (1.12) (1.38) (-1.21) EXT 3.5xl0"4 -0.006 0.003 -19173 (0.16) (-0.40) (0.27) (-1.21) R2 0.741 0.375 0.426 0.417 P Value 0.000*** 0.000*** 0.000*** 0.000*** N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 152 TABLE 85. Regression Results of Violent Political Behavior on Ethnic Diversity, All Intervening Variables, and All Interactional Terms" Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant lxlO'4 -0.128 0.001 95765 DIV 0.007 0.185 5.6xl0"4 -30417 (0.21) (1.00) (0.04) (-0.18) ECO -7xlQ"7 -9xl0"7 -6xl0"7 0.426 (-0.44) (-0.09) (-0.76) (0.05) DISQ -lxlO-6 -4xl0"4 -7.5xl0"4 -2833 (-0.00) (0.01) (-0.22) (-0.07) COER 0.345 3.290 0.060 -635480 (5.09)*** (8.12)*** (1.87)* (-1.70)* ELT -0.007 0.020 -0.004 76701 (-0.76) (0.36) (-0.83) (1.54) GEO 0.005 0.011 0.002 -2923 (0.17) (1.22) (0.15) (-0.04) EXT 5xl0"4 0.022 2xl0"3 -47912 (0.17) (1.22) (0.15) (-2.87)*** IECO 4xl0"6 lxlO"5 2xl0"6 6.810 (0.55) (0.21) (0.45) (0.15) IDISQ -0.002 -0.136 0.002 -301325 (-0.05) (-0.77) (0.17) (-1.84)* ICOER -0.180 10.050 -0.150 -2324869 (-0.59) (5.50)*** (-1.05) (-1.38) IELT 0.024 -0.039 0.012 856264 (0.88) (-0.24) (0.95) (5.67)*** IGEO -0.070 -0.300 -0.031 -109661 (-1.14) (-0.81) (-1.06) (-0.32) IEXT 0.006 0.139 0.002 -317176 (0.38) (1.43) (0.28) (-3.52)*** 153 TABLE 16. (contiuned) R2 0.751 0.587 0.454 0.674 P Value 0.000*** 0.000*** 0.000*** 0.000*** R2 - R2 0.010 0.212 0.029 0.257 c r F Value 0.43 3.85*** 0.57 5.00*** N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 154 we find that governmental coercion has the strongest impacts on riot, armed attack, and assassination, while the role of elite has the strongest impacts on death from violence. As for the interaction relationship hypothesized, governmental coercion would reinforce ethnic diversity to case armed attack. And while the role of ethnic elites does intensify the tendency of ethnic diversity to cause death from violence, economic and political discrepancies and external links of ethnic groups would reduce that tendency, which is opposite to what we have hypothesized. Why did not the negative interaction relationships show up earlier in the regressions with one intervening variable considered at one time? In the process of regression tests, we have carefully constructed each interaction term in such a way that it is not linearly related to either of its components. For instance, IDISQ, the interaction term for economic and political discrepancies, is made up of the multiplicity of (DIV - 0.3342106) and (DIS - 0.2715854), where 0.3342106 is the mean value of ethnic diversity and 0.2715854 is the mean of economic and political discrepancies (see the regression equation on page 128). The other interaction terms are similarly constructed. Nonetheless, while some interaction terms are thus not linearly related to their composite intervening variables, some are still highly correlated with their composite intervening variables. In other words, the standardizing 155 process of the intervening variables before they are combined into any interaction term may not be adequate. Perhaps, each component of the interaction needs to be divided by the standard deviation of the two composite intervening variables. Thus, a more satisfactory interaction term of economic and political discrepancies together with ethnic diversity, for example, would take the following form: (71) IDIS = [(DIV-0.3342106)/0.21]*[(DIS-0.2715854)/0.52] where 0.21 nd 0.52 are the standard deviation of ethnic diversity and economic and political discrepancies respectively. Even if the problem of multicolinearity between the interaction term and its two composite variables is solved, we have to guard against any possible correlation between any of the two interaction terms when all variables and interaction terms, even if modified, are included in one regression line. (See Table 86 in Appendix I for the Pearson correlation coefficients among all variables and interaction terms.) Accordingly, we shall be very cautious when drawing inference from the regression analyses. CHAPTER VII SUMMARY AND CONCLUSIONS A. Summary of the Study We started with the recognition of the widely shared contention that the existence of ethnic diversity is to be blamed for most tragedies of violent political behavior. And yet, after briefly reviewing the literature on ethnic relationships, we were surprised to find that there had been few efforts at theorizing their causal relationships and subjecting them to systemic empirical tests. We believed that with the construction of a general model tested against a more extensive data set, the presumed causal relationship between ethnic diversity and violent political behavior would be illuminated. Our present study was the first cross-national test of the above-mentioned relationships within a theoretical framework. We introduced a general model of violent political behavior. It was hypothesized that there was a linear and positive relationships between ethnic diversity and violent political behavior. Two schools of thought were solicited to buttress this causal link — Wilsonian/Value and Psychological perspectives. The former school took two 156 157 forms: A Wilsonian view, arguing for self-determination, contended that only ethnically homogeneous countries were compatible to democracy and peace while ethnically heterogeneous countries were invariably conflict-prone. A refined value perspective argued that ethnic conflicts were the result of disagreement over different values systems in connection with diverse ethnic groups within the country. The later psychological school, more sophisticated and less deterministic, suggested that conflict was the consequence of persistent deprivation of cherished political, economic, or cultural rights. As we held the belief that mono-causal explanation was too deterministic to offer adequate understanding of the complicated relationships involved, theories were sought to guide the selection of possible intervening variables that would serve to exacerbate or relieve the presumed detrimental impacts of ethnic diversity. Nine potentially immediate causes of deprivation/frustration were incorporated in a preliminary assessment: economic development, economic and political discrepancies, governmental coercion, role of elites, geographical concentration of ethnic groups, external links of ethnic groups, process of modernization, favorable international conditions, and historical explanations. After looking into relevant treatises on these intervening variables, only the former six intervening 158 variables were retained in the revised model. Process of modernization was dropped out because of theoretical ambiguity and empirical complexity. The variable favorable international conditions was omitted from the model because we found these conditions to be constants rather than variables unless observed across time. Finally, we discarded historical explanations because they were too deterministic to offer any insights amendable for cross- national comparison. Recognizing the difficulty in directly measuring the psychological state of deprivation/ frustration, we decided to leave out this intermediate stage in the working model for testing. The methods employed in this study were cross- tabulations and bivariate and multivariate regressions with Pearson correlation coefficients for the preliminary investigation. They were chosen for their relative ease in interpreting the interaction relationships embedded in most of our hypotheses. The unit of analysis being the state rather than the ethnic group, the sample, limited by datum availability, had 132 countries. While inclined to define ethnicity in such a broad way as to include identity based upon linguistic, religious, and racial differences, we were handicapped by data obtainable and forced to examine only two most common facets of ethnic identity: language and religion. The concept diversity was constructed to consider both the number of ethnic groups 159 within a country and their relative sizes. Violent political behavior was represented by two dimensions: its level and magnitude, each being measured by riot, armed attack, and assassination, and by death from violence, respectively, for the 1963-1982 period. The level of economic development was indicated by per capita Gross National Product in 1978. The concept economic and political discrepancies was the average of ordinal measures of economic discrimination and political discrimination pertaining to ethnic groups. Governmental coercion was measured by the imposition of political sanction by the government. The role of ethnic elites was derived from the multiplication of ordinal measure of the separatist potential of ethnic groups by the proportion of these groups. Geographical concentration of ethnic groups was a crude dummy variable largely based on our compilation of the existence of territorial based ethnic groups or ones with separatist potential within a country. The concept external links of ethnic groups was gauged by an ordinal measure of group dispersion across states. B. Findings and Prospects Findings Statistical results from cross-tabulations fail to provide any evidence that ethnic diversity and violent 160 political behavior are associated as suggested in Hypothesis 1. Although economic development is found to have an interaction relationship with ethnic diversity and death from violence, its exacerbating influence is opposite to what we expected in Hypothesis 2. Similarly, while there is an interaction relationship among economic and political discrepancies, ethnic diversity, and assassination, the mitigating effect contradicts ordinary wisdom. Thus Hypothesis 3 is also rejected. Governmental coercion exercises disparate impacts on the relationship between ethnic diversity and violent political behavior. While governmental coercion is found to inhibit the detrimental effect of ethnic diversity on armed attack as we have hypothesized, it, on the contrary, tends to exacerbate the negative influence of ethnic diversity on riot. In other words, governmental coercion is useful in containing armed attacks but not more spontaneous riots. Thus, Hypothesis 4 is partially confirmed. Hypothesis 5 is not supported since the role of ethnic elites does not have any impact on the relationship between ethnic diversity and any one of the four indicators of violent political behavior. Geographical concentration of ethnic groups is found to intensify the relationships between ethnic diversity and riot, armed attack, and death from violence. In other words, except assassination, the other three facets of 161 violent political behavior are all even more susceptible to the potentially ill effect of ethnic diversity when the ethnic groups are geographically concentrated. Hypothesis 6 is thus strongly supported. Finally, we discover that there is no interaction relationship among external links of ethnic groups, ethnic diversity, and violent political behavior since there is no change in the relationship between the latter two after we control for the value of external links. Hypothesis 7 is thus not substantiated. . Statistical results from regressions, being more sophisticated and stringent, support fewer hypotheses than those from cross-tabulations do. Ethnic diversity is found to be linearly and positively related only with death from violence. Hypothesis 1 is thus minimally substantiated. Further investigation does confirm that ethnic diversity and death from violence are curvilinearly related. Nonetheless, the U-shaped relationship is opposite to what we have looked for. Hypothesis 1.1 is thus rejected. While there is no linear relation between economic development and violent political behavior, the hypothesized inverse U-shaped relationship between them is not supported either. Similarly, there is no linear relationship between economic and political discrepancies and violent political behavior. Neither economic development nor the variable economic and political discrepancies has any impact on the 162 relationship between ethnic diversity and violent political behavior. Hypotheses 2 and 3 are thus not supported. While governmental coercion is found to intensify the relationship between ethnic diversity and armed attack, it would lessen the relationship between ethnic diversity and assassination. In other words, although governmental coercion may diminish the relationship between ethnic and assassination, it may intensify the impact of ethnic diversity on armed attack. Thus Hypothesis 4 is partially supported. It is noted that there does exist a curvilinear relationship between governmental coercion and violent political behavior. However, the inverse U-shaped relationship is opposite to what we have hypothesized. While the role of ethnic elites is positively related to violent political behavior, it also has an interaction relationship with ethnic diversity and violent political behavior in so far as death from violence is employed as indicator of the latter. Hypothesis 5 is thus modestly supported. Finally, both geographical concentration and external links of ethnic groups are found to be positively related to violent political behavior. However, they show different interaction impacts. Since geographical concentration of ethnic groups is observed to aggravate the relationship between ethnic diversity and death from violence, Hypothesis 6 is thus minimally supported. Finally, Hypothesis 7 is not 163 confirmed because none of the four interaction terms pertaining to external links and relevant measures of violent political behavior is statistically significant. In summary, when both cross-tabulations and regressions are taken into account, we confirm that ethnic diversity is associated with death from violence and that there is an inverse U-shaped relationship between ethnic diversity and violent political behavior. Although economic development is negatively related to violent political behavior as hypothesized, its interaction term is positively related to violent political behavior. While the variable economic and political discrepancies is positively related to violent political behavior, its interaction term is negatively related to violent political behavior. Governmental coercion is found to be correlated with violent political behavior, but its interaction relationships with ethnic diversity differ for different dimensions of violent political behavior: whereas governmental coercion may inhibit assassinations, it will provoke riots. While the role of ethnic elites is positively related to violent political behavior, its interaction term is also positively related to violent political behavior. While geographical concentration of ethnic groups is positively related to violent political behavior, its interaction is also positively related to violent political behavior. While the external links of ethnic links is positively related to 164 violent political behavior, there is no sign showing that it has any influence on the relationship between ethnic diversity and violent political behavior. Discussions of Findings Why is the relationship between ethnic diversity and death from violence U-shaped rather than inversely U-shaped as we have hypothesized? One plausible explanation is that ethnic violence is usually waged by a tiny minority group. At the same time, the central government may calculate that it is much easier to pacify a small ethnic group in remote areas than to contain a larger ethnic group or many minority groups at a time. Accordingly, even though the degree of ethnic diversity is negligible in such a country, the magnitude of violence may be very extensive. With the increase of the degree of ethnic diversity, the magnitude of violence may decrease because there are more and more ethnic groups to cope with. It would be costly to wage violence with many ethnic groups at one time for both the central government and the ethnic groups. Nevertheless, when the downward slope of the curve of ethnic diversity reaches its bottom, the expected positive relationship between ethnic diversity and violence begins to show up. In other word, with the multiplicity of many ethnic groups, it is more difficult to reach consensus on any issue. More violence is thus expected. 165 Why does economic development reinforce the conflict potential of ethnic diversity rather than alleviate it? One possible explanation is that with the expansion of economic development, members of the ethnic groups may begin to demand something other than materials, such as the cultural heritages of the ethnic groups or even the quest for territorial autonomy, which may have been heretofore neglected by them or suppressed by the central government. Guarded with economic affluence, ethnic groups may be more daring to demand what they deem due. The expansion of economic activities also suggests more contacts between members belonging to different ethnic groups. It also means there are more economic gains to compete for. In other words, although ethnic groups may compete for scarce resources in periods of economic recession, the competition may be more intensive in time of prosperity. Under such a condition, ethnic diversity is more conducive to violent political behavior. Why is ethnic diversity less conducive to violent political behavior in countries with a higher degree of economic and political discrepancies than in ones with a lower degree of discrepancies? This seemingly counter intuitive finding may be due to habitual perceptions. Ethnic groups in the disadvantaged region may perceive the discrepancies more strongly in the beginning even though the differences in the distribution of economic resources and 166 political power are small. But when the discrepancies widen after a persistent period of time, they may no longer maintain the original state of discontent as they may have taken their disadvantaged position for granted. Accordingly, the degree of deprivation may be in reality higher at this point, but the level of frustration may actually decrease. Why does a. higher degree of governmental coercion lead to a higher level of violent political behavior as measured by riot, armed attack, and assassination? As we have discussed earlier, governmental coercion may inhibit violent political behavior through threatening the use of punishment. However, what we have observed is actual, rather than threatened, governmental coercion. Therefore the degree of resentment felt by ethnic groups being punished by the central government may be exacerbated as the coercive activities waged upon them intensify. This is confirmed by the case of Iran, where revolts in the frontier regions seemed to be the results of the coercive efforts at nation-building made by the central government. Finally, why is it that in countries where ethnic groups have more external ties ethnic diversity is no more conducive to violent political behavior than in countries where ethnic groups have fewer external ties? One plausible explanation is that a country with ethnic groups tied to more neighboring countries may not necessarily have a higher 167 degree of ethnic diversity. For instance, an ethnic group may be spread in several countries but constitute a small proportion of the population in every country. Therefore, external ties of ethnic groups may facilitate violent political behavior, but not contribute to violent potential for ethnic diversity. Prospects While the empirical results of this study have largely confirmed our hypotheses, we would like to replicate the tests by comparing data for the time period 1963-1972 with 1963-1982 as listed in Appendix D. Most of the African and Asian countries got independence in the former period while counties in the Western World began to experience ethnic unrest. We thus expect elements of the model would have different patterns of relationships in the two periods. Efforts should be made to overcome limitations of available data. As we have noted earlier, cross-national event data pertaining to ethnic conflict are either incomplete or available for a time frame too short to offer enough variations across countries. In this study, consequently, we have to rely on data of violent political behavior as proxy dependent variables. The adequacy of our model will be more rigorously evaluated if it can be tested against a data set of ethnic conflict. It is time to expand and update the data set of minority conflict compiled by 168 Feierabend et al. (no date) for 1955-1965. Although we have largely confirmed what we have hypothesized, any improvement in the measures of the independent and intervening variables would make the test results of the model more robust since measurement errors in these variables also present a potential threat to the validity of our findings. As we have recognized, some indicators of these variables are indirectly measured, such as that of the role of ethnic elites and that of external links of ethnic groups. Since we are unable to look into how these measures were originally collected, there is no way to verify their validity and reliability. Since the measure of the role of ethnic elites is partially made up of the percentage of the ethnic groups that are pursuing separatism, it is thus highly correlated with our measure of ethnic diversity, the Pearson correlation coefficient being 0.59. (See Table 2 on page 92.) For a better measure of the role of ethnic elites, we need to identify the existence of ethnic cultural clubs, ethnic parties, and separatist movements. We need to ascertain their agendas concerning ethnic interests by content-analyzing either interviews or their writings that can be compiled from newspapers, popular magazines, or campaign leaflets. So far, we have failed to incorporate the subjective dimension of ethnic consciousness, which may be more 169 important in determining willingness to defend ethnic interests. Similarly, we have not tapped the concept of deprivation/frustration. To arrive at such measures, a large scale survey is needed for each country. Unfortunately, such a tremendous task is not foreseen in a global fashion in the immediate future. Meanwhile, we may like to select certain countries as test cases, such as Sri Lanka, Malaysia, and Belgium . We have to admit unashamedly that our model, parsimonious as it is, fails to incorporate possible reciprocal relationships between the variables under investigation. For instance, we would expect that the activities of the government and of the ethnic elite can reinforce or mitigate each other in a feedback loop, which would demand the construction of a causal model, which in turn would require the compilation of data of a higher order. This is a problem of model specification. As we have suggested earlier, more sophisticated techniques, such as causal analysis and confirmatory factor analysis may be worthwhile trying. Finally, we have so far avoided discussion of the coterminality or cross-cutting cleavages among various facets of ethnicity. In other words, we have combined linguistic and religious diversity in an additive fashion. And yet, the two, as well as other ethnicity characteristics, may mutually reinforce one another or may 170 cancel one another out. Thus, in addition to ethnic diversity, other facets of ethnicity, such as coterminality, need to be taken into account. Before any comprehensive compilation of all dimensions of ethnicity is conducted, we may begin the analysis of the phenomenon of coterminality, in the minimum, between linguistic and religious characteristics of ethnic identity. Admittedly, the concept of coterminality is ambiguous enough to promise its immediate delineation. Again, we may start its examination in some restricted cases with prominent indications of such a coterminality. We plead guilty if our test results are interpreted by some policy makers as a rationale for measures which would eradicate elements of ethnic diversity. For instance, some may judge that since geographical concentration of ethnic groups may intensify violence potential of ethnic diversity, they may expect to prevent the zeal of separatism if the ethnic groups are expelled from their home lands and spread throughout the country. This is what the central government did to the Kurds in Iraq. However, as we have shown, repressive measures imposed by the government may contain assassinations, but they may also cause riots and armed attacks. It appears that governmental coercion can only lead to ethnic violence rather than suppress it. Perhaps ethnic identity can be diminished by governmental policies which 171 encourage the integration of ethnic groups, but they cab never be eradicated by force. The embrace of a tolerant attitude of pluralism is warranted. Fortunately, we have found that the role of ethnic elites does matter. Very important in multiethnic countries is whether ethnic elites accept life in a multiethnic state. If they decide that their group interests can be best protected within the same state as other ethnic groups, they must work with them in creating a consensus as to how their welfare can be maximized altogether. This seems to be what the Czechs and the Slovaks have realized in the aftermath of the Soviet coup in August 1991. If one group believes that separation is the only way to guarantee its welfare, forced marriage can only jeopardize the interests of both parties. This is what the ethnic conflict in Yugoslavia has shown the world. With the independence of Latvia, Estonia, and Lithuania and their worldwide recognition, we believe the idea of self- determination has begun to show its fruit. LIST OF REFERENCES Abu-Laban, Baha, and Donald Mottershead. 1981. "Cultural Pluralism and Varieties of Ethnic Politics." Canadian Ethnic Studies, Vol. 13, No. 3, pp. 44-63. Agresti, Alan, and Barbara Finlay Agresti. 1979. Statistical Methods for the Social Sciences. San Francisco, Calif.: Dellen. Ake, C. 1974. "Modernization and Political Instability: A Theoretical Explanation." World Politics. Vol. 26, No. 3, pp. 576-91. Allardt, Erik. 1979. Implications of the Ethnic Revival in Modern Industrial Society: A Comparative Study of Linguistic Minorities in Western Europe. Helsinki: Societas Scientiarum Fennica. Almond, Gabriel A., and Sidney Verba. 1965. Civic Culture. Boston, Mass.: Little, Brown and Co. Asher, Herbert B. 1983. Causal Modeling. 2nd ed. Beverly Hills, Calif.: Sage Publications. Attalides, Michael A. 1979. Cyprus: Nationalism and International Politics. New York: St. Martin's Press. Anderson, Charles W., Fred R. van der Mohden, and Crawford Young. 1967. Issues of Political Development. Englewood Cliffs, N.J.: Prentice-Hall. Babbie, Earl R. 1979. The Practice of Social Research. 2nd ed. Belmont, Calif.: Wadsworth Publishing Co. Bagley, C. 1972. "Racism and Pluralism: A Dimensional Analysis of Forty-Eight Countries." Race, Vol. 13, No. 3, pp. 347-56. Banks, Arthur S., and Robert B. Texter. 1963. A Cross- Polity Survey. Cambridge, Mass.: MIT Press. Banuazizi, Ali, and Myron Weiner. 1986. The State, Religion, and Ethnic Politics: Afghanistan. Iran, and Pakistan. Syracuse, N.Y.: Syracuse University Press. 172 173 Barrett, David B., ed. 1982. World Christian Encyclopedia. Oxford: Oxford University Press. Barrows, Walter L. 1976. "Ethnic Diversity and Political Instability in Black Africa." Comparative Political Studies, Vol. 9, No. 2, pp. 139-70. Bartelsen, Judy S., ed. 1977. Nonstate Nations in International Politics. New York: Praeger. Berrington, Hugh. 1985. "Centre-Periphery Conflict and British Politics," in Yves Meny and Vincent Wright, eds. Centre-Periphery Relation in Western Europe, pp. 171-206. London: Allen & Unwin. Berry, William D. 1984. Nonrecursive Causal Models. Beverly Hills, calif.: Sage Publcations. Birch, Anthony H. 1978. "Minority Nationalist Movements and Theories of Political Integration." World Politics. Vol. 30, No. 3, pp. 325-43. Blalock, Hubert M., Jr. 1979. Social Statistics, rev. 2nd ed. New York: McGraw-Hill Book Co. 1971. Causal Models in the Social Sciences. Chicago: Aldine-Atherton. Boas, Franz. 1931. "Race." Encyclopedia of the Social Sciences. Vol. 13, pp. 25-34. New York: Macmillan. Boehm, Max Hildebert. 1931. "Minorities, National." Encyclopedia of the Social Sciences. New York: Macmillian. Bolukasi, Suha. 1988. Turkish-American Relations and Cyprus. Lanham, Md.: University Press of America. Bonacich, Edna. 1980. "Class Approaches to Ethnicity and Race." Insurgent Sociologist, Vol. 10, No. 2, pp. 9- 23. Brass, Paul R. 1976. "Ethnicity and Nationality Formation." Ethnicity. Vol. 3, No. 3, pp. 25-41. Bunge, Frederica M. 1983. Indian Ocean: Five Island Countries. Washington, D.C.: Amerincan Univerity. Claude, Inis L., Jr. 1955. National Minorities: An International Problem. Cambridge, Mass.: Harvard University Press. 174 Coleman, James S. 1960. "The Politics of Sub-Saharan Africa," in Galriel A. Almond, and James S. Coleman, eds. The Politics of the Developing Areas, pp. 247- 368. Princeton, N.J.: Princeton University Press. Coleman, James S., and Carl G. Rosberg, Jr. 1964. "Conclusions," in James S. Coleman, and Carl G. Rosberg, Jr., eds. Political Parties and National Integration in Tropical Africa, pp. 655-91. Berkeley, Calif.: University of California Press. Connor, Walker. 1984. "Econ- or Ethno-Nationalism?" Ethnic and Racial Studies. Vol. 7, No. 3, pp. 342-59. 1983. "Nationalism: Competitors and Allies." Canadian Review of Studies in Nationalism. Vol. 10, No. 2, pp. 277-82. 1979. "An Overview of the Ethnic Composition and Problems of Non-Arab Asia," in Tai S. Kang, ed. Nationalism and the Crises of Ethnic Minorities in Asia, pp. 11-27. Westport, Conn.: Greenwood Press. 1977. "Ethnonationalism in the First World: The Present in Historical Perspective." in Milton J. Esman, ed. Ethnic Conflict in the Western World, pp. 19-45. Ithaca: Cornell University Press. 1976. "The Political Significance of Ethnonationlism within Western Europe," in Abdul Said and Luiz R. Simmons, eds. Ethnicity in An International Context. pp. 110-33. New Brunswick, N.J.: Transaction Books. 1973. "The Politics of Ethno Nationalism." Journal of International Affairs, Vol. 27, No. 1, pp. 1-21. 1972. "Nation-Building or Nation-Destroying?" World Politics. Vol. 24, No, 3, pp. 319-55. 1970. "Ethnic Nationalism as a Political Force." World Politics. Vol. 133, No. 2, pp. 91-97. Copper, Mark N. 1974a. "Plural Societies and Conflict: Theoretical Considerations and Cross-National Evidence." International Journal of Group Tensions, Vol. 4, No. 4, pp. 408-30 1974b. "Racialism and Pluralism as Dimensions of Nations: A Further Investigation." Race, Vol. 15, No. 3, pp. 370-81. 175 Coulter, Philip B. 1989. Measuring Ineguality: A Methodological Handbook. Boulder, Colo.: Westview. Covell, Maureen. 1981. "Ethnic Conflict and Elite Bargaining: The Case of Belgium." Western European Politics, Vol. 4, No. 3, pp. 197-218. Coverdale, John F. 1977. "Regional Challenges to the Nation-State: The case of Spain." in Arthur Cyr, ed. The States in Europe, pp. 43-56. Chicago: Chicago Council on Foreign Relations. Da Silva, Milton. 1977. "The Basques as a Non-State Nation," in Judy S. Bertelsen, ed. Nonstate Nations in International Politics, pp. 98-130. New York: Praeger. 1975. "Modernization and Elite Conflict: The Case of the Basques." Comparative Politics. Vol. 7, No. 2, pp. 227-51. Dahl, Robert A. 1971. Polyarchy: Participation and Opposition. New Haven, Conn.: Yale University Press. Davies, James C. 1969. "Political Stability and Instability." Journal of Conflict Resolution, Vol. 13, No. 1, pp. 1-18. 1962. "Toward a Theory of Revolution." American Sociological Review. Vol. 27, No. 1, pp. 5-19. Dec.T}-, Gyula, comp. 1988. Statistical Report on the Languages of the World as of 1985.. Pt. 3. Bloomington, I no:.: Euroligua. Demographic Yearbook. 1956, 1963, 1964, 1971, 1973, 1979, 1983, 1988. New York: United Nations. Denktash, R. R. 1982. The Cyprus Triangle. London: K. Rustem & Bro. Deutsch, Karl W. 1971. "Social Mobilization and Political Development," in Jason L. Finkle, and Richard W. Gable, eds. Political Development and Social Change, pp. 384- 405. New York: John Wiley & Son. 1966. Nationalism and Social Communication: An Inguiry into the Foundations of Nationality. 2nd ed. Cambridge, Mass.: M.I.T. Press. Doob, Leonard W. 1964. Patriotism and Nationalism: Their Psychological Foundations. New Haven, Conn.: Yale University Press. 176 Douglas, J. Neville H. 1983. "Political Integration and Division in Plural Societies — Problems of Recognition, Measurement ans Salience," in Nurit Klit, and Stanely Waterman, eds. Pluralism and Political Geography, pp. 47-68. London: Croom Helm. Duchacek, Ivo D. 1977. "Antagonistic Cooperation: Territorial And Ethnic Communties." Publius, Vol. 7, No. 4, pp. 3-29. Duvall, Raymond,and Mary Welfling. 1973. "Social Mobilization, Political Institutionalization,and Conflcit in Black Africa: A Simple Dynamic Model." Journal of Conflict Resolution, Vol. 17, No. 4, pp. 673-702. Eichenberg, Richard C, Brigitta Widmaier, and Ulrich Widmaier. 1984. "Projecting Domestic Conflict Using Cross-Section Data: A Project Report," in J. David Singer, and Richard J. Stool, eds. Quantitative Indicators in World Politics: Timely Assurance and Early Warning, pp. 11-33. New York: Praeger. Enloe, Cynthis H. 1986. Ethnic Conflict and Political Development. Lanham, Md.: University Press of American. Esman, Milton J. 1982. "The Politics of Official Bilingualism in Canada." Political Science Quarterly. Vol. 97, No. 2, pp. 233-53. 1976. "Communal Conflict in Souteast Asia," in Nathan Glazer, and Daniel P. Moynihan, eds. Ethnicity: Theory and Experience, pp. 391-419. Cambridge, Mass.: Harvard University Press. 1973. "The Management of Communal Conflict." Public Policy. Vol. 21, No. 1, pp. 49-79. Esman, Milton J., and Itamar Rabinovich, eds. 1988. Ethnicity, Pluralism, and the State in the Middle East. Itahca, N.Y.: Cornell University Press. Europa World Year Book. 1989. Rochester, N.Y.: Europa Publications. Feierabend, Ivo K, and Rosalind L. Feierabend. 1972. "Systemic Conditions of Political Aggression: An Application of Frustration-Agrression Theory," in Ivo K. Feierabend, Rosalind L. Feieeeerabend, and Ted Robert Gurr, eds. Anger. Violence, and Politics: Theories and Research, pp. 136-83. Englewood Cliffs, 177 N.J.: Prentice-Hall. 1966. "Aggressive Behaviors within Polities, 1948-62: A Cross-National Study." Journal of Conflict Resolution. Vol. 10, No. 3, pp. 249-71. Feierabend, Ivo K, Rosalind L. Feierabend., and Rose Kelly, no date. Data Bank of Minority Group Conflict, 1955- 1965. Ann Arbor, Mich.: ICPSR. ICPSR Study No. 5209. Fenwick, Rudy. 1981. "Social Change and Ethnic Nationalism: An Historical Analysis of the Separatist Movement in Quebec." Comparative Studies in Society and History, Vol. 23, No. 2, pp. 196-216. Fernandez, J. W. 1969. "Contemportary African Religion: Confluents of Inquiry," in G. M. Carter, and A. parden, eds. Expanding Horizons in African Studies, pp. 27-47. Evanston, 111.: Northwestern University Press. Fishman, Joshua A. 1966. "Some Constrasts between Linguistically Homogeneous and Lingusitically Heterogeneous Polities," in Stanley Lieberson, ed. Explanations in Sociolingusitcs. pp. 18-30. The Hague: Mouton. Fishman, Joshua A., and Frank R. Solano. 1990. "Cross- Polity Perspective on the Importance of Linguistic Heterogeneity as a 'Contrbuting Factor' in Civil Strife." Canadian Review of Studies in Nationalism, Vol. 27, No. 1-2, pp. 131-46. 1989. "Cross-Polity Linguistic Homogeniety/Heterogeneity and Per-Capita Gross National Product: An Empirical Study ." Language Problems and Language Planning, Vol. 13, No. 2, pp. 103-18. Foltz, William J. 1974. "Ethnicity, States, and Conflict," in Wendell Bell, and Walter A. Freeman, eds. Ethnicity and Nation-building, pp. 103-16. Beverly Hills, Calif.: Sage Piblications. Gastil, Raymond D. 1979. Freedom in the World: Political Rights and Civil Rights. New Yrok: Freedom House. 1978. "The Comparative Survey of Freedom — VIII." Freedom at Issue, No. 44, pp. 2-19. 1977. "The Comparative Survey of Freedom — VII." Freedom at Issue, No. 39, pp. 5-17. 1976. "The Comparative Survey of Freedom — VI." 178 Freedom at Issue, No. 34, pp. 11-20. 1975. "The Comparative Survey of Freedom V." Freedom at Issue, No. 29, pp. 3-9. Geertz, Clifford , ed. 1963. Old Societies and New States: The Quest for Modernity in Asia and Afica. New York: Free Press. Gillespie, John V. 1971. "An Introduction to Macro-Corss- National Research." in John V. Gillespie, and Betty Nosvold, eds. Macro-Quantitative Analysis, pp. 13-27. Beverly Hills, Calif.: Sage Publications. Glaser, Kurt. 1983. "Political Rights and Realities in Multi-Ethnic Countries." Korea and World Affairs. Vol. 7, No. 3, pp. 416-44. Greenberg, Joseph H. 1956. "The Measurement of Linguistic Diversity." Language, Vol. 32, No. 1, pp. 109-15. Grove, D. John. 1991. "Education and the Ethnic Division of Labor in Reform-Minded Societies." Comparative Political Studies, Vol. 21, No. 1, pp. 56-75. 1979. "Ethnic Socioeconomic Redistribution: A Crosss-Cultural Study." Comparative Politics. Vol. 12, No. 1, pp. 87-98. 1977. "A Cross-National Explanation of Cross- Cutting and Reinforcing Cultural Cleavages." International Jorunal of Comparative Sociology. Vol. 18, No. 3-4, pp. 217-27. 1974. "Different Political and Economic Patterns of Ethnic and Race Relations: A Cross-National Analysis." Race. Vol. 15, No. 3, pp. 303-28. 1973-74. "The Race vs. Ethnic Debate: A Cross- National Analysis of Two Theoretical Approaches." Studies in Race and Nations. Vol. 5, No. 4, pp. 1-44. Gurr, Ted Robert. 1968. "A Casual Model of Civil Strife: A Comparative Analysis Using New Indices." American Political Science Review, Vol. 62, No. 4, pp. 1104-24. 1966. New Error-Compensated Measures for Comparing Nations: Some Correlates of Civil Violence. Princeton, N.J.: Center of International Studies, Princeton University. Research Monography No. 25. Gurr, Ted Robert, and Raymond Duval. 1973. "Civil Conflict 179 in the 1970's: A Reciprocal Theoretical System with ' Parameter Estimates." Comaprative Political Studies. Vol. 6, No. 2, pp. 135-69. Gurr, Ted Robert, and Erika Gurr. 1983. "Indicators of Group Discrimination and Potential Separatism," in Charles Lewis Taylor, and David A. Jodice, World Handbook of Political and Social Indicators. 3rd ed. Vol. 1, pp. 50-57. New Haven, Conn.: Yale University Press. Guseman, Patricia K., Rudolf J. Freund, and Albert Schaffer. 1976. "The Measurement of Diversity: Ethnic and Socioeconomic Mixing in Residential Areas." Social Science Research, Vol. 5, No. 1, pp. 21-34. Halley, Laurence. 1985. Ancient Affections: Ethnic Groups and Foreign Policy. New York: Praeger. Hanushek, Eric A., and John E. Jackson. 1977. Statistical Methods for Social Scientists. Orlando, Fla.: Academic Press. Haug, Marie R. 1967. "Social and Cultural Pluralism as a Concept in Social Systems Analysis." American Journal of Sociology. Vol. 73, pp. 294-304. Hechter, Michael. 1986. "Theories of Ethnic Relations." in John F. Stacks, Jr., ed. The Primodial Challenge, pp. 13-24. New York: Greenwood Press. . 1975. Internal Colonialism: The Celtic Fringe in British National Development, 1536-1966. Berkeley, Calif.: University of California Press. Hibbs, Douglas A. 1973. Mass Political Violence: A Cross- National Causal Analysis. New York: John Wiley & Sons. Horowitz, Donald L. 1985. Ethnic Groups in Conflict. Berkeley, Calif.: University of California Press. Hudson, Michael C. 1977. Arab Politics: The Search for Legitimacy. New Haven, Conn.: Yale University Press. Huntington, Samuel P. 1968. Political Order in Changing Society. New Haven, Conn.: Yale University Press. Inglehart, Ronald F., and Margaret Woodward. 1967. "Language Conflicts and Political Community." Comparative Studies in Society and History, Vol. 10, No. 1, pp. 27-45. Jackman, Robert W. 1985. "Cross-National Statistical 180 Research and the Study of Comparative Politics." American Jounral of Political Science, Vol. 29, No. 1, pp. 161-82. 1978. "The Predictability of Coup d'etat: A Model with African Data." American Political Science Review, Vol. 72, No. 4, pp. 1262-75. Jackson, Robert H. 1984. "Ethnicity," in Giovanni Sartori, ed. Social Science Concept: A Systematic Analysis, pp. 205-33. Beverly Hils, Calif.: Sage Publications. Jacob, Philip E., and Henry Teune. 1964. "The Integrative Process: Guidelines for Analysis of the Bases of Political Community," in Philip E. Jacob, and James V. Toscano, eds. The Integration of Political Communities, pp. 1-45. Philadelphia: J. B. Lippincott. Jensen, Loyd. 1982. Explaining Foreign Policy. Englewood Cliffs, N.J.: Prentice-Hall. Johnston, J. 1984. Econometric Methods. 3rd ed. New York: MaGraw-Hill Book Co. Joseph, Joseph S. 1985. Cyprus: Ethnic Conflict and International Concern. New York: Peter Lang. Khalilzad, Zalmay. 1984-85. "The Politics of Ethnicity in Southwest Asia: Political Development or Political Decay?" Political Science Quarterly, Vol. 99, No. 4, pp. 657-79. Koutsoyiannis, A. 1977. Theory of Econometrics. 2nd ed. Totowa, N.J.: Barnes & Noble Books. Kuper, Leo. 1969. "Plural Societies: Perspectives and Problesm," in Leo Puper, and M. G. Smith, eds. Pluralism in Africa, pp. 7-20. Berkeley, Calif.: University of California Press. Kuper, Leo, and M. G. Smith, eds. 1969. Pluralism in Africa. Berkeley, Calif.: University of California Press. Kurian, George Thomas. 1984. The New Book of World Rankings. New York: Facts on File. 1979. The Book of World Rankings. New York: Facts on File. Laakso, Markku, and Rein Taagepera. 1979. "'Effective' Number of Parties: A Measure with Application to West 181 Europe." Comparative Political Studies. Vol. 12, No. 1, pp. 3-27 Leng, Lee Yong. 1983. "Ethnic Differences and the State- Minority Relationship in Southeastern Asia." Ethnic and Racial Studies, Vol. 6, No. 2, pp. 213-20. Lerner, Daniel. 1958. The Passing of Traditional Society: Modernizing the Middle East. New York: Free Press. Lieberson, Stanley, and Lynn K. Hansen. 1974. "National Development, Mother Tongue Diversity, and the Comparative Study of Nations." American Sociological Review. Vol. 39, No. 4, pp. 523-41. Lieberson, Stanley, and James F. O'Connor. 1975. "Language Diversity in a Nation and Its Regions," in Jean-Guy Savard, and Richard Vigneault, eds. Multilingual Political Systems: Problems and Solutions, pp. 151-83. Quebec: Les Presses de L'universite Laval. Lijphart, Arend. 1977. "Political Theories and the Explanation of Ethnic Conflict in the Western World: Falsified Predictions and Plausible Postdictions," in Milton J. Esman, ed. Ethnic Conflict in the Western World, pp. 46-64. Ithaca, N.Y.: Cornell University Press. 1971. "Cultural Diversity and Theories of Political Integration." Canadian Journal of Political Science, Vol. 4, No. 1, pp. 1-14. 1968. "Typologies of Democratic Systems." Comparative Political Studies. Vol. 1, No. 1, pp. 3-44. Linz, Juan. 1985. "From Primordialism to nationalism," in Edward A. Tiryakian and Ronald Rogowski, eds. New Nationalism of the Developed West, pp. 204-53. Boston: Allen & Unwin. 1973. "Early State-Building and Late Peripheral Nationalism against the State:The Case of Spain," in S. N. Eisnstadt and Stein Rokkan, eds. Building States and Nations, Vol. 2, pp. 32-116. Beverly Hills, Calif.: Sage. Long, J. Scott. 1983a. Confirmatory Factor Analysis. Beverly Hills, Calif.: Sage Publications. 1983b. Covariance Structural Models. Beverly Hills, Calif.: Sage Publications. 182 McAllister, Ian, and Anthony Mughan. 1984. "Values, Protest and Minority Nationalism in Wales." British Journal of Political Science, Vol. 14, Pt. 2, pp. 230- 43. Mayo, Patricia Elton. 1974. The Roots of Identity: Three National Movements in Contemporary European Politics. London: Allen Lane. Merritt, Richard L. 1970. Systemic Approaches to Comparative Politics. Chicago: Rand McNally & Co. Mill, John Stuart. 1958. Considerations on Representative Government. Indianapolis, Ind.: Bobbs-Merill. Morris, H. S. 1968. "Ethnic Groups." International Encyclopedia of the Social Sciences, Vol. 5, pp. 167- 72. New York: Macmillan Co. & Free Press. Morrison, D. G., R. C. Mitchell, J. N. Poden, and H. M. Stevenson. 1972. Black Africa: A Comparative Handbook. New York: Free Press. Morrison, D. G., and H. M. Stevenson. 1974. "Measuring Social and Political Requirements for System Stability: Empirical Validation of an Index Using Latin American and African Data." Comparative Political Studies, Vol. 7, No. 2, pp. 252-63. 1972. "Integration and Instability: Patterns of African Political Development." American Political Science Review. Vol. 46, No. 3, pp. 902-27. Mughan, Anthony. 1983. "Accommodation or Diffusion in the Management of Linguistic Conflict in Belgium." Political Studies, Vol. 31, No. 3, pp. 434-52. 1979 "Modernization and Ethnic in Belgium." Political Studies. Vol. 27, No. 1, pp. 21-37. Muir, Richard. 1975. Modern Political Geography. London: Macmillan Press. Muller, Siegfries H. 1964. The World's Living Languages. New York: Frederick Ungar. Murdock, George P. 1931. "Ethnic Communties" Encyclopedia of the Social Sciences, Vol. 5, pp. 607-14. New York: Macmillan. Murphy, Alexander B. 1988. "Evolving Regionalism in Linguistically Divided Belgium," in R. J. Johnston, et 183 al. eds. Nationalism, Self-Determination and Political Geography, pp. 135-50. London: Croom Helm. Nairn, T. 1977. The Break-Up of Britian. London: New Left Books. Neter, John, and William Wasserman. 1974. Applied Linear Statistical Models. Homewood, 111.: Richard D. Irwin. Nielsson, Gunnar P. 1987. "On the Concepts of Ethnicity, Nation and State." Paper presented ad the workshop on "The Sociology of Nationalism," SECOND WORLD BASQUE CONFERENCE, Victory, Basque Country, Spain, September, 1987. 1985. "States and 'Nation-Groups': A Global Taxonomy," in Edward A. Tiryakian, and Ronald Rogowski, eds. New Nationalism of the Developed West, pp. 27-56. Boston, Mass.: Allen & Unwin. Nielsson, Gunnar P., and Ralph Jones. 1988. "From Ethnic Categories to Nation: Patterns of Poltical Modernazation." Paper presented at the 29th Annual Convention of the INTERNATIONAL STUDIES ASSOCIATION, St. Louis, Missouri, March 29-April 2, 1988. Olson, Mancur, Jr. 1963. "Rapid Growth as a Destabilizing Force." Journal of Economic History, Vol. 23, pp. 529- 52. Payne, Stanley. 1971. "Catalan and Basque Nationalism" Journal of Contemporary History. Vol. 6, No. 1, pp. 15- 51. Pedhazur, Elazar J. 1983. Multiple Regression in Behavioral Research. 2nd ed. New York: Holt, Rinehart & Winston. Powell, G. Bingham, Jr. 1982. Contemportary Democracies: Participation. Stability, and Violence. Cambridge, Mass.: Harvard University Press. Purcell, H. D. 1968. Cyprus. New York: Frederick A. Praeger. Rae, Douglas, and Michael Taylor. 1970. The Analysis of Poltical Cleavage. New Haven, Conn.: Yale University Press. Rawkins, Philip. 1979. "An Approach to the Political Sociology of the Welsh Nationalist Movement." Political Studies. Vol. 27, No. 3, pp. 440-57. 184 Riggs, Fred W. 1986. "What Is Ethnic" What Is National? • Let's Turn the Tables." Canadian Review of Studies in Nationalism, Vol. 13, No, 1, pp. 111-23. Roberts, Janet. 1962. "Sociocultural Change and Communications Problems," in Frank A. Rice, ed. Study of the Role of Second Languages in Asia, Afirca and Latin America, pp. 105-23. Washington, D.C.: Center for Applied Linguistics of Modern Language Association of America. Rose, Arnold M. 1968. "Minorities." International Encyclopedia of the Social Sciences, Vol. 10, pp. 365- 71. New York: Macmillan Co. & Free Press. Rosh, Robert M. 1987. "Ethnic Cleavages as a Composition of Global Military Expenditures." Journal of Peace Research. Vol. 24, No. 1, pp. 21-30. Rothchild, Joseph. 1981. Ethnopolitics: A Conceptual Framework. New York: Columbia University Press. Rudolph, Joseph R., Jr. 1977. "Ethnic Sub-States and the Emergent Politics of Tri-Level Interactions in Western Europe." Western Political Quarterly. Vol. 30, No. 4, pp. 537-57. , and Robert J. Thompson. 1985. "Ethnoterritorial Movements and the Policy Process: Accommodating Nationalist Demands in the Developed World." Comparative Politics. Vol. 17, No. 3, pp. 291-312. Rummel, R. J. 1966. "Dimensions of Conflict Behavior within Nations, 1946-59." Journal of Conflict Resolution. Vol. 10, No. 1, pp. 65-73. Rustow, Dankwart. 1967. A World of Nations. Washington, D.C.: The Brookings Institute. Ryan, Stephen. 1990. Ethnic Conflict and International Relations. Aldershot, England: Dartmouth. 1988. "Emerging Ethnic Conflict: The Neglected . International Dimension." Review of International Studies. Vol. 14, pp. 161-77. Said, Abdul, and Luiz R. Simmons. 1976. Ethnicity in an International Context. New Brunswick, N.J.: Transaction Books. Salih, Halil Ibrahim. 1978. Cyprus: The Impacts of Diverse Nationalism on a State. University, Ala.: University 185 of Alabama Press. Sanders, David. 1981. Patterns of Political Instability. New York: St. Martin's Press. Sani, Giacomo, and Giovanni Sartori. 1983. "Polarization, Fragmentation and Competition in Western Democracies," in Hans Daalder, and Peter Mair, eds. European Party Systems. pp. 307-40. London: Sage. Sartori, Giovanni. 1976. Parties and Party System: A Framework for Analysis. Cambridge: Cambridge University Press. Schermerhorn, R. A. 1970. Comparative Ethnic Relations: A Framework for THeorv and Research. Nev; York: Random House. See, Katherine O'Sullivan. 1986. First World Nationalisms: Class and Ethnic Politics in Northern Ireland and Quebec. Chicago: University of Chicago Press. Shabad, Goldie. 1986. "After Autonomy: The Dynamics of Regionalism in Spain," in Stanley G. Payne, ed. The Politics of Democratic Spain, pp. 111-80. Chicago: Chicago Council on Foreign Relations. Shabad, Goldie, and Richard Gunther. 1982. "Language, Nationalism, and Political Conflict in Spain." Comparative Politics. Vol. 14, No. 4, pp. 443-77. Shields, Frederick L., eds. 1984. Ethnic Separation and World Politics. Lanham, Md.: University Press of America. Singh, Baldave. 1981. "Race, Ethnicity and Class: Clarifying Relationships or Continuous Muddling Through." Journal of Ethnic Studies. Vol. 9, No. 2, pp. 1-19. Sinnott, Richard, and E. E. Davis. "Political Mibilization, Political Institutionalization and the Maintenance of Ethnic Conflict." Ethnic and Racial Studies, Vol. 4, No. 4, pp. 398-414. Smith, Anthony D. 1986. "History and Liberty: Dilemmas of Loyalty in Western Democracies." Ethnic and Racial Studies, Vol. 9, No. 1, pp. 43-65. 1983. State and Nation in the Third World. Bringhton: Wheatsheaf Books. 186 1982. "Nationalism, Ethnic Separatism and the Intelligentsia," in Colin H. Williams, ed. National Separatism, pp. 17-42. Cardiff: University of Wales Press. 1981. The Ethnic Revival. Cambridge: Cambridge University Press. Smooha, Sammy. 1975. "Pluralism and Conflict: A Theoretical Explanation." Plural Societies, Vol. 6, No. 3, pp. 69-89. Snyder, Louis L. 1983. "Nationalism and the Flawed Concept of Ethnicity." Canadin Review of Studies in Natinalism, Vol, 10, No. 2, pp. 253-65. Stack, John F., Jr. 1981. "Ethnic Groups as Emerging Transnational Actors," in John F. Stack, Jr., ed. Ethnic Identities in a Transnational World, pp. 18-45. Westport, Conn.: Greenwood Press. Stavrinides, Zenon. 1976. The Cyprus Conflict: National Identity and Statehood, no publisher. Stein, Carol A. 1976. A Simple Useer's Guide to LISREL. Amherst, Mass.: Department of Communications Studies, University of Massachusetts. Stephens, Meic. 1976. Linguistic Minorities in Western Europe. Llandyzul, Wales: Gomer Press. Suhrke, Astri, and Lela Garner Nobel, eds. 1977. Ethnic Conflict in International Relations. New York: Praeger. Tanter, Raymond. 1966. "Dimensions of Conflict Behaviors within and Between Nations, 1958-1960." Journal of Conflict Resolution. Vol. 10, No. 1, pp. 41-64. Taylor, Charles Lewis, and David A Jodice. 1983. World Handbook of Political and Social Indicators. 3rd. ed. Vol. 1. New Haven, Conn.: Yale University Press. no date. World Handbook of Political and Social Indicators III: 1948-1982. Ann Arbor, Mich.: ICPSR. ICPSR Study No. 7761. Taylor, Charles Lewis, and Michael C. Hudson. 1972. World Handbook of Political and Social Indicators. 2nd. ed. New Haven, Conn.: Yale Univ. Press. Thompson, John L. P. 1983. "The Plural Society Approach to 187 Class and Ethnic Political Mobilization." Ethnic and Racial Studies, Vol. 6, No. 2, pp. 127-53. Tilly, Charles. 1978. From Moblization to Revolution. Reading, Mass.: Addison-Wesley. Timms, D. 1970. "Quantatative Techniques in Urban Social Geography," in Richard J. Chorley, and Peter Haggett, eds. Frontiers in Geographical TEaching, pp. 239-65. London: Methuen. Tiryakian, Edward A. 1980. "Quebec, Wales and Scotland: Three Nations in Search of A State." International Journal of Comparative Sociology. Vol. 21, No. 1-2, pp. 1-13. Tu'fte, Edward R. 1974. Data Analysis for Politics and Policy. Englewood Cliffs, N.J.: Prentice-Hall. Urwin, Derek W. 1985. "The Price of A Kingdom: Territory, Identity and the Centre-periphery Dimension in Western Europe," in Yves Meny and Vincent Wright, eds. Centre- Periphery Relation in Western Europe, pp. 151-70. London: Allen & Unwin. 1983. "Harbinger, Fossil or Fleabite: •Regionalism' and the West European Party Mosaic," in Hans Daalder and Peter Mair, eds, Western European Party Systems, pp. 220-56. Beverly Hills, Calif.: Sage. Van den Berghe, Pierre L. 1967. Race and Racism: A Comparative Perspective. New York: John Wiley & Sons. Van den Haag, E. 1973. "Notes on Group and Ethnic Relationships." Plural Societies. Vol. 4, No. 3, pp. 3-18. Van Haegendoren, M. 1982. "Ethnicity, Language and State," Plural Societies. Vol. 13, Nos. 1-4, pp. 47-56. Wai, Dunstan M. 1978. "Sources of Communal Conflicts and Seccessionist Politics in Africa." Ethnic and Racial Studies. Vol. 1, No. 1, pp. 286-305. Wallerstein, Immanuel. 1971. "Ethnicity and National Integration in West Africa," in Jason L. Finkle, and Richard W. Gable, eds. Political Development and Social Change, pp. 669-76. New York: John Wiley & Son. Weber, M. 1968. "Ethnic Groups," in G. Roth, and C. 188 Wittich, eds. Economy and Society. Vol. 1, Ch. 5, pp. 385-98. New York: Bedmister Press. Weekes, Richard V. 1984. Muslim Peoples: A World Ethnographic Survey. Westport, Conn.: Greenwood. Weiner, Myron, and Bert F. Hoselitz. 1961. "Economic Development and Political Stability in India." Dissent. Vol. 8, No. 2, pp. 172-84. Wildgen, John K. 1971. "The Measurement of Hyperfractionalization." Comparative Political Studiesf Vol. 4, No. 2, pp. 233-43. Wilkenfeld, Jonathan. 1980. Foreign Policy Behavior: The Inter-State Bahavior Analysis Model. Beverly Hills, Calif.: Sage Publications. Wolf, Ken. 1986. "Ethnic Nationalism: An Anaysis and a Defense." Canadian Review of Studies in Nationalism, Vol. 13, No. 1, pp. 99-109. World Almanac and Book of Facts. 1990. New York: World Almanac. Wyckoff, Theodore. 1980. "Standardized List of National Political Units in the Twentieth Centurt: The Russett- Singer-Small List of 1968 Updated." Internatonal Social Science Journal. Vol. 32, No. 4, pp. 833-46. Yinger, J. Milton. 1983. "Ethnicity and Social Change: The Interaction of Structural, Cultural, and Personality Factors." Ethnic and Racial Studies. Vol. 6, No. 4, pp. 395-409. Young, Crawford. 1976. The Politics of Cultural Pluralism. Madison, Wis.: University of Wisconsin Press. Zolberg, Aristide R. 1977. "Ethnic Regionalisms in Western Europe," in Arthur Cyr, ed. The State in Europe, pp. 23-42. Chicago: Chicago Council on Foreign Relations. APPENDIX A LIST OF COUNTRIES INCLUDED AND EXCLUDED IN THE STUDY 189 190 (1) Countries included1 (N=132): 700 AFGN AFGHANISTAN 339 ALBN ALBANIA 615 ALGR ALGERIA 540 ANGL ANGOLA 160 ARGN ARGENTINA 900 AUSL AUSTRALIA 305 AUST AUSTRIA 053 BRBD BARBADOS 211 BLGM BELGIUM 434 BNIN BENIN 145 BOLV BOLIVIA 571 BTSN BOTSWANA 140 BRZL BRAZIL 355 BLGR BULGARIA 775 BRMA BURMA 516 BRND BURUNDI 471 CMRN CAMEROON 020 CNDA CANADA 482 CAFR CENTRAL AFRICA REPUBLIC 483 CHAD CHAD 155 CHLE CHILE 710 CHNA CHINA 100 CLMB COLOMBIA 484 CNGO CONGO 094 CRCA COSTA RICA 040 CUBA CUBA 352 CYPR CYPRUS 315 CZCH CZECHOSLOVAKIA 390 DNMK DENMARK 042 DMNR DOMINICAN REPUBLIC 130 ECDR ECUADOR 651 EGPT EGYPT 092 ELSL EL SALVADOR 530 ETHP ETHIOPIA 375 FNLD FINLAND 220 FRNC FRANCE 481 GBON GABON 420 GMBA GAMBIA 265 GDR GERMANY, EAST 260 FRG GERMANY, WEST 452 GHNA GHANA 350 GRCE GREECE 090 GTML GUATEMALA 438 GNEA GUINEA 110 GYNA GUYANA 1. See the code book in the ICPSR version of World Handbook of Political and Social Indicators III (ICPSR 7761) and Wyckoff (1980). 191 041 HATI HAITI 091 HNDS HONDURAS 310 HNGR HUNGARY 395 ICLD ICELAND 750 INDA INDIA 850 INDS INDONESIA 630 IRAN IRAN 645 IRAQ IRAQ 205 IRLD IRELAND 666 ISRL ISRAEL 325 ITLY ITALY 437 IVCT IVORY COAST 051 JMCA JAMAICA 740 JPAN JAPAN 663 JRDN JORDAN 811 KMPC KAMPUCHEA 501 KNYA KENYA 731 KORN KOREA, NORTH 732 KORS KOREA, SOUTH 690 KWAT KUWAIT 812 LAOS LAOS 660 LBNN LEBANON 570 LSTO LESOTHO 450 LBRA LIBERIA 620 LBYA LIBYA 212 LXBG LUXEMBOURG 580 MDGS MADAGASCAR 553 MLWI MALAWI 820 MLYS MALAYSIA 781 MLDV MALDIVES 432 MALI MALI 338 MLTA MALTA 435 MRTN MAURITANIA 590 MRTS MAURITIUS 070 MXCO MEXICO 712 MNGL MONGOLIA 600 MRCO MOROCCO 541 MZBQ MOZAMBIQUE 790 NPAL NEPAL 210 NTHL NETHERLANDS 920 NZLD NEW ZEALAND 093 NCRG NICARAGUA 436 NGER NIGER 475 NGRA NIGERIA 385 NRWY NORWAY 770 PKST PAKISTAN 095 PNMA PANAMA 910 PPNG PAPUA NEW GUINEA 150 PRGY PARAGUAY 135 PERU PERU 840 PHLP PHILIPPINES 290 PLND POLAND 235 PRTG PORTUGAL 360 RMNA RUMANIA 517 RWND RWANDA 670 SDAR SAUDI ARABIA 433 SNGL SENEGAL 451 SRLE SIERRA LEONE 830 SNGP SINGAPORE 520 SMLA SOMALIA 560 SAFR SOUTH AFRICA 365 USSR SOVIET UNION 230 SPAN SPAIN 780 SRLK SRI LANKA 625 SDAN SUDAN 380 SWDN SWEDEN 225 SWTZ SWITZERLAND 652 SYRA SYRIA 713 TWAN TAIWAN 510 TNZN TANZANIA 800 TLND THAILAND 461 TOGO TOGO 052 TRNT TRINIDAD AND TOBAGO 616 TNSA TUNISIA 640 TRKY TURKEY 500 UGND UGANDA 200 UK UNITED KINGDOM 002 USA UNITED STATES 439 UPVL UPPER VOLTA 165 URGY URUGUAY 101 VNZL VENEZUELA 678 YMNS YEMAN (SANA) 680 YMNA YEMAN (ADEN) 345 YGSL YUGOSLAVIA 490 ZAIR ZAIRE 551 ZMBA ZAMBIA 552 ZIMB ZIMBABWE (2) Countries excluded: 232 ANDORRA 058 ANTIGUA AND BARBUDA 031 BHMS BAHAMAS 692 BHRN BAHRAIN 771 BNGL BANGLADESH 080 BELIZE 760 BHTN BHUTAN 835 BRUNEI 402 CVRD CAPE VERDE 581 CMRS COMOROS 522 DJIBOUTI 054 DOMINICA 411 EQGN EQUATORIAL GUINEA 950 FIJI FIJI 055 GRND GRENADE 404 GNBS GUINEA-BISSAU 946 KIRIBATI 223 LIECHTENSTEIN 221 MONACO 565 NAMIBIA 970 NAURU 698 OMAN OMAN 694 QTAR QATAR 060 SAINT CHRISTOPHER (ST. KITTS) AND NEVIS 056 SAINT LUCIA 057 SAINT VINCENT AND THE GRENADINES 331 SAN MARINO 403 STPR SAO TOME AND PRINCIPE 591 SYCH SEYCHELLES 115 SRNM SURINAME 572 SWAZ SWAZILAND 955 TONGA 947 TUVALU 696 UAE UNITED ARAB EMIRATES 328 VANTICAN CITY 935 VANUATU 815 VNM VIETNAM 816 VNMN VIETNAM, NORTH 817 VNMS VIETNAM, SOUTH 990 WESTERN SAMOA APPENDIX B SOURCES OF INDICATORS OF ETHNIC DIVERSITY 194 195 Allardt (1979) (n=64 Western European territorial linguistic minorities) patterns of ethnic mobilization Anderson et al. (1967) (n=84) cultural pluralism Banks & Texter (1963) religious homogeneity (n=106) racial homogeneity (n=109) linguistic homogeneity (n=114) religious, racial, and linguistic homogeneity (n=106) Barrows (1976) (n=32) ethnic distribution: linguistic fractionalization (from Taylor & Hud son) ethnic fractionalization (from Morrison et al.) ethnic polarization ethnic pluralism Connor (1979) (n=29 non-Arab Asian states) ethnolinguistic composition Decsy (1988) (n=271 states and state-like units) linguistic composition Grove (1974) (n=113) cultural pluralism: ethr.ic pluralism racial pluralism Gurr (1966) group discrimination (n=119) potential separatism (n=118) Haug (1967) (n=114) cultural pluralism Kurian (1984) ethnic homogeneity (n=135) religious homogeneity: Christians (n=205) Muslims (n=77) Jews (n=85) Kurian (1979) ethnic homogeneity (n=135) religious homogeneity: Christians (n=135) Muslims (n=67) Jews (n=87) 196 Muller (1964) (n=183) linguistic composition Nielson (1985) (n=164) ethnic fragmentation Rosh (1987) (n=40) ethnic cleavage Rustow (1967) (n=131) linguistic unity Singh (1981) (n=17) ethnic/racial composition Taylor and Hudson (1972) ethnic & linguistic fractionalization (n=129) percentage of the Christian community (n=136) percentage of the Islamic community (n=135) Taylor and Jodice (1983) (n=136) ethnic & linguistic fractionalization Weekes (1984) (n=112) percentage of Muslim population Weigert et al. (1957) (n=83) religious composition APPENDIX C COMPONENTS OF THE VARIABLES 197 198 A. Components of the Independent and Intervening Variables (named "INDPT.REVALL" in our file) Variable Name Year \Aid e Source VI Country Code ia mm l 3 VI V2 Country Name 4 V2 V3 Political Right 1973 1 V40 V4 Political Right 1974 1 V41 V5 Political Right 1975 1 V42 V6 Political Right 1976 1 V43 V7 Political Right 1977 1 V44 V8 Political Right 1978 1 V45 V9 Political Right 1979 1 V46 VI0 Civil Right 1973 1 V47 VII Civil Right 1974 1 V48 V12 Civil Right 1975 1 V49 VI3 Civil Right 1976 1 V50 V14 Civil Right 1977 1 V51 V15 Civil Right 1978 1 V52 VI6 Civil Right 1979 1 V53 V17 Political Discrimination (%) 1975 3 V56 V18 Political Discrimination (intensity) 1975 1 V57 VI9 Economic Discrimination (%) 1975 3 V60 V20 Economic Discrimination (intensity) 1975 1 V61 V21 Separatism (%) 1975 3 V64 V22 Separatism (intensity) 1975 1 V65 V23 Ethno-linguistic Fractionalization 1960 4 V109 V24 GNP per capita 1978 5 Vlll V25 Muslim (%) 3 V26 Other Religions (%) 3 V27 Christian (%) 3 V28 External Links 1 V29 Geographical Concentration 1 The data for the above components of the independent and intervening variables1 used in this study, except variables 26 through 30, are derived from the ICPSR version of World Handbook of Political and Social Indicators III (Taylor and Jodice: ICPSR 7761), Part 1: Aggregated Data (National Characteristics), which may be obtained from the Polimetric Lab of the Department of Political Science, the Ohio State University. The following information is needed to retrieve the data from the tape: tape #: PL0087 slot #: A167 dsn & file #: ODICT.S7761AG #60 1. There is one exception, the measure of "Imposition of Political Sanction, which is listed in the next section. 199 ODATA.S7761AG #61 B. Components of the Dependent Variables (named "ZVENTT1.REV" for data in the 1963-1972 period, "EVENTT2.REV" for data in the 1973-1982 period, and "EVENTT3.REV" for data in the 1963-1982 period) Variable Name Year Wide Source V30 Country Code 3 VI V31 Country Name 4 V2 V32 Riot 4 V7 V33 Armed Attack (successful) 5 V8 V34 Assassination (successful) 4 V9 V35 Political Strike 4 V10 V36 Imposition of Political Sanction 4 VI8 V37 Political Execution 6 V19 V38 Death from Domestic Group Violence 6 V21 The data for the above components of the dependent variables2 in this study are derived from the ICPRS version of World Handbook of Political and Social Indicators III (Taylor and Jodice; ICPSR 7761), Part 3: Annual Political Events, which may be obtained from the Polimetric Lab of the Department of Political Science, the Ohio State University. While the book contains data for the 1948-1977 period, the ICPSR version has updated the data set and now encompasses the 1948-1982 period. We have extracted the data for three separate time periods: 1963-1972, 1973-1982, and 1963-1982. The following information is needed to retrieve the data from the tape: tape #: PL0193 slot #: L068 dsn & file #: ODICT.S7761AE #30 ODATA.S7761AE #31 C. The Measures of Population (named "POPALL" in our file) Variable Name Year Wide Source V39 Country Code 3 VI V40 Country Name 4 V2 V41 Population (for 1963-1972) 1965 6 V88 V42 Population (for 1963-1982) 1970 6 V89 2. As we have mentioned earlier, the measure of "Imposition of Political Sanction" is obtained in this source. It is further noted that the measure of "Political Strike" is not used in this study. 200 V43 Population (for 1973-1982) 1975 6 V90 These three measure of population are used to standardize the dependent variables, if applicable. The data, as those described in section A, can also be obtained from the ICPRS version of World Handbook of Political and Social Indicators III (Taylor and Jodice: ICPSR 7761), Part 1: Aggregated Data (National Characteristics). Thus, the same set of information is needed to retrieve them. D. Construction of the Measures DIV = [ V23 + ( 1 - V252 - V262 - V272 ) ] / 2 ECO = V24 DIS = [ ( V17 * V18 ) + ( V19 * V20 ) ] / 2 COE = Ln [ ( V36 / V42 ) + 1 ] ELT = V21 * V22 GEO = V29 EXT = V28 GOV = ( V3 + V4 + . . . + V20 ) / 14 RIOT = Ln [ ( V32 / V42 ) + 1 ] ATT = Ln [ ( V33 / V42 ) + 1 ] ASS = Ln [ ( V34 / V42 ) + 1 ] DEATH = V37 + V38 APPENDIX D ORIGINAL DATA BEFORE MANIPULATIONS 201 202 DATA IN THE FILE OF "INDPT.REVALL" 2USA 11111111111111.121.121.000.505 7120 .00 .10 .9011 20CNDA11111111111U1.000.271.282.755 6930 .00 .09 .8131 40CUBA77777767777666.000.000.000.038 800 .00 .15 .8500 41HATI76666776666666.999.999.999.014 190 .00 .21 .7940 42DMNR33444422222322.811.812.000.037 720 .00 .02 .9800 51JMCA11111222222333.771.772.000.046 1110 .00 .49 .5140 52TRNT22222223222222.999.999.999.558 2000 .00 .25 .7530 53BRBD11111111111111.999.999.999.218 1410 .00 .01 .9940 70MXCO54444443333444.000.000.000.305 1050 .00 .13 .8711 90GTML22444433233344.562.563.000.644 570 .00 .02 .9811 S1HNDS76666663333333.999.999.999.162 360 .00 .29 .7100 92ELSL22223343333334.102.102.000.166 460 .00 .001 .0000 93NCRG45555553444555.142.143.000.180 700 .00 .12 .8800 94CRCA11111111111111.052.053.000.072 960 .00 .29 .7100 95PNMA77777656666655.142.142.000.285 1290 .00 .05 .9510 100CLMB22222222223333.202.203.000.060 580 .00 .13 .8700 101VNZL22221112222222.152.152.000.107 2280 .00 .10 .9000 110GYNA24443342223333.999.999.999.584 510 .09 .34 .5730 130ECDR77776653444443.613.613.000.534 590 .00 .11 .8921 135PERU77666655564444.462.462.000.590 760 .00 .06 .9421 140BRZL55444445545554.142.142.000.071 1030 .00 .001 .0000 145BOLV55666654455443.601.602.000.678 360 .00 .01 .9921 150PRGY45555556555665.502.502.000.145 580 .00 .05 .9500 155CHLE17777762555555.052.053.000.140 990 .00 .10 .9000 160ARGN62226663244565.000.000.000.307 1550 .00 .05 .9510 165URGY35556664555666.000.000.000.198 1300 .00 .16 .8400 200UK 11111111111111.042.042.173.325 3780 .00 .29 .7111 205IRLD11111112222111.000.000.000.045 2390 .00 .001 .0010 210NTHL11111111111111.000.411.000.102 5750 .00 .22 .7801 211BLGM11111111111111.000.000.371.551 6270 .00 .05 .9511 212LXBG22222111111111.999.999.999.155 6020 .00 .04 .9600 220FRNC11111112222112.074.073.061.261 5950 .01 .15 .8441 225SWTZ11111111111111.044.043.000.504 8410 .00 .13 .8731 230SPAN55555226655323.000.000.264.436 2750 .00 .001 .0031 235PRTG55552226633222.000.000.071.006 1570 .00 .05 .9501 260FRG 11111111111112.044.043.000.026 6670 .00 .03 .9741 265GDR 77777777777776.000.000.000.017 3910 .00 .07 .9341 290PLND66666666666655.000.000.000.028 2600 .00 .07 .9300 305AUST11111111111111.034.033.000.126 4870 .00 .03 .9701 310HNGR66666666666655.000.000.000.098 2150 .00 .14 .8600 315CZCH77777777776666.351.061.321.490 3610 .00 .34 .6611 325ITLY11112222222112.000.000.000.038 2810 .00 .09 .9141 338MLTA11111222111222.999.999.999.083 1390 .00 .01 .9900 339ALBN77777777777777.999.999.999.093 510 .52 .15 .3310 203 345YGSL66666666666655.521.000.742.754 1550 .10 .00 .9021 350GRCE67222226522222.000.000.000.099 2340 .03 .03 .9440 352CYPR22443333344444.999.999.999.349 1240 .18 .06 .7631 355BLGR77777777777777.081.000.042.220 2110 .11 .03 .8611 360RMNA77777776666666.121.000.093.252 1240 .00 .001.0011 365USSR66677776666666.491.000.472.666 2550 .11 .65 .2411 375FNLD22222222222222.000.000.000.159 5420 .00 .02 .9801 380SWDN11121111111111.000.000.000.083 8150 .00 .001.0000 385NRWY11111111111111.000.000.000.039 6760 .00 .001.0001 390DNMK11111111111111.000.000.000.049 6810 .00 .02 .9801 395ICLD11111111111111.999.999.999.054 5930 .00 .05 .9500 420GMBA22222222222222.999.999.999.728 180 .46 .50 .0420 432MALI77777776667777.999.999.999.778- 90 .62 .37 .0120 433SNGL66666546654433.999.999.999.723 360 .75 .20 .0520 434BNIN77777775567777.999.999.999.618 130 .17 .68 .1530 435MRTN66566666666666.999.999.999.335 320 .87 .12 .0130 436NGER66777776666666.999.999.999.733 130 .75 .24 .0120 437IVCT66666666665555.999.999.999.859 540 .23 .60 .1720 438GNEA77777777777777.999.999.999.750 130 .65 .34 .0120 439UPVL33665524444543.999.999.999.678 110 .25 .70 .0540 450LBRA66666666534444.999.999.999.830 410 .25 .59 .1620 451SRLE46666565555555.999.999.999.769 200 .23 .72 .0521 452GHNA67777666655554.000.000.393.706 590 .10 .68 .2231 461TOG077777775566666.999.999.999.711 250 .05 .70 .2530 471CMRN66667664444555.151.000.201.892 280 .18 .48 .3421 475NGRA66666554445443.000.000.944.869 340 .45 .45 .1021 481GBON66666666666666.999.999.999.688 2540 .00 .30 .7020 482CAFR77777777777777.999.999.999.686 220 .04 .71 .2520 483CHAD66677767776666.999.999.999.826 120 .58 .36 .0631 484CNG075555777666666.999.999.999.657 510 .00 .38 .6231 490ZAIR77777776667666.000.000.564.901 140 .01 .40 .5931 500UGND77777777777777.762.000.504.899 230 .05 .41 .5431 501KNYA55555554445555.000.431.432.833 220 .10 .63 .2721 510TNZN66666666666666.000.000.031.926 170 .25 .48 .2721 516BRND77777777776666.999.999.999.036 110 .01 .41 .5810 517RWND77777766655555.999.999.999.137 100 .01 .64 .3540 520SMLA77777776666777.999.999.999.077 110 .96 .03 .0110 530ETHP55677776656677.999.999.999.694 100 .40 .25 .3521 540ANGL66666776666677.999.999.999.783 370 .00 .57 ,4331 541MZBQ66667776666777.999.999.999.655 180 .12 .73 .1530 551ZMBA55555555545555.000.000.102.818 420 .01 .75 .2431 552ZIMB66666655555555.952.953.173.544 550 .01 .64 .3511 553MLWI77777766666666.999.999.999.620 130 .08 .58 .3430 560SAFR44444555555566.833.824.333.877 1270 .01 .37 .6221 204 570LSTO75555554344444 .999.999.999.222 160 .00 .25 .7510 571BTSN322222243333331.999.999.999.50 6 350 .00 .78 .2210 580MDGS55556553445555 .000.000.154.062 200 .05 .54 .4101 590MRTS33333222222224 .999.999.999.580 610 .02 .61 .3730 600MRCO55555434555534 .000.000.262.534 470 .81 .16 .0331 615ALGR66676666666666 .000.071.154.435 870 .85 .11 .0441 616TNSA66666665555555 .000.000.000.158 730 .78 .21 .0140 620LBYA77777766776666 .999.999.999.228 5530 .92 .06 .0240 625SDAN66666656666655 .000.341.264.735 270 .58 .38 .0431 630IRAN55566666666655 .000.000.274.756 1660 .84 .15 .0121 640TRKY32222224433333 .091.000.092.255 900 .90 .09 .0131 645IRAQ77777777777776 .042.000.184.362 1250 .95 .01 .0431 651EGPT66665556644445 .071.071.000.044 260 .85 .00 .1540 652SYRA77666557777666 .101.000.062.223 720 .68 .22 .1041 660LBNN22244442224444 .000.471.454.135 1070 .42 .13 .4541 663JRDN66666666666666 .071.000.404.047 460 .79 .16 .0541 666ISRL22222223333332 .431.431.000.199 3790 .07 .89 .0411 670SDAR66666666666666 .999.999.999.059 40101 .00 .00 .0040 678YMNS45566664445555 .999.999.999.037 2001 .00 .00 .0030 680YMNA77777777777777 .999.999.999.015 250 .95 .00 .0540 690KWAT44446664333543 .999.999.999.18515190 .95 .03 .0240 700AFGN47777775666667 .999.999.999.658 150 .92 .08 .0011 710CHNA77777667777766 .999.999.999.118 380 .02 .97 .0141 712MNGL77777777777777 .999.999.999.383 860 .10 .90 .0010 713TWAN66665555555544 .861.000.000.350 930 .00 .94 .0640 731KORN77777777777777 .999.999.999.000 450 .00 .94 .0640 732KORS54555556665655 .000.000.000.000 560 .00 .91 .0940 740JPAN22222221111111 .000.000.000.015 4450 .00 .99 .0100 750INDA22223223335522 .000.000.362.886 140 .10 .87 .0331 770PKST33354665555545 .053.000.053.645 160 .87 .12 .0121 775BRMA77766775556666 .000.000.304.475 110 .03 .93 .0411 780SRLK22222223334323 .094.202.102.467 190 .05 .88 .0711 781MLDV33344452224445 .999.999.999.180 1101 .00 .00 .0000 790NPAL66666665555555 .999.999.999.699 110 .001 .00 .0011 800TLND76526665333654 .082.081.033.664 350 .04 .95 .0111 811KMPC66677775567777 .999.999.999.297 90 .00 .99 .0111 812LAOS55567775556777 .999.999.999.600 90 .00 .99 .0141 820MLYS22333333334443 .462.532.462.716 760 .44 .51 .0531 830SNGP55555555555555 .000.000.000.419 2450 .11 .81 .0830 840PHLP45555556555555 .061.062.044.745 380 .03 .13 .8421 850INDS55555555555555 .032.031.364.764 220 .85 .09 .0621 900AUSL11111111111111 .000.000.000.316 5700 .01 .13 .8610 910PPNG33332222222222 .999.999.999.421 470 .00 .52 .4811 920NZLD11111111111111 .082.082.000.373 4280 .00 .19 .8110 DATA IN THE FILE OF "EVENTT3.REV" (1963-1982) 2USA 780 737 9 1501150 0 392 20CNDA 23 71 1 24 83 0 10 40CUBA 1 78 2 0 100 343 402 41HATI 2 81 1 0 138 289 279 42DMNR 63 215 4 10 131 35 4234 51JMCA 18 30 1 2 19 0 916 52TRNT 2 3 0 1 7 0 2 53BRBD 0 0 0 0 0 0 0 70MXCO 55 114 3 14 52 0 440 90GTML 13 286 35 2 64 30 217 91HNDS 5 34 0 0 25 0 53 92ELSL 7 271 20 3 35 0 122 93NCRG 47 192 5 10 84 14 211 94CRCA 10 18 0 3 15 0 0 95PNMA 54 19 3 3 50 0 56 100CLMB 46 455 5 12 101 0 1062 101VNZL 31 340 6 4 116 1 217 110GYNA 42 151 3 7 73 0 337 130ECDR 33 34 1 11 76 0 87 135PERU 48 134 1 34 102 0 737 140BRZL 42 87 4 17 225 32 88 145BOLV 50 142 5 48 141 4 764 150PRGY 1 9 2 0 23 1 34 155CHLE 76 57 6 37 147 488 719 160ARGN 99 370 67 69 231 0 4586 165URGY 27 34 6 31 68 1 60 200UK 39712405 53 136 928 1 1454 205IRLD 13 102 5 3 114 0 53 210NTHL 9 34 1 2 52 0 16 211BLGM 14 26 0 7 20 0 74 212LXBG 0 0 0 1 0 0 0 220FRNC 93 2184 8 75 296 1 54 225SWTZ 8 19 0 0 38 0 0 230SPAN 274 600 18 157 888 9 163 235PRTG 160 121 1 31 434 0 43 260FRG 67 577 4 4 318 0 55 265GDR 2 4 0 0 233 1 11 290PLND 44 21 0 240 274 1 22 305AUST 1 9 1 0 23 0 7 310HNGR 1 1 0 0 52 0 1 315CZCH 30 29 0 22 276 0 28 325ITLY 155 529 47 80 259 0 161 338MLTA 6 2 0 1 8 0 1 339ALBN 0 1 0 0 9 1 4 345YGSL 11 10 1 3 132 350 29 350GRCE 106 212 5 17 710 0 42 352CYPR 29 595 7 3 64 0 1365 355BLGR 2 2 0 0 18 1 1 360RMNA 0 1 0 6 18 0 0 365USSR 36 31 1 1 712 83 22 375FNLD 0 0 0 0 8 0 0 380SWDN 7 5 1 3 22 0 6 385NRWY 1 3 0 1 17 0 1 390DNMK 1 1 0 3 10 0 0 395ICLD 0 0 0 2 0 0 0 420GMBA 0 2 0 1 2 0 0 432MALI 1 2 0 4 14 0 3 433SNGL 22 1 2 16 35 1 14 434BNIN 14 8 1 6 38 1 45 435MRTN 4 4 0 2 16 0 0 436NGER 0 8 3 1 17 15 23 437IVCT 1 2 0 3 19 0 1 438GNEA 1 1 2 0 27 64 2 439UPVL 1 5 0 8 14 0 0 450LBRA 2 0 0 3 25 35 0 451SRLE 12 3 0 4 43 12 15 452GHNA 21 20 0 22 113 12 53 461TOGO 5 2 2 0 19 0 1 471CMRN 1 3 0 1 4 3 6 475NGRA 152 622 7 27 225 72 1995163 481GBON 9 7 0 0 14 0 39 482CAFR 16 2 0 2 20 9 2 483CHAD 18 70 0 0 22 4 2260 484CNGO 21 27 8 4 46 20 517 490ZAIR 28 443 1 3 156 1109 8869 500UGND 74 72 3 2 105 512 29747 501KNYA 51 86 • 4 5 119 0 363 510TNZN 2 20 4 4 80 9 29 516BRND 4 17 2 0 26 119 81860 517RWND 0 16 0 0 5 14 26319 520SMLA 6 13 3 0 27 18 12 530ETHP 24 244 13 14 57 467 33156 540ANGL 14 255 0 2 33 67 9606 541MZBQ 6 148 0 1 28 24 4607 551ZMBA 40 105 1 13 78 0 1295 552ZIMB 62 317 3 18 297 59 3794 553MLWI 11 32 1 1 28 9 81 560SAFR 169 126 1 36 492 19 776 207 89. 570LSTO 9 12 1 0 15 0 24 90. 571BTSN 0 2 0 0 7 0 1 91. 580MDGS 11 8 1 1 12 0 183 92. 590MRTS 5 9 0 1 8 0 28 93. 600MRCO 20 19 1 17 93 56 49 94. 615ALGR 10 75 3 12 103 11 183 95. 616TNSA 23 4 1 2 39 13 3 96. 620LBYA 10 20 0 3 52 71 28 97. 625SDAN 48 122 0 9 138 127 4284 98. 630IRAN 178 280 33 13 285 2692 265 99. 640TRKY 74 168 17 11 207 10 219 100. 645IRAQ 7 231 6 2 154 559 4672 101. 651EGPT 34 26 3 4 217 16 137 102. 652SYRA 33 86 7 16 132 238 1566 103. 660LBNN 55 2167 26 55 155 0 60721 104. 663JRDN 31 148 0 2 95 15 2110 105. 666ISRL 93 260 1 50 230 0 216 106. 670SDAR 4 12 1 0 23 112 1 107. 678YMNS 7 149 9 0 60 138 7243 108. 680YMNA 51 431 10 15 93 73 1419 109. 690KWAT 1 7 1 3 34 0 1 110. 700AFGN 15 241 4 4 48 3504 45 111. 710CHNA 136 480 0 13 459 1092 2389 112. 712MNGL 1 0 0 0 8 0 0 113. 713TWAN 5 13 0 3 59 5 473 114. 731KORN 0 2 0 0 12 0 13 115. 732KORS 126 55 1 25 324 22 178 116. 740JPAN 57 43 0 12 70 0 32 117. 750INDA 447 1572 31 610 545 0 4280 118. 770PKST 237 542 25 52 434 103 310877 119. 775BRMA 39 171 1 3 87 0 1837 120. 780SRLK 23 32 1 9 95 1 5245 121. 781MLDV 5 1 0 0 2 0 0 122. 790NPAL 23 16 0 6 39 7 56 123. 800TLND 16 188 3 10 91 4 1429 124. 811KMPC 14 1204 1 3 112 691 55884 125. 812LAOS 8 428 7 0 63 1 22456 126. 820MLYS 67 76 2 1 143 0 476 127. 830SNGP 4 5 0 0 50 3 0 128. 840PHLP 44 394 11 2 188 1 4192 129. 850INDS 63 214 0 1 341 16 575922 130. 900AUSL 8 21 1 4 34 0 0 131. 910PPNG 1 9 0 0 4 0 7 132. 920NZLD 2 1 0 0 5 0 0 208 DATA IN THE FILE OF "EVENTTl.REV" (1963-1972) 1. 2 USA 653 425 4 130 770 0 342 2. 20CNDA 21 51 •1 20 66 0 9 3. 40CUBA 0 73 2 0 90 343 400 4. 41RATI 2 72 1 0 118 289 254 5. 42DMNR 59 207 2 9 120 35 4219 6. 51JMCA 2 14 0 0 8 0 9 7. 52TRNT 2 3 0 1 7 0 2 8. 53BRBD 0 0 0 0 0 0 0 9. 70MXCO 43 37 0 9 31 0 192 10. 90GTML 2 134 25 1 47 26 191 11. 91HNDS 4 8 0 0 12 0 4 12. 92ELSL 0 2 0 0 5 0 100 13. 93NCRG 7 16 0 0 13 0 151 14. 94CRCA 2 0 0 0 5 0 0 15. 95PNMA 45 15 3 1 35 0 56 16. 100CLMB 25 162 1 5 50 0 994 17. 101VNZL 27 333 6 4 109 Q 208 18. 110GYNA 41 147 0 2 59 0 337 19. 130ECDR 23 25 0 4 49 0 30 20. 135PERU 10 100 0 8 35 0 585 21. 140BRZL 35 74 4 11 186 10 67 22. 145BOLV 36 109 5 21 91 0 641 23. 150PRGY 1 7 0 0 14 1 5 24. 155CHLE 51 20 3 16 63 1 23 25. 160ARGN 81 182 5 48 .164 0 110 26. 165URGY 26 24 6 23 37 0 46 27. 200UK 199 2028 1 32 373 0 554 28. 205IRLD 6 44 0 2 59 0 2 29. 210NTHL 2 13 0 0 21 0 0 30. 211BLGM 7 11 0 4 13 0 69 31. 212LXBG 0 0 0 0 0 0 0 32. 220FRNC 37 80 0 31 132 1 12 33. 225SWTZ 2 4 0 0 19 0 0 34. 230SPAN 56 64 0 31 258 3 11 35. 235PRTG 10 19 1 6 121 0 4 36. 260FRG 48 55 1 3 132 0 37 37. 265GDR 2 4 0 0 186 1 11 38. 290PLND 15 9 0 14 157 1 20 39. 305AUST 1 2 0 0 13 0 3 40. 310HNGR 1 1 0 0 47 0 1 41. 315CZCH 28 29 0 21 207 0 28 42. 325ITLY 86 117 1 54 95 0 41 43. 338MLTA 5 1 0 1 4 0 0 44. 339ALBN 0 0 0 0 7 0 4 209 45. 345YGSL 7 7 1 3 71 350 29 46. 350GRCE 44 80 0 5 366 0 8 47. 352CYPR 22 262 1 0 29 0 436 48. 355BLGR 2 2 0 0 16 1 1 49. 360RMNA 0 1 0 0 8 0 0 50. 365USSR 14 12 1 0 343 82 14 51. 375FNLD 0 0 0 0 4 0 0 52. 380SWDN 6 3 1 2 11 0 2 53. 385NRWY 0 1 0 1 8 0 0 54. 390DNMK 1 0 0 0 6 0 0 55. 395ICLD 0 0 0 1 0 0 0 56. 420GMBA 0 0 0 1 1 0 0 57. 432MALI 0 2 0 0 6 0 3 58. 433SNGL 15 0 2 15 30 1 14 59. 434BNIN 12 7 1 6 34 0 39 60. 435MRTN 4 0 0 2 8 0 0 61. 436NGER 0 7 3 0 12 8 22 62. 437IVCT 1 1 0 2 17 0 1 63. 438GNEA 1 0 0 0 24 64 1 64. 439UPVL 0 5 0 0 5 0 0 65. 450LBRA 0 0 0 0 11 0 0 66. 451SRLE 8 1 0 0 34 4 8 67. 452GHNA 8 11 0 4 83 4 52 68. 461TOGO 5 2 2 0 18 0 1 69. 471CMRN 0 2 0 0 3 3 6 70. 475NGRA 129 609 6 4 182 35 1995119 71. 481GBON 8 6 0 0 12 0 39 72. 482CAFR 0 0 0 0 8 8 2 73. 483CHAD 17 22 0 0 12 0 2048 74. 484CNGO 21 24 6 4 40 3 478 75. 490ZAIR 25 429 1 2 139 1083 8769 76. 500UGND 63 47 2 1 78 4 5990 77. 501KNYA 46 82 3 3 96 0 317 78. 510TNZN 1 18 4 1 74 9 29 79. 516BRND 3 11 2 0 24 119 81860 80. 517RWND 0 15 0 0 3 14 26307 81. 520SMLA 6 5 3 0 22 1 12 82. 530ETHP 1 36 2 0 16 0 207 83. 540ANGL 0 128 0 0 11 30 867 84. 541MZBQ 0 109 0 0 19 0 3488 85. 551ZMBA 39 98 1 11 58 0 1281 86. 552ZIMB 49 105 0 15 233 42 375 87. 553MLWI 11 32 1 1 24 9 81 88. 560SAFR 35 30 1 4 302 15 150 210 89. 570LSTO 8 4 1 0 12 0 24 90. 571BTSN 0 1 0 0 4 0 0 91. 580MDGS 5 4 0 1 7 0 40 92. 590MRTS 5 9 0 0 6 0 28 93. 600MRCO 19 16 0 12 80 23 48 94. 615ALGR 8 71 2 10 96 11 181 95. 616TNSA 13 1 1 0 23 0 3 96. 620LBYA 6 15 0 3 34 0 18 97. 625SDAN 43 111 0 7 110 11 3669 98. 630IRAN 23 21 3 0 110 48 216 99. 640TRKY 25 43 0 2 87 6 100 100. 645IRAQ 4 191 3 1 132 235 4048 101. 651EGPT 13 1 1 2 130 4 72 102. 652SYRA 28 37 2 15 104 48 1481 103. 660LBNN 38 77 2 34 84 0 225 104. 663JRDN 30 145 0 1 89 14 2102 105. 666ISRL 12 51 0 6 61 0 40 106. 670SDAR 0 7 0 0 14 48 0 107. 678YMNS 7 139 5 0 49 66 7238 108. 680YMNA 51 430 10 15 87 50 1419 109. 690KWAT 0 1 1 2 21 0 1 110. 700AFGN 5 0 0 1 3 0 0 111. 710CHNA 131 474 0 10 419 1026 2382 112. 712MNGL 1 0 0 0 8 0 0 113. 713TWAN 2 10 0 2 32 4 472 114. 731KORN 0 2 0 0 12 0 13 115. 732KORS 68 38 0 12 161 6 168 116. 740JPAN 50 15 0 8 53 0 21 117. 750INDA 303 1480 10 583 322 0 3659 118. 770PKST 171 498 19 35 260 97 309754 119. 775BRMA 37 86 1 0 69 0 827 120. 780SRLK 11 27 0 7 59 1 5098 121. 781MLDV 5 1 0 0 0 0 0 122. 790NPAL 9 2 0 5 23 5 36 123. 800TLND 0 88 0 0 49 0 549 124. 811KMPC 8 449 1 1 62 104 37769 125. 812LAOS 0 373 6 0 29 0 22355 126. 820MLYS 66 52 1 1 122 0 439 127. 830SNGP 4 4 0 0 35 3 0 128. 840PHLP 32 233 8 2 88 0 862 129. 850INDS 57 214 0 1 315 16 575912 130. 900AUSL 8 8 0 2 18 0 0 131. 910PPNG 1 4 0 0 2 0 4 132. 920NZLD 0 1 0 0 1 0 0 211 DATA IN THE FILE OF "EVENTT2.REV" (1973-1982) 1. 2 USA 127 312 5 20 380 0 50 2. 20CNDA 2 20 0 4 17 0 1 3. 40CUBA 1 5 0 0 10 0. 2 4. 41HATI 0 9 0 0 20 0 25 5. 42DMNR 4 8 2 1 11 0 15 6. 51JMCA 16 16 1 2 11 0 907 7. 52TRNT 0 0 0 0 0 0 0 8. 53BRBD 0 0 0 0 0 0 0 9. 70MXCO 12 77 3 5 21 0 248 10. 90GTML 11 152 10 1 17 4 26 11. 91HNDS 1 26 0 0 13 0 49 12. 92ELSL 7 269 20 3 30 0 22 13. 93NCRG 40 176 5 10 71 14 60 14. 94CRCA 8 18 0 3 10 0 0 15. 95PNMA 9 4 0 2 15 0 0 16. 100CLMB 21 293 4 7 51 0 68 17. 101VNZL 4 7 0 0 7 1 9 18. 110GYNA 1 4 3 5 14 0 0 19. 130ECDR 10 9 1 7 27 0 57 20. 135PERU 38 34 1 26 67 0 152 21. 140BRZL 7 13 0 6 39 22 21 22. 145BOLV 14 33 0 27 50 4 123 23. 150PRGY 0 2 2 0 9 0 29 24. 155CHLE 25 37 3 21 84 487 696 25. 160ARGN 18 188 62 21 67 0 4476 26. 165URGY 1 10 0 8 31 1 14 27. 200UK 19810377 52 104 555 1 900 28. 205IRLD 7 58 5 1 55 0 51 29. 210NTHL 7 21 1 2 31 0 16 30. 211BLGM 7 15 0 3 7 0 5 31. 212LXBG 0 0 0 1 0 0 0 32. 220FRNC 56 2104 8 44 164 0 42 33. 225SWTZ 6 15 0 0 19 0 0 34. 230SPAN 218 536 18 126 630 6 152 35. 235PRTG 150 102 0 25 313 0 39 36. 260FRG 19 522 3 1 186 0 18 37. 265GDR 0 0 0 0 47 0 0 38. 290PLND 29 12 0 226 117 0 2 39. 305AUST 0 7 1 0 10 0 4 40. 310HNGR 0 0 0 0 5 0 0 41. 315CZCH 2 0 0 1 69 0 0 42. 325ITLY 69 412 46 26 164 Q 120 43. 338MLTA 1 1 0 0 4 0 1 44. 339ALBN 0 1 0 0 2 1 0 45. 345YGSL 4 3 0 0 61 0 0 46. 350GRCE 62 132 5 12 344 0 34 47. 352CYPR 7 333 6 3 35 0 929 48. 355BLGR 0 0 0 0 2 0 0 49. 360RMNA 0 0 0 6 10 0 0 50. 365USSR 22 19 0 1 369 1 8 51. 375FNLD 0 0 0 0 4 0 0 52. 380SWDN 1 2 0 1 11 0 4 53. 385NRWY 1 2 0 0 9 0 1 54. 390DNMK 0 1 0 3 4 0 0 55. 395ICLD 0 0 0 1 0 0 0 56. 420GMBA 0 2 0 0 1 0 0 .57. 432MALI 1 0 0 4 8 0 0 58. 433SNGL 7 1 0 1 5 0 0 59. 434BNIN 2 1 0 0 4 1 6 60. 435MRTN 0 4 0 0 8 0 0 61. 436NGER 0 1 0 1 5 7 1 62. 437IVCT 0 1 0 1 2 0 0 63. 438GNEA 0 1 2 0 3 0 1 64. 439UPVL 1 0 0 8 9 0 0 65. 450LBRA 2 0 0 3 14 35 0 66. 451SRLE 4 2 0 4 9 8 7 67. 452GHNA 13 9 0 18 30 8 1 68. 461TOGO 0 0 0 0 1 0 0 69. 471CMRN 1 1 0 1 1 0 0 70. 475NGRA 23 13 1 23 43 37 44 71. 481GBON 1 1 0 0 2 0 0 72. 482CAFR 16 2 0 2 12 1 0 73. 483CHAD 1 48 0 0 10 4 212 74. 484CNGO 0 3 2 0 6 17 39 75. 490ZAIR 3 14 0 1 17 26 100 76. 500UGND 11 25 1 1 27 508 23757 77. 501KNYA 5 4 1 2 23 0 46 78. 510TNZN 1 2 0 3 6 0 0 79. 516BRND 1 6 0 0 2 0 0 80. 517RWND 0 1 0 0 2 0 12 81. 520SMLA 0 8 0 0 5 17 0 82. 530ETHP 23 208 11 14 41 467 32949 83. 540ANGL 14 127 0 2 22 37 8739 84. 541MZBQ 6 39 0 1 9 24 1119 85. 551ZMBA 1 7 0 2 20 0 14 86. 552ZIMB 13 212 3 3 64 17 3419 87. 553MLWI 0 0 0 0 4 0 0 88. 560SAFR 134 96 0 32 190 4 626 89. 570LSTO 1 8 0 0 3 0. 0 90. 571BTSN 0 1 0 0 3 0 1 91. 580MDGS 6 4 1 0 5 0 143 92. 590MRTS 0 0 0 1 2 0 0 93. 600MRCO 1 3 1 5 13 33 1 94. 615ALGR 2 4 1 2 7 0 2 95. 616TNSA 10 3 0 2 16 13 0 96. 620LBYA 4 5 0 0 18 71 10 97. 625SDAN 5 11 0 2 28 116 615 98. 630IRAN 155 259 30 13 175 2644 49 99. 640TRKY 49 125 17 9 120 4 119 100. 645IRAQ 3 40 3 1 22 324 624 101. 651EGPT 21 25 2 2 87 12 65 102. 652SYRA 5 49 5 1 28 190 85 103. 660LBNN 17 2090 24 21 71 0 60496 104. 663JRDN 1 3 0 1 6 1 8 105. 666ISRL 81 209 1 44 169 0 176 106. 670SDAR 4 5 1 0 9 64 1 107. 678YMNS 0 10 4 0 11 72 5 108. 680YMNA 0 1 0 0 6 23 0 109. 690KWAT 1 6 0 1 13 0 0 110. 700AFGN 10 241 4 3 45 3504 45 111. 710CHNA 5 6 0 3 40 66 7 112. 712MNGL 0 0 0 0 0 0 0 113. 713TWAN 3 3 0 1 27 1 1 114. 731KORN 0 0 0 0 0 0 0 115. 732KORS 58 17 1 13 163 16 10 116. 740JPAN 7 28 0 4 17 0 11 117. 750INDA 144 92 21 27 223 0 621 118. 770PKST 66 44 6 17 174 6 1123 119. 775BRMA 2 85 0 3 18 0 1010 120. 780SRLK 12 5 1 2 36 0 147 121. 781MLDV 0 0 0 0 2 0 0 122. 790NPAL 14 14 0 1 16 2 20 123. 800TLND 16 100 3 10 42 4 880 124. 811KMPC 6 755 0 2 50 587 18115 125. 812LAOS 8 55 1 0 34 1 101 126. 820MLYS 1 24 1 0 21 0 37 127. 830SNGP 0 1 0 0 15 0 0 128. 840PHLP 12 161 3 0 100 1 3330 129. 850INDS 6 0 0 0 26 0 10 130. 900AUSL 0 13 1 2 16 0 0 131. 910PPNG 0 5 0 0 2 0 3 132. 920NZLD 2 0 0 0 4 0 0 DATA IN THE FILE OF "POPALL 1. 2 USA :L9430 3 204879 213925 2. 20CNDA 19644 21406 22801 3. 40CUBA 7802 8565 9481 4. 41HATI 3950 4235 4552 5. 42DMNR 3703 4343 5118 6. 51JMCA 1760 1882 2029 7. 52TRNT 908 955 1009 8. 53BRBD 235 239 245 9. 70MXCO 42859 50313 59204 10. 90GTML 4583 5298 6129 11. 91HNDS 2209 2553 3037 12. 92ELSL 2954 3516 4108 13. 93NCRG 1701 1970 2318 14. 94CRCA 1495 1737 1994 15. 95PNMA 1261 1458 1678 16. 100CLMB 18691 22075 25890 17. 101VNZL 9105 10559 12213 18. 110GYNA 633 709 791 19. 130ECDR 5095 6031 7090 20. 135PERU 11440 13248 15326 21. 140BRZL 82541 95204 109730 22. 145BOLV 4246 4780 5410 23. 155CHLE 8510 9369 10253 24. 160ARGN 22179 23748 25384 25. 165URGY 2802 2955 3108 26. 200UK 54520 55480 56427 27. 205IRLD 2876 2954 3131 28. 210NTHL 12292 13032 13599 29. 211BLGM 9464 9638 9846 30. 212LXBG 332 339 342 31. 220FRNC 48758 50670 52913 32. 225SWTZ 5857 6267 6535 33. 230SPAN 31913 33779 35433 34. 235PRTG 9235 8628 8762 35. 260FRG 59012 60700 61682 36. 265GDR 17019 17058 17127 37. 290PLND 31496 32473 33841 38. 305AUST 7255 7447 7538 39. 310HNGR 10153 10338 10534 40. 315CZCH 14159 14339 14793 41. 325ITLY 51944 53565 55023 42. 338MLTA 320 326 329 43. 339ALBN 1903 2169 2482 44. 345YGSL 19434 20371 21322 215 45. 350GRCE 8551 8793 8930 46. 352CYPR 594 633 673 47. 355BLGR 8201 8490 8793 48. 360RMNA 19027 20244 21178 49. 365USSR230936 242768 255038 50. 375FNLD 4564 4606 4652 51. 380SWDN 7734 8043 8291 52. 385NRWY 3723 3877 4007 53. 390DNMK 4758 4929 5026 54. 395ICLD 192 204 216 55. 402CVRD 240 268 292 56. 420GMBA 422 463 509 57. 432MALI 4530 5047 5697 58. 433SNGL 3490 3925 4418 59. 434BNIN 2365 2686 3074 60. 435MRTN 1050 1162 1283 61. 436NGER 3524 4016 4600 62. 437IVCT 3835 4310 4885 63. 438GNEA 3510 3921 4416 64. 439UPVL 4858 5384 6032 65. 450LBRA 1376 1523 1708 66. 451SRLE 2367 2644 2983 67. 52GHNA 7740 8628 9873 68. 461TOGO 1697 1960 2248 69. 471CMRN 5356 5836 6433 70. 475NGRA 48779 55073 63049 71. 481GBON 484 500 521 72. 482CAFR 1452 1612 1790 73. 483CHAD 3368 3640 3947 74. 484CNGO 1069 1191 1345 75. 490ZAIR 18835 21638 24450 76. 500UGND 8578 9806 11353 77. 501KNYA 9527 11247 13251 78. 510TNZN 11616 13273 15388 79. 516BRND 3210 3350 3765 80. 517RWND 3198 3679 4233 81. 520SMLA 2500 2789 3170 82. 530ETHP 22078 24855 28134 83. 540ANGL 5101 5670 6394 84. 541MZBQ 7449 8234 9223 85. 551ZMBA 3723 4295 5004 86. 552ZIMB 4393 5308 6272 87. 553MLWI 3932 4360 4909 88. 560SAFR 18337 21500 24663 216 89. 570LSTO 954 1043 1148 90. 571BTSN 554 617 691 91. 580MDGS 6079 6932 8020 92. 590MRTS 761 824 899 93. 600MRCO 13139 15126 17504 94. 615ALGR 11923 14330 16792 95. 616TNSA 4620 5137 5747 96. 620LBYA 1617 1938 2255 97. 625SDAN 13540 15695 18268 98. 630IRAN 24662 28359 32923 99. 640TRKY 31151 35232 39882 100. 645IRAQ 7976 9356 11067 101. 651EGPT 29389 33329 37543 102. 652SYRA 5320 6247 7259 103. 660LBNN 2151 2469 2869 104. 663JRDN 1955 2280 2688 105. 666ISRL 2563 2958 3417 106. 670SDAR 6750 7740 8966 107. 678YMNS 5016 5767 6668 108. 680YMNA 1252 1436 1660 109. 690KWAT 475 760 1085 110. 700AFGN 15097 16978 19280 111. 710CHNA710324 771840 838803 112. 712MNGL 1070 1248 1446 113. 713TWAN 12429 14035 16453 114. 731KORN 12100 13892 15852 115. 732KORS 28088 31365 34663 116. 740JPAN 98881 104331 111120 117. 750INDA482365 543132 613217 118. 770PKST 52415 60449 70560 119. 775BRMA 24754 27748 31240 120. 780SRLK 11164 12514 13986 121. 781MLDV 100 110 120 122. 790NPAL 10100 11232 12572 123. 800TLND 30641 35745 42093 124. 811KMPC 6142 7060 8110 125. 812LAOS 2652 2962 3303 126. 820MLYS 9080 10466 12093 127. 830SNGP 1880 2075 2248 128. 840PHLP 32030 37604 44437 129. 850INDS105070 119467 136044 130. 900AUSL 11387 12552 13809 131. 910PPNG 2148 2413 2716 132. 920NZLD 2628 2820 3031 APPENDIX E DATA AFTER MANIPULATIONS 217 218 OBS DIV ECO DISQ COER COR ELT 1 0.3425 7120 0.120 0.00560 1.00000 0.00 2 0.5454 6930 0.135 0.00387 1.00000 0.56 3 0.1465 800 0.000 0.01161 6.71429 0.00 4 0.1729 190 • 0.03207 6.21429 • 5 0.0381 720 1.215 0.02972 2.78571 0.00 6 0.2729 1110 1.155 0.01005 1.85714 0.00 7 0.4665 2000 • 0.00730 2.07143 • 8 0.1189 1410 • 0.00000 1.00000 • 9 0.2656 1050 0.000 0.00103 3.78571 0.00 10 0.3416 570 1.400 0.01201 3.21429 0.00 11 0.2869 360 • 0.00974 4.57143 • 12 0.0830 460 0.200 0.00991 2.85714 0.00 13 0.1956 700 0.350 0.04176 4.57143 0.00 14 0.2419 960 0.125 0.00860 1.00000 0.00 15 0.1900 1290 0.280 0.03372 6.14286 0.00 16 0.1431 580 0.500 0.00456 2.28571 0.00 17 0.1435 2280 0.300 0.01093 1.78571 0.00 18 0.5677 510 • 0.09800 3.00000 • 19 0.3649 590 1.830 0.01252 5.07143 0.00 20 0.3514 760 0.920 0.00767 5.35714 0.00 21 0.0355 1030 0.280 0.00236 4.50000 0.00 22 0.3489 360 0.900 0.02907 4.85714 0.00 23 0.1200 580 1.000 0.00245 5.14286 0.00 24 0.1600 990 0.125 0.00617 5.28571 0.00 25 0.2010 1550 0.000 0.07527 4.21429 0.00 26 0.2334 1300 0.000 0.00122 5.21429 0.00 27 0.3684 3780 0.080 0.27319 1.00000 0.51 28 0.0225 2390 0.000 0.00871 1.28571 0.00 29 0.2226 5750 0.205 0.00538 1.00000 0.00 30 0.3230 6270 0.000 0.05732 1.00000 0.37 31 0.1159 6020 • 0.00000 1.35714 • 32 0.2664 5950 0.245 0.04615 1.35714 0.06 33 0.3651 8410 0.140 0.00112 1.00000 0.00 34 0.2180 2750 0.000 0.09796 4.21429 1.04 35 0.0505 1570 0.000 0.00712 3.57143 0.07 36 0.0421 6670 0.140 0.01847 1.07143 0.00 37 0.0736 3910 0.000 0.00715 6.92857 0.00 38 0.0791 2600 0.000 0.03613 5.85714 0.00 39 0.0921 4870 0.105 0.00222 1.00000 0.00 40 0.1694 2150 0.000 0.00362 5.85714 0.00 41 0.4694 3610 0.205 0.00514 6.71429 0.32 42 0.1009 2810 0.000 0.58471 1.57143 0.00 43 0.0514 1390 • 0.00368 1.42857 • 44 0.3456 510 • 0.00044 7.00000 • 219 OBS DIV ECO DISQ COER COR ELT 45 0.4670 1550 0.260 0.01490 5.85714 1.48 46 0.1068 2340 0.000 0.75219 3.14286 0.00 47 0.3677 1240 • 0.00751 3.35714 • 48 0.2337 2110 0.040 0.00089 7.00000 0.08 49 0.1260 1240 0.060 0.00007 6.50000 0.27 50 0.5869 2550 0.245 0.14374 6.28571 0.94 51 0.0991 5420 0.000 0.00099 2.00000 0.00 52 0.0415 8150 0.000 0.00566 1.07143 0.00 53 0.0195 6760 0.000 0.003443 1.00000 0.00 54 0.0441 6810 0.000 0.047856 1.00000 0.00 55 0.0745 5930 • 0.000000 1.00000 • 56 0.6324 180 • 0.004310 2.00000 • 57 0.6283 90 • 0.002770 6.78571 • 58 0.5590 360 • 0.008878 5.00000 • 59 0.5521 130 • 0.014048 6.64286 • 60 0.2818 320 • 0.013675 5.92857 • 61 0.5564 130 • 0.004224 6.35714 • 62 0.7086 540 • . 0.004399 5.71429 • 63 0.6059 130 • 0.006862 7.00000 • 64 0.5615 110 • 0.002597 4.14286 • 65 0.6969 410 • 0.016282 5.14286 • 66 0.5976 200 • 0.016132 5.28571 • 67 0.5926 590 0.000 0.013012 5.85714 1.17 68 0.5780 250 • 0.009647 6.35714 • 69 0.7568 280 0.075 0.000685 5.28571 0.20 70 0.7270 340 0.000 0.004077 4.85714 3.76 71 0.5540 2540 • 0.027615 6.00000 • 72 0.5589 220 • 0.012331 7.00000 • 73 0.6782 120 • 0.006026 6.42857 • 74 0.5641 510 • 0.037896 6.00000 • 75 0.6964 140 0.000 0.007184 6.57143 2.24 76 0.7184 230 0.760 0.010651 7.00000 2.00 77 0.6766 220 0.215 0.010525 4.78571 0.86 78 0.7801 170 0.000 0.006009 6.00000 0.03 79 0.2657 110 • 0.007731 6.71429 • 80 0.3024 100 * 0.001358 6.07143 • 81 0.0772 110 • 0.009634 6.71429 • 82 0.6745 100 • 0.002291 6.21429 • 83 0.6366 370 • 0.005803 6.28571 • 84 0.5426 180 • 0.003395 6.42857 • 85 0.5989 420 0.000 0.017998 4.92857 0.20 86 0.5059 550 2.375 0.054444 5.42857 0.51 87 0.5808 130 * 0.006401 6.42857 • 88 0.6778 1270 2.885 0.022626 4.78571 0.99 220 OBS DIV ECO DISQ COER COR ELT 89 0.2985 160 • 0.014279 4.57143 • 90 0.4246 350 • 0.011281 2.64286 • 91 0.2999 200 0.000 0.001730 4.78571 0.60 92 0.5353 610 *• 0.009662 2.50000 • 93 0.4257 470 0.000 0.006130 4.50000 0.52 94 0.3494 870 0.035 0.007162 6.07143 0.60 95 0.2527 730 0.000 0.007563 5.50000 0.00 96 0.1888 5530 • 0.026478 6.57143 • 97 0.6263 270 0.170 0.008754 5.78571 1.04 98 0.5139 1660 0.000 0.010000 5.64286 1.08 99 0.2184 900 0.045 0.005858 2.71429 0.18 100 0.2289 1250 0.040 0.016326 6.92857 0.72 101 0.1495 260 0.070 0.006490 5.14286 0.00 102 0.3511 720 0.050 0.020910 6.28571 0.12 103 0.3696 1070 0.235 0.060887 3,14286 1.80 104 0.1974 460 0.035 0.040822 6.00000 1.60 105 0.2002 3790 0.430 0.074880 2.42857 0.00 106 0.0295 4010 • 0.002967 6.00000 • 107 0.0185 200 • 0.010350 5.00000 • 108 0.0550 250 • 0.062752 7.00000 • 109 0.1406 15190 • 0.043765 4.21429 • 110 0.4026 150 • 0.002823 6.28571 • 111 0.0883 380 • 0.000595 6.71429 • 112 0.2815 860 • 0.006390 7.00000 • 113 0.2314 930 0.430 0.004195 5.14286 0.00 114 0.0564 450 • 0.000863 7.00000 • 115 0.0819 560 0.000 0.010277 5.21429 0.00 116 0.0174 4450 0.000 0.000671 1.50000 0.00 117 0.5591 140 0.000 0.001003 2.71429 0.72 118 0.4368 160 0.075 0.007154 4.57143 0.15 119 0.3038 110 0.000 0.003130 6.14286 1.20 120 0.3426 190 0.380 0.007563 2.50000 0.20 121 0.0900 110 • 0.018019 3.50000 • 122 0.3495 110 • 0.003466 5.50000 • 123 0.3799 350 0.120 0.002543 4.78571 0.09 124 0.1584 90 0 0.015743 6.42857 • 125 0.3099 90 a 0.021046 6.00000 • 126 0.6299 760 0.990 0.013571 3.07143 0.92 127 0.3722 2450 0.000 0.023811 5.00000 0.00 128 0.5108 380 0.090 0.004987 5.00000 0.16 129 0.5149 220 0.045 0.002850 5.00000 1.44 130 0.2797 5700 0.000 0.002705 1.00000 0.00 131 0.4601 470 • 0.001656 2.28571 • 132 0.3404 4280 0.160 0.001771 1.00000 0.00 221 OBS GEO EXT RIOT ATT ASS DEATH 1 1 1 0.00380 0.00359 0.00004 392 2 1 3 0.00107 0.00331 0.00005 10 3 0 0 0.00012 0.00907 0.00023 745 4 0 4 0.00047 0.01895 0.00024 568 5 0 0 0.01440 0.04832 0.00092 4269 6 0 4 0.00952 0.01581 0.00053 916 7 0 3 0.00209 0.00314 0.00000 2 8 0 4 0.00000 0.00000 0.00000 0 9 1 1 0.00109 0.00226 0.00006 440 10 1 1 0.00245 0.05258 0.00658 247 11 0 0 0.00196 0.01323 0.00000 53 12 0 0 0.00199 0.07425 0.00567 122 13 0 0 0.02358 0.09300 0.00253 225 14 0 0 0.00574 0.01031 0.00000 0 15 0 1 0.03637 0.01295 0.00206 56 16 0 0 0.00208 0.02040 0.00023 1062 17 0 0 0.00293 0.03169 0.00057 218 18 0 3 0.05755 0.19308 0.00422 337 19 1 2 0.00546 0.00562 0.00017 87 20 1 2 0.00362 0.01006 0.00008 737 21 0 0 0.00044 0.00091 0.00004 120 22 1 2 0.01041 0.02927 0.00105 768 23 0 0 0.00011 0.00096 0.00021 35 24 0 0 0.00320 0.00240 0.00025 1207 25 0 1 0.03295 0.11797 0.02242 4586 26 0 0 0.00049 0.00061 0.00011 61 27 1 1 0.12610 1.64854 0.01778 1455 28 0 1 0.00100 0.00780 0.00038 53 29 0 0.00093 0.00352 0.00010 16 30 1 0.04047 0.07390 0.00000 74 31 0 0.00000 0.00000 0.00000 0 32 4 0.01473 0.29899 0.00128 55 33 3 0.00024 0.00056 0.00000 0 34 3 0.03126 0.06723 0.00208 172 35 0 0.00263 0.00199 0.00002 43 36 4 0.00392 0.03327 0.00023 55 37 4 0.00006 0.00012 0.00000 12 38 0 0.00589 0.00282 0.00000 23 39 0 0.00010 0.00087 0.00010 7 40 0 0.00007 0.00007 0.00000 1 41 1 0.00056 0.00054 0.00000 28 42 4 0.38897 0.96420 0.13468 161 43 0 0 0.00276 0.00092 0.00000 1 44 0 1 0.00000 0.00005 0.00000 5 222 OBS GEO EXT RIOT ATT ASS DEATH 45 1 2 0.00125 0.00114 0.00011 379 46 0 4 0.15483 0.28887 0.00787 42 47 3 0.00341 0.06774 0.00082 1365 48 1 0.00010 0.00010 0.00000 2 49 1 0.00000 0.00000 0.00000 0 50 1 0.00779 0.00671 0.00022 105 51 0 0.00000 0.00000 0.00000 0 52 0 0.00180 0.00129 0.00026 6 53 0 0.000203 0.00061 0.000000 1 54 0 0.004890 0.00489 0.000000 0 55 0 0 0.000000 0.00000 0.000000 0 56 0 2 0.000000 0.00431 0.000000 0 57 0 2 0.000198 0.00040 0.000000 3 58 0 2 0.005589 0.00025 0.000509 15 59 0 3 0.005199 0.00297 0.000372 46 60 0 3 0.003436 0.00344 0.000000 0 61 0 2 0.000000 0.00199 0.000747 38 62 0 2 0.000232 0.00046 0.000000 1 63 0 2 0.000255 0.00026 0.000510 66 64 0 4 0.000186 0.00093 0.000000 0 65 0 2 0.001312 0.00000 0.000000 35 66 1 2 0.004528 0.00113 0.000000 27 67 1 3 0.002431 0.00232 0.000000 65 68 0 3 0.002548 0.00102 0.001020 1 69 1 2 0.000171 0.00051 0.000000 9 70 1 2 0.002756 0.01123 0.000127 1995235 71 0 2 0.017840 0.01390 0.000000 39 72 0 2 0.009877 0.00124 0.000000 11 73 1 3 0.004933 0.01905 0.000000 2264 74 1 3 0.017479 0.02242 0.006695 537 75 1 3 0.001293 0.02027 0.000046 9978 76 1 3 0.007518 0.00732 0.000306 30259 77 1 2 0.004524 0.00762 0.000356 363 78 1 2 0.000151 0.00151 0.000301 38 79 0 1 0.001193 0.00506 0.000597 81979 80 0 4 0.000000 0.00434 0.000000 26333 81 0 1 0.002149 0.00465 0.001075 30 82 1 2 0.000965 0.00977 0.000523 33623 83 1 3 0.002466 0.04399 0.000000 9673 84 0 3 0.000728 0.01781 0.000000 4631 85 1 3 0.009270 0.02415 0.000233 1295 86 1 1 0.011613 0.05801 0.000565 3853 87 0 3 0.002520. 0.00731 0.000229 90 88 1 2 0.007830 0.00584 0.000047 795 223 OBS GEO EXT RIOT ATT ASS DEATH 89 0 1 0.008592 0.01144 0.000958 24 90 0 1 0.000000 0.00324 0.000000 1 91 1 0 0.001586 0.00115 0.000144 183 92 0 3 0.006050 0.01086 0.000000 28 93 1 3 0.001321 0.00126 0.000066 105 94 1 4 0.000698 0.00522 0.000209 194 95 0 4 0.004467 0.00078 0.000195 16 96 0 4 0.005147 0.01027 0.000000 99 97 1 3 0.003054 0.00774 0.000000 4411 98 1 2 0.006257 0.00982 0.001163 2957 99 1 3 0.002098 0.00476 0.000482 229 100 1 3 0.000748 0.02439 0.000641 5231 101 0 4 0.001020 0.00078 0.000090 153 102 1 4 0.005269 0.01367 0.001120 1804 103 1 4 0.022032 0.63004 0.010476 60721 104 1 4 0.013505 0.06289 0.000000 2125 105 1 1 0.030956 0.08425 0.000338 216 106 0 4 0.000517 0.00155 0.000129 113 107 0 3 0.001213 0.02551 0.001559 7381 108 0 4 0.034899 0.26247 0.006940 1492 109 0 4 0.001315 0.00917 0.001315 1 110 1 1 0.000883 0.01410 0.000236 3549 111 1 4 0.000176 0.00062 0.000000 3481 112 0 1 0.000801 0.00000 0.000000 0 113 0 4 0.000356 0.00093 0.000000 478 114 0 4 0.000000 0.00014 0.000000 13 115 0 4 0.004009 0.00175 0.000032 200 116 0 0 0.000546 0.00041 0.000000 32 117 3 0.000823 0.00289 0.000057 4280 118 2 0.003913 0.00893 0.000413 310980 119 1 0.001405 0.00614 0.000036 1837 120 1 0.001836 0.00255 0.000080 5246 121 0 0.044452 0.00905 0.000000 0 122 1 0.002046 0.00142 0.000000 63 123 1 0.000448 0.00525 0.000084 1433 124 1 0.001981 0.15746 0.000142 56575 125 4 0.002697 0.13497 0.002360 22457 126 3 0.006381 0.00724 0.000191 476 127 3 0.001926 0.00241 0.000000 3 128 2 0.001169 0.01042 0.000292 4193 129 2 0.000527 0.00179 0.000000 575938 130 0 1 0.000637 0.00167 0.000080 0 131 1 1 0.000414 0.00372 0.000000 7 132 0 1 .0.000709 0.00035 0.000000 0 APPENDIX F CROSS-TABULATION AND REGRESSION RESULTS 225 TABLE 7. Cross-Tabulation of Economic Development with Riot Riot Economic Development Low High Total Low 35 31 66 High 31 35 66 Total 66 66 132 X2 = 0.49 P = 0.49 0 = 0..0 6 TABLE 8. Cross-Tabulation of Economic Development with Armed Attack Armed Attack Economic Development Low High Total Low 32 34 66 High 34 32 66 Total 66 66 132 2 X = 0.12 P = 0..7 3 Assassination Economic Development Low High Total Low 33 33 66 High 33 33 66 Total 66 66 132 X2 = 0.00 P = 1..0 0 0 = 0,.0 0 TABLE 10. Cross-Tabulation of Economic Development with Death from Violence" Death from Violence Economic Development Low High Total Low 28 38 66 High 38 28 66 Total 66 66 132 X2 = 3.03 P = 0.,08 * 0 = -0.1. 5 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 227 TABLE 11. Cross-Tabulation of Economic and Political Discrepancies with Riot* Economic and Riot Political Discrepancies Low High Total Low 24 17 41 High 15 26 41 Total 41 41 82 X2 = 3.96 P = 0.05** 0 = 0,.2 2 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 12. Cross-Tabulation of Economic and Political Discrepancies with Armed Attack" Economic and Armed Attack Political Discrepancies Low High Total Low 24 17 41 High 14 27 41 Total 41 41 82 X2 = 4.90 P = 0..03* * 0 = 0..2 5 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 228 TABLE 13. Cross-Tabulation of Economic and Political Discrepancies with Assassination" Economic and Assassination Political Discrepancies Low High Total Low 23 18 41 High 14 27 41 Total 41 41 82 X2 = 3.99 P = 0,.05* * 0 = 0.22 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 14. Cross-Tabulation of Economic and Political Discrepancies with Death from Violence Economic and Death from Violence Political Discrepancies Low High Total Low 20 21 41 High 13 28 41 Total 33 49 82 X2 = 2.49 P = 0.,1 2 0 = 0..1 7 22' TABLE 15. Cross-Tabulation of Governmental Coercion with Riot" Riot Governmental Sanction Low High Total Low 52 14 66 High 14 52 66 Total 66 66 132 X2 = 43.7 6 P = 0.00*** 0 = 0,.5 8 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 16. Cross-Tabulation of Governmental Coercion with Armed Attack" Armed Attack Governmental Sanction Low High Total Low 51 15 67 High 15 51 65 Total 66 66 132 X2 = 39.2 7 P = 0..00 * 0 = 0,.5 5 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 230 TABLE 17. Cross-Tabulation of Governmental Coercion with Assassination" Assassination Governmental Sanction Low High Total Low 45 21 66 High 21 45 66 Total 66 66 132 X2 = 17.4 6 P = 0..00** * 0 = 0..3 6 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 18. Cross-Tabulation of Governmental Coercion with Death from Violence Death from Violence Governmental Sanction Low High Total Low 37 29 66 High 29 37 66 Total 66 66 132 X2 = 1.94 P = 0.,1 6 0 = 0..1 2 231 TABLE 19. Cross-Tabulation of Role of Ethnic Elites with Riot Role of Riot Ethnic Elites Low High Total Low 21 22 43 High 18 21 39 Total 39 43 132 X2 = 0.06 P = 0,.8 1 0 = 0..0 3 TABLE 20. Cross-Tabulation of Role of Ethnic Elites with Armed Attack Role of Armed Attack Ethnic Elites Low High Total Low 23 20 43 High 15 24 39 Total 38 44 82 X2 = 1.86 P = 0..1 7 0 = 0.,1 5 232 TABLE 21. Cross-Tabulation of Role of Ethnic Elites with Assassination Role of Assassination Ethnic Elites Low High Total Low 19 24 43 High 18 21 39 Total 37 45 82 x2 = o..0 3 P = 0.86 0 = •0.02 TABLE 22. Cross-Tabulation of Role of Ethnic Elites with Death from Violence" Role of Death from Violence Ethnic Elites Low High Total Low 23 20 43 High 10 29 39 Total 33 49 82 X2 = 6.60 P = 0.101** * 0=0. 28 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 233 TABLE 23. Cross-Tabulation of Geographic Concentration with Riot Riot Geographic Concentration Low High Total Low 36 29 65 High 30 37 67 Total 66 66 132 X2 = 1-49 P = 0.22 0 = 0.,1 1 TABLE 24. Cross-Tabulation of Geographic Concentration with Armed Attack" Armed Attack Geographic Concentration Low High Total Low 39 26 65 High 27 40 67 Total 66 66 132 X2 = 5.12 P = 0,.02 " 0 = 0..2 0 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 234 TABLE 25. Cross-Tabulation of Geographic Concentration with Assassination Assassination Geographic Concentration Low High Total Low 34 31 65 High 32 35 67 Total 66 66 132 X2 = 0.27 P = 0..6 0 0 = 0..0 5 TABLE 26. Cross-Tabulation of Geographic Concentration with Death from Violence" Death from Violence Geographic Concentration Low High Total Low 44 21 65 High 22 45 67 Total 66 66 132 X2 = 16.03 P = 0.00*** 0 = 0.35 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 235 TABLE 27. Cross-Tabulation of External Links with Riot" Riot External Links Low High Total Low 45 35 80 High 21 31 52 Total 66 66 132 X2 = 3.17 P = 0.08* 0 = 0..1 6 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 28. Cross-Tabulation of External Links with Armed Attack Armed Attack External Links Low High Total Low 44 36 80 High 22 30 52 Total 66 66 132 X2 = 2.03 P = 0..1 5 0 = 0.,1 2 23 TABLE 29. Cross-Tabulation of External Links with Assassination Assasssinatio n External Links Low High Total Low 41 39 80 High 25 27 52 Total 66 66 132 X2 = 0.13 P = 0..7 2 0 = 0..0 3 TABLE 30. Cross-Tabulation of External Links with Death from Violence" Death from Violence External Links Low High Total Low 46 34 80 High 20 32 52 Total 66 66 132 X2 = 4.57 P = 0.,03 " 0=0,,1 9 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes 237 TABLE 32. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Economic Development Economic Economic Development Development Low High Riot Riot Ethnic Diversi-t y Low High Total Low High Total Low 13 8 21 24 21 45 High 22 23 45 7 14 21 Total 35 31 66 31 35 66 2 x = 0.9 7 X2 = 2.30 P =•• 0 .32 P =• 0.13 0 =: 0.1 2 0 =• 0.19 TABLE 33. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Economic Development Economic Economic Development Development Low High Armed Attack Armed Attack Ethnic Diversity Low High rotal Low High Total Low 9 12 21 26 19 45 High 23 22 45 8 13 21 Total 32 34 66 34 32 66 x2 = 0.39 x2 = 2.22 = 0.53 = 0.14 P =-0.08 P = 0.18 238 TABLE 34. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Economic Development Economic Economic Development Development Low High Assassination Assassination Ethnic Diversity Low High Total Low High Total Low 10 11 21 21 24 45 High 23 22 45 12 9 21 Total 33 33 66 33 33 66 X2 = 0.07 X2 = 0.63 P == 0.79 P == 0.43 0 ==-0.0 3 0 ==-0.1 0 TABLE 35. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Economic Development" Economic Economic Development Development Low High Death from Violence Death from V;Lolenc e Ethnic Diversity Low High 'Tota l Low High Total Low 7 14 21 29 16 45 High 21 24 45 9 12 21 Total 28 38 66 38 28 66 X2 = 1-04 X2 - 2.73 P = 0.31 P =•- 0.10* 0 =-0.13 0 == 0.20 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 239 TABLE 36. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Economic and Political Discrepancies Economic and Political Economic and Political Discrepancies Discrepancies Low High Riot Riot Ethnic Diversity Low High Total Low High Total Low 17 11 28 6 11 17 High 7 6 13 9 15 24 Total 24 17 41 15 26 41 x2 = 0.17 x2 = 0.02 = 0.68 = 0.89 P = 0.07 P =-0.02 TABLE 37. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Economic and Political Discrepancies Economic and Political Economic and Political Discrepancies Discrepancies Low High Armed Attack Armed Attack Ethnic Diversity Low High Total Low High Total Low 18 10 28 6 11 17 High 6 7 13 8 16 24 . Total 24 17 41 14 27 41 X2 - 1.20 X2 - 0.02 P = 0.27 P = 0.90 0 = 0.17 0 = 0.02 240 TABLE 38. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Economic and Political Discrepancies" Economic and Political Economic and Political Discrepancies Discrepancies Low High Assassination Assassination Ethnic Diversity Low High Total Low High Total Low 16 12 28 3 14 17 High 7 6 13 11 13 24 Total 23 18 41 14 27 41 C N XC M •© - l " I ' 0.04 x2 =' 3.52 0.84 P == 0.06* 0.03 0 ==-0.2 9 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 39. Cross-Tabulation of Ethnic Diversity with Death from violence Controlling for Economic and Political Discrepancies Economic and Political Economic and Political Discrepancies Discrepancies Low High Death from Violence Death from Violence liunnic Diversity Low High Total Low High Total Low 16 12 28 7 10 17 High 4 9 13 6 18 24 Total 20 21 41 13 28 41 "XCM-© - l " I ' 2.47 X2 - 1.20 0.12 P = 0.27 0.25 0 = 0.17 241 TABLE 40. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Governmental Coercion" Governmental Governmental Coercion Coercion Low High Riot Riot Ethnic Diversity Low High Total Low High Total Low 27 5 33 10 24 34 High 25 9 34 4 28 32 Total 52 14 66 14 52 66 X2 = 1. 16 X2 = 2.82 P == 0 .28 P =•• 0.09* 0 == 0 .13 0 == 0.21 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 41. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Governmental Coercion" Governmental Governmental Coercion Coercion Low High Armed Attack Armed Attack Ethnic Diversity Low High Total Low High Total Low 30 2 32 5 29 34 High 21 13 34 10 22 32 Total 51 15 66 15 51 66 X2 = 9.60 X2 => 2.56 P = 0.00*" P == 0.11 0 = 0.38 =-0.20 0 = a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 242 TABLE 42. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Governmental Coercion Governmental Governmental Coercion Coercion Low High Asssassinatio ri Assassination Ethnic Diversity Low High Total Low High Total Low 22 10 32 9 25 34 High 23 11 34 12 20 32 Total 45 21 66 21 45 66 2 X2 = 0.01 x = 0.92 P == 0.92 = 0.34 0 == 0.01 P ==-0.1 2 0 = TABLE 43. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Governmental Coercion Governmental Governmental Coercion Coercion Low High Death from Violence Death from ViLolenc e Ethnic Diversity Low High Total Low High Total Low 20 12 32 16 18 34 High 17 17 34 13 19 32 Total 37 29 66 29 37 66 X2 = 1-05 0.27 P = 0.31 » 0.60 0 = 0.13 =I M 0.07 -©••o x 243 TABLE 44. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Role of Ethnic Elites ]Rol e of ]Rol e of ]Ethni c Elites ]Ethni c Elites Low High Riot Riot Ethnic Diversity Low High Total Low High Total Low 18 17 35 5 5 10 High 3 5 8 13 16 29 Total 21 22 43 18 21 39 x2 = 0.51 x2 = 0.08 = 0.48 = 0.78 P P = 0.11 = 0.05 = 0.38 0 Fisher P TABLE 45. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Role of Ethnic Elites Role of ]Rol e of Ethnic Elites ]Ethni c Elites Low High Armed Attack Armed Attack Ethnic Diversity Low High Total Low High Total Low 19 16 35 5 5 10 High 4 4 8 10 19 29 Total 21 22 43 15 24 39 x2 = 0.05 x2 = 0.76 = 0.83 = 0.38 P = 0.03 P = 0.14 Role of Role of Ethnic Elites Ethnic Elites Low High Assassination Assassination liunnxDiversitc y — Low High Total Low High Total Low 14 21 35 5 5 10 High 5 3 8 13 16 29 Total 19 24 43 18 21 39 X2 = 1.34 x2 = 0.08 P = 0.25 P = 0.78 0 =-0.18 TABLE 47. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Role of Ethnic Elites Role of Role of Ethnic Elites Ethnic Elites Low High Death from Violence Death from Violence Ethnic Diversity Low High Total Low High Total Low 19 16 35 4 6 10 High 4 4 8 6 23 29 Total 23 20 43 10 29 39 X2 = 0.05 X2 = 1.45 P = 0.83 P = 0.23 0 = 0.03 0 = 0.19 Fis3he r P =0.57 Fisher P = 0.21 245 TABLE 48. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Geographic Concentration" Geographic Geographic Concentration Concentration Low High Riot Riot Ethnic Diversity Low High Total Low High Total Low 24 20 44 13 9 22 High 12 9 21 17 28 45 Total 36 29 65 30 37 67 X2 =' 0.0 4 2.71 P == 0 .84 = 0.10* =-0 .02 =I Q.20 0 = •©••o x a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 49. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Geographic Concentration" Geographic Geographic Concentration Concentration Low High Armed Attack Armed Attack Ethnic Diversity Low High Total Low High Total Low 23 21 44 12 10 22 High 16 5 21 15 30 45 Total 39 26 65 27 40 67 X2 =•• 3.39 X2 = 2.76 P == 0.07* P == 0.10* 0 ==-0.2 3 0 =* 0.20 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 246 TABLE 50. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Geographic Concentration Geographic Geographic Concentration Concentration Low High Assassination Assassination Ethnic Diversity Low High Total Low High Total Low 20 24 44 11 11 22 High 14 7 21 21 24 45 Total 34 31 65 32 35 67 2 X2 - 2.56 x = 0.07 P == 0.11 P == 0.80 0 ==-0.2 0 0 == 0.03 TABLE 51. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Geographic Concentration" Geographic Geographic Concentration Concentration Low High Death from Violence Death from V:Lolenc e Ethnic Diversity Low High Total Low High Total Low 25 19 44 11 11 22 High 19 2 21 11 34 45 Total 44 21 65 22 45 67 2 x = 7.36 X2 = 4.38 = 0.01*" P =-- 0.04" P =-0.34 0 == 0.26 a: * denotes P < 0. 1, ** denotes P < 0.05, and *** denotes P < 0.01. 247 TABLE 52. Cross-Tabulation of Ethnic Diversity with Riot Controlling for External Links External Links External Links Low High • Riot Riot Ethnic DiversiLt y Low High Total Low High Total Low 25 16 41 12 13 25 High 20 19 39 9 18 27 Total 45 35 80 21 31 52 2 x = 0.76 X2 1.16 0.38 P 0.28 P 0.10 : 0.15 0 TABLE 53. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for External Links External Links External Links Low High Armed Attack Armed Attack Ethnic Diversity Low High Total Low High Total Low 23 18 41 12 13 25 High 21 18 39 10 17 27 Total 44 36 80 22 30 52 X2 0.41 X2 -« 0.64 P 0.84 P == 0.42 0 0.02 0 == 0.11 248 TABLE 54. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for External Links External Links External Links Low High Assassination Assassination Ethnic Diversj-t y Low High Total Low High Total Low 20 21 41 11 14 25 High 21 18 39 14 13 27 Total 41 39 80 25 27 52 x2 = 0.21 x2 = 0.32 = 0.65 = 0.57 P =-0.05 P =-0.08 TABLE 55. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for External Links External Links External Links Low High Death from Violence Death from Violence DiversjLt y Low High Total Low High Total Low 26 15 41 10 15 25 High 20 19 39 10 17 27 Total 46 34 80 20 32 52 C N XC M •© • l I ' " 1.20 X2 - 0.05 0.27 P = 0.83 0.12 0 = 0.03 249 TABLE 57. Regression Results of Violent Political Behavior on Ethnic Diversity* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.018 0.065 0.004 -27309 DIV -0.022 -0.057 -0.006 156546 (-1.39) (-0.78) (-1.27) (2.13)" R2 0.015 0.005 0.012 0.034 P Value 0.168 0.437 0.205 0.035" N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 250 TABLE 58. Regression Results of Violent Political Behavior on Ethnic Diversity in Quadratic Form* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.018 0.028 0.005 29211 DIV -0.027 0.257 -0.011 -324461 (-0.44) (0.90) (-0.55) (-1.14) DIV2 0.000 -0.433 0.006 664256 (0.99) (-1.14) (0.23) (1-74)* R2 0.015 0.015 0.013 0.056 P Value 0.386 0.389 0.438 0.025" R2 - R2 0.000 0.010 0.000 0.022 - o r F Value 0.00 1.27 0.05 2.59* N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 251 TABLE 59. Regression Results of Violent Political Behavior on Ethnic Diversity and Economic Development* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.017 0.052 0.004 -22953 DIV -0.020 -0.039 -0.006 150860 (-1.21) (-0.50) (-1.19) (1.91)* ECO 3xl0"7 4.6xl0-6 -9xl0"9 -1.437 (0.22) (0.65) (-0.02) (0.20) R2 0.015 0.008 0.012 0.034 P Value 0.378 0.601 0.449 0.108* R2 - R2 0.000 0.003 0.000 0.000 o r F Value 0.05 0.41 0.00 0.04 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 252 TABLE 60. Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Economic Development* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.015 0.037 0.004 -13682 DIV -0.017 -0.001 -0.006 126795 (-0.96) (-0.01) (-1.11) (1.52) ECO lxlO"6 lxlO"5 6xl0-9 -6.141 (0.53) (1.38) (0.01) (-0.70) IECO 5xl0'6 6x10-" lxlO"7 -38.050 (0.58) (1.44) (0.04) (-0.90) R2 0.018 0.024 0.012 0.040 P Value 0.516 0.378 0.661 0.155 R2 - R2 0.003 0.016 0.000 0.006 o r F Value 0.34 2.05 0.00 0.80 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 253 TABLE 61. Regression Results of Violent Political Behavior on Ethnic Diversity and Economic and Political Discrepancies* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.024 0.079 0.006 -47402 DIV -0.030 -0.042 -0.009 325354 (-1.15) (-0.34) (-1.03) (2.65)"* DIS -0.003 -0.020 -0.001 -57895 (-0.31) (-0.41) (-0.34) (-1.19) R2 0.020 0.004 0.017 0.087 P Value 0.447 0.840 0.506 0.028" R2 - R2 0.005 0.000 0.005 0.049 c r F Value 0.71 0.00 0.61 6.54" N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 254 TABLE 62. Regression Results of Violent Political Behavior . on Ethnic Diversity and Interaction with Economic and Political Discrepancies* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.024 0.079 0.006 -47714 DIV -0.029 -0.040 -0.008 317250 (-1.08) (-0.32) (-0.97) (2.57)** DIS -0.008 -0.026 -0.003 -33094 (-0.67) (-0.45) (-0.63) (-0.57) IDIS 0.038 0.048 0.010 -187000 (0.74) (0.20) (0.63) (-0.78) R2 0.027 0.005 0.022 0.094 P Value 0.541 0.943 0.624 0.052" R2 - R2 0.007 0.001 0.005 0.045 c r F Value 0.55 0.04 0.40 3.67* N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 255 TABLE 63. Regression Results of Violent Political Behavior on Ethnic Diversity and Governmental Coercion" Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.011 0.011 1.6xl0"4 -26282 DIV -0.003 0.005 -0.002 155375 (-0.34) (0.08) (-0.48) (2.09)" COER 0.379 1.238 0.089 -23425 (18.55)*" (8.45)*" (9.36)*" (-0.13) R2 0.731 0.359 0.412 0.034 P Value 0.000"* 0.000"* 0.000"* 0.109* R2 - R2 0.717 0.355 0.400 0.000 F Value 93.80*" 71.40*" 87.60"* 0.01 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 256 TABLE 64. Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Governmental Coercion* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.002 -0.059 0.002 -12599 DIV -0.005 0.140 -0.005 128826 (-0.51) (2.55)*" (-1.10) (1.61)* COER 0.359 2.818 0.058 -332944 (8.28)*" (10.53)*" (2.90)*" (-0.86) ICOER -0.105 8.098 -0.159 -1586641 (0.53) (6.70)*" (-1.77)" (-0.90) R2 0.732 0.526 0.426 0.040 P Value 0.000*" 0.000*" 0.000*" 0.156 R2 - R2 0.001 0.166 0.014 0.006 c r F Value 0.29 33.20*" 3.07" 0.81 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 257 TABLE 65. Regression Results of Violent Political Behavior on Ethnic Diversity and Role of Elites* Dependent Variabl e Inde pendent Armed Assassi Death from Variable Riot- Attack nation Violence Constant 0.024 0.085 0.006 -4931 DIV -0.036 -0.124 -0.010 -162505 (-1.11) (-0.83) (-0.98) (1.33) ELT 0.002 0.039 4x10-" 245281 (0.20) (0.83) (0.12) (6.39)*" R2 0.019 0.011 0.016 0.387 P Value 0.460 0.006 0.531 0.000*" R2 - R2 0.005 0.079 0.004 0.353 c r F Value 0.63 0.79 0.47 47.13"* N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 258 TABLE 66. Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Role of Elites* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.024 0.096 0.006 -40214 DIV -0.035 -0.167 -0.009 -29563 (-1.07) (-1.10) (-0.90) (-0.28) ELT 0.002 0.096 3.3xl0-4 65488 (0.10) (1.49) (-0.07) (1.47) IELT 0.002 -0.252 0.003 789250 (0.05) (-1.29) (0.24) (5.86)*" R2 0.020 0.032 0.017 0.574 P Value 0.672 0.473 0.726 0.000*" R2 - R2 0.000 0.021 0.001 0.187 c r F Value 0.08 1.63 0.06 23.83*" N 82 82 82 82 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 259 TABLE 67. Regression Results of Violent Political Behavior on Ethnic Diversity and Geographic Concentration* Dependent Variablee Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.016 0.048 0.090 -35881 DIV -0.027 -0.095 0.270 137928 (-1.63)* (-1.27) (1.00) (1.80)* GEO 0.008 0.060 0.168 29148 (1.80) (1.88)* (0.45) (0.90) R2 0.024 0.031 0.032 0.040 P Value 0.216 0.130 0.126 0.074* R2 - R2 0.009 0.026 0.019 0.006 e r F Value 1.17 3.42* 2.52 0.80 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 260 TABLE 68. Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with Geographic Concentration* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.017 0.051 0.004 -44405 DIV -0.027 -0.096 -0.008 139032 (-1.64)* (-1.27) (-1.59)* (1.83)* GEO 0.008 0.060 0.003 2954 (1.08) (1.87)" (1.30) (0.91) IGEO -0.033 -0.117 -0.012 276024 (-1.01) (-0.78) (-1.20) (1.82)* R2 0.031 0.036 0.036 0.064 P Value 0.253 0.198 0.194 0.037** R2 - R2 0.008 0.005 0.004 0.024 c r F Value 1.01 0.61 0.58 3.23* N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 261 TABLE 69. Regression Results of Violent Political Behavior on Ethnic Diversity and External Links* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.011 0.043 0.002 -20543 DIV -0.030 -0.082 -0.009 164041 (-1.85)* (-1.10) (-1.69)* (2.16)" EXT 0.005 0.016 0.001 -4743 (2.05)" (1.42) (1.86)* (-0.42) R2 0.046 0.020 0.038 0.035 P Value 0.049" 0.272 0.081* 0.101* R2 - R2 0.031 0.015 0.026 0.001 c r - F Value 4.06" 1.98 3.38* 0.17 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 262 TABLE 70. Regression Results of Violent Political Behavior on Ethnic Diversity and Interaction with External Links* Dependent Variable Inde pendent Armed Assassi Death from Variable Riot. Attack nation Violence Constant 0.016 0.063 0.003 -5977 DIV -0.024 -0.062 -0.007 178790 (-1.47) -0.80) (-1.38) (2.28)" EXT 0.002 0.005 6.7x10-* -12564 (0.63) (0.35)" (0.68) (-0.83) IEXT -0.022 -0.081 -0.006 -58891 (-1.43) -1.12) (-1.13) (-0.80) R2 0.061 0.029 0.048 0.040 P Value 0.045" 0.280 0.098 0.157 R2 - R2 0.015 0.009 0.010 0.005 F Value 2.01 1.23 0.01 0.62 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. APPENDIX G STATISTICAL RESULTS WHEN THE INDIRECT MEASURE OF GOVERNMENTAL COERCION IS USED 263 264 TABLE 72. Cross-Tabulation of Governmental Coercion (Indirect Measure) with Riot Riot Governmental Coercion Low High Total Low 32 36 68 High 34 30 64 Total 66 66 132 x2 = o,.4 9 P = 0.49 0 = -0.06 TABLE 73. Cross-Tabulation of Governmental Coercion (Indirect Measure) with Armed Attack Armed Attack Governmental Coercion Low High Total Low 31 37 68 High 35 29 64 Total 66 66 132 x2 = i..0 9 P = 0.30 0 = -0.09 265 TABLE 74. Cross-Tabulation of Governmental Coercion (Indirect Measure) with Assassination Assassination Governmental Coercion Low High Total Low 33 35 68 High 33 31 64 Total 66 66 132 x2 = o..1 2 P = 0..7 3 0 = -0.03 TABLE 75. Cross-Tabulation of Governmental Coercion (Indirect Measure) with Death from Violence Death from Violence Governmental Coercion Low High Total Low 34 34 68 High 32 32 64 Total 66 66 132 X2 = 0.00 P = 1..0 0 0 = 0..0 0 266 TABLE 76. Cross-Tabulation of Ethnic Diversity with Riot Controlling for Governmental Coercion" (Indirect Measure) Governmental Governmental Coercion Coercion Low High Riot Riot Ethnic Diversity Low High Total Low High Total Low 18 19 37 19 10 29 High 14 17 31 15 20 35 Total 32 36 68 34 30 64 C N XCM-© - x2 = 0.0 8 l " » 3.27 P =•• 0 .77 = 0.07* 0 =• 0 .04 = 0.23 a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. TABLE 77. Cross-Tabulation of Ethnic Diversity with Armed Attack Controlling for Governmental Coercion (Indirect Measure) Governmental Governmental Coercion Coercion Low High Armed Attack Armed Attack nunnic Diversity Low High Total Low High Total Low 16 21 37 19 10 29 High 15 16 31 16 19 35 Total 31 37 68 35 29 64 X2 = 0.18 X2 = 2.51 P == 0.67 P == 0.11 0 ==-0.5 1 0 == 0.20 267 TABLE 78. Cross-Tabulation of Ethnic Diversity with Assassination Controlling for Governmental Coercion (Indirect Measure) Governmental Governmental Coercion Coercion Low High Asisassinatio ni Asisassinatio n Ethnic Diversity Low High Total Low High Total Low 15 22 37 16 13 29 High 18 13 31 17 18 35 Total 33 35 68 33 31 64 M 2.07 0.28 • 0.15 • 0.60 <-0.1I 8 'I H 0.07 -©-•o x •©••o x TABLE 79. Cross-Tabulation of Ethnic Diversity with Death from Violence Controlling for Governmental Coercion" (Indirect Measure) Governmental Governmental Coercion Coercion Low High Death from Violence Death from VjLolenc e Ethnic Diversity Low High Total Low High Total Low 22 15 37 14 15 29 High 12 19 31 18 17 35 Total 34 34 68 32 32 64 K ) 2.91 X2 = 0.06 •- 0.09* P = 0.80 --i I 0.21 0 =-0.03 •©-•o x a: * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 268 TABLE 80. Regression Results of Violent Political Behavior on Ethnic Diversity and Governmental Coercion" (Indirect Measure) Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.029 0.131 0.007 -20870 DIV -0.013 -0.005 -0.004 161679 (-0.80) (-0.06) (-0.82) (2.09)" COER -0.003 -0.019 -0.001 -1837 (-1.83)* (-2.35)" (-1.34) (-0.22) R2 0.040 0.046 0.026 0.034 P Value 0.074* 0.049* 0.185 0.107* R2 - R2 0.025 0.041 0.014 0.000 c r F Value 3.27* 5.79" 1.78 0.05 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 269 TABLE 81. Regression Results of Violent Political . Behavior on Ethnic Diversity and Interaction with Governmental Coercion" (Indirect Measure) Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.028 0.143 0.006 -17513 DIV -0.014 0.010 -0.005 165805 (0.86) (0.14) (-0.96) (2.10)** COER -0.003 -0.021 -0.001 -2579 (-1.60)* (-2.56)"* (-1.02) (-0.30) IGOV 0.004 -0.042 0.002 -11420 (0.44) (-1.05) (0.81) (-0.28) R2 0.041 0.054 0.031 0.035 P Value 0.146 0.069* 0.259 0.210 R2 - R2 0.002 0.008 0.005 0.001 o r F Value 1.20 1.10 0.66 0.07 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. APPENDIX H STATISTICAL RESULTS WHTH THE QUADRATIC FORMS OF ECONOMIC DEVELOPMENT AND GOVERNMENTAL COERCION 270 271 TABLE 82. Regression Results of Violent Political Behavior on Ethnic Diversity, Economic Development, and Economic Development in Quadratic Form" Dependent Variable Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant 0.012 0.026 0.003 -18665 DIV -0.015 -0.014 -0.005 146681 (-0.90) (-0.18) (-0.99) (1.81)* ECO 4.6xl0-6 3xl0"5 8xl0"7 -5.031 (1.40) (1.69)* (0.76) (-0.32) ECO2 -5xl0"10 -2xl0"9 -lxlO"10 -0.0004 (-1.46) (-1.57) (-0.86) (0.26) R2 0.031 0.027 0.018 0.034 P Value 0.256 0.326 0.506 0.212 R2 - R2 0.016 0.019 0.006 0.000 o r F Value 2.08 2.45 0.009 0.000 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. 272 TABLE 83. Regression Results of Violent Political Behavior on Ethnic Diversity, Governmental Coercion, and Governmental Coercion in Quadratic Form* Dependent Variabl. e Inde pendent Armed Assassi Death from Variable Riot Attack nation Violence Constant -0.003 -0.035 -0.010 -21592 DIV -0.004 -0.007 -0.002 156578 (-0.49) (-0.15) (-0.58) (2.10)" COER 0.671 4.737 0.188 -377671 (10.25)"* (12.09)*" (5.96)*" (-0.59) COER2 -0.458 -5.482 -0.155 555079 (-4.66)*" (-9.33)*" (-3.29)*" (0.58) R2 0.770 0.619 0.458 0.036 P Value 0.000*" 0.000*" 0.000*" 0.191 R2 - R2 0.039 0.260 0.046 0.002 c r F Value 18.56*" 51.92*" 10.01*" 0.265 N 132 132 132 132 a: Numbers in parentheses are t statistics. * denotes P < 0.1, ** denotes P < 0.05, and *** denotes P < 0.01. APPENDIX I PEARSON CORRELATION COEFFICIENTS AMONG ALL VARIABLES INCLUDING THE INTERACTION TERMS 213 Table 86. Pearson Correlation Coefficients Among All Variables Including the Interaction Terms3 DIV ECO DISQ COEH ELT GEO EXT HRIOT HATT oiv 1.00000 -0.35569 0.19795 -0.12350 0.58900 0.27170 0.23777 -0.12078 -0.06824 o.o 0.0001 0.0746 0.1583 0.0001 0.C016 0.0060 0.1678 0.4369 132 133 82 133 82 133 132 133 132 ECO -0.35569 1.00000 -0.20783 0.09176 -0.27981 0.049U -0.09639 0.05128 0.07728 0.0001 0.0 0.0610 0.2954 0.0109 0.5760 0.2715 0.4851 0.3785 132 132 82 132 83 133 132 132 132 DISQ 0.19795 -0.20783 1.00000 -0.06393 0.00222 0.05722 -0.03142 -0.06151 -0.05442 0.0746 0.0610 0.0 0.5744 0.9842 0.6096 0.7793 0.5830 0.6273 82 83 82 83 82 82 82 82 83 COER -0.12350 0.09176 -0.06293 1.00000 •0.05330 0.03409 0.17285 0.85507 0.59951 0.1583 0.3954 0.5744 0.0 0.6344 0.7840 0.0475 0.0001 0.0001 132 133 82 133 82 132 132 132 133 ELT 0.58900 -0.37981 0.00232 -0.05330 1.00000 0.40485 0.28176 -0.06342 0.04631 0.0001 0.0109 0.9842 0.6344 0.0 0.0002 0.0103 0.5714 0.6795 83 03 82 83 82 82 82 83 82 GEO 0.37170 0.04911 0.05722 0.03409 0.40485 1.00000 0.08625 0.05796 0.13809 0.0016 0.5760 0.6096 0.7840 0.0003 0.0 0.3354 0.5093 0.1143 132 133 82 133 83 132 133 133 132 EXT 0.33777 -0.09639 -0.03142 0.17385 0.38176 0.08635 1.00000 0.14339 0.10393 0.0060 0.3715 0.7793 0.0475 0.0103 0.3354 0.0 0.1036 0.2357 133 133 82 132 82 133 133 133 132 HUOT -0.13078 0.06138 -0.06151 0.85507 -0.06342 0.05796 0.14339 1.00000 0.69596 0.1678 0.4BS1 0.5830 0.0001 0.5714 0.5093 0.1036 0.0 0.0001 133 132 82 132 82 132 133 133 132 WOt -0.06834 0.07728 -0.05442 0.59951 0.04631 0.13809 0.10393 0.69596 1.00000 0.4369 0.3785 0.6273 0.0001 0.6795 0.1143 0.33S7 0.0001 0.0 133 132 82 132 83 133 133 132 132 NAS3 -0.11100 0.03791 -0.06184 0.64092 -0.06360 0.07872 0.12992 0.91321 0.58513 0.2051 0.6661 0.5610 0.0001 0.S763 0.3696 0.1376 0.0001 0.0001 132 133 82 132 83 132 133 132 132 SHEATH 0.18353 -0.08164 -0.07345 •0.03360 0.61064 0.12426 0.00877 -0.02701 -0.01637 0.0352 0.3520 0.5119 0.7021 0.0001 0.1557 0.9305 0.7585 0.8533 132 132 82 132 63 132 133 132 133 IECO -0.06595 -0.53065 0.12269 0.01451 -0.04313 -0.06820 0.01351 0.03740 0.07301 0.4525 0.0001 0.2722 0.8688 0.7004 0.4372 0.8778 0.6703 0.4119 132 132 82 133 82 132 133 132 133 IDISQ 0.03147 0.07545 0.54408 0.06183 0.02151 0.08583 0.00343 0.04503 -0.00800 0.7315 0.5005 0.0001 0.5811 0.8479 0.4433 0.9756 0.6679 0.9431 82 82 82 83 83 83 83 82 63 ICOER -0.07838 0.03300 0.07333 -0.86198 -0.0095S -0.03801 -0.3513S -0.74574 -0.33557 0.3717 0.7072 0.5127 0.0001 0.9333 0.6653 0.0036 0.0001 0.0001 132 132 82 132 82 133 133 132 133 IELT 0.3908S -0.10092 0.01832 -0.02255 0.70459 0.OS153 0.03955 -0.02143 -0.0F263 0.0080 0.3670 0.8703 0.8406 0.0001 0.6457 0.7922 0.8485 0.63B7 82 82 82 82 82 83 82 82 82 IGEO -0.01003 -0.06493 0.17494 -0.02434 0.28549 -0.00855 -0.05944 -0.08743 -0.06776 0.9092 0.4595 0.1160 0.7818 0.0093 0.9224 0.4984 0.3188 0.4401 132 132 83 133 82 132 132 132 132 izn 0.03079 0.01328 -0.03656 -0.36447 -0.04863 -0.06139 -0.62483 -0.21135 -0.15629 0.7260 0.8798 0.8138 0.0033 0.6644 0.4844 0.0001 0.0150 0.0735 133 132 83 133 82 132 133 132 132 Table 86. (continued) IDISQ -0.11100 0.18353 -0.06595 0.03847 -0.07838 0.29085 -0.01003 0.03079 0.2051 0.0352 0.4535 0.7315 0.3717 0.0080 0.9092 0.7260 133 132 133 S3 133 S3 133 132 0.03791 -0.08164 -0.53065 0.07545 0.03300 -0.10093 -0.06493 0.01328 0.6661 0.3520 0.0001 0.5005 0.7072 0.3670 0.4595 0.8798 133 132 133 82 132 S3 132 132 DISQ -0.06184 -0.07345 0.12369 0.54408 0.07332 0.01832 0.17494 -0.02656 0.5810 0.5119 0.3722 0.0001 0.5127 0.8703 0.1160 0.812S S3 82 82 82 S3 82 82 82 0.64093 -0.03360 0.01451 0.06183 -0.86198 -0.02255 -0.02434 -0.26447 0.0001 0.7021 0.8688 0.5811 0.0001 0.8406 0.7818 0.0023 133 133 132 82 133 82 132 133 -0.06360 0.61064 -0.04313 0.02151 -0.00955 0.70459 0.2SS49 -0.04863 0.5763 0.0001 0.7004 0.8479 . 0.9333 0.0001 0.0093 0.6644 83 82 82 82 82 82 82 83 0.07873 0.13436 -0.06820 0.08583 -0.03801 0.05152 -0.00B55 -0.06139 0.3696 0.1557 0.4372 0.4433 0.6653 0.64S7 0.9224 0.4844 133 132 132 82 133 82 133 133 0.13993 0.00077 0.01351 0.00343 -0.35135 0.029SS -0.05944 -0.63482 0.1376 0.9305 0.8778 0.9756 0.0036 0.7922 0.4984 0.0001 133 132 132 82 132 82 133 132 0.91331 -0.02701 0.03740 0.04503 -0.74574 -0.02143 -0.08743 -0.31135 0.0001 0.7585 0.6703 0.6879 0.0001 0.8485 0.3188 0.0150 133 133 132 82 132 82 133 133 0.58513 -0.01637 0.07301 -0.00800 -0.33557 -0.05363 -0.06776 -0.15629 0.0001 0.8533 0.4119 0.9431 0.0001 0.6387 0.4401 0.0735 132 133 133 83 133 82 133 132 1.00000 -0.017S3 0.01137 0.03364 -0.60245 -0.00SS6 -0.10390 -0.18233 0.0 0.8393 0.8979 0.7641 0.0001 0.9370 0.3358 0.0364 132 132 133 83 132 S3 133 133 -0.01782 1.00000 -0.06404 -0.13035 -0.04156 0.74833 0.15338 -0.03333 0.8392 0.0 0.4657 0.3431 0.6361 0.0001 0.0791 0.7906 132 132 133 83 132 83 133 133 0.01127 -0.06404 1.00000 -0.13610 0.04338 -0.33063 0.09461 -0.1S138 0.8979 0.4657 0.0 0.3590 0.6214 0.0371 0.3805 0.0834 132 133 133 83 132 83 133 133 IDISQ 0.03364 -0.13035 -0.13610 1.00000 -0.03653 0.03676 -0.03393 -0.07115 0.7641 0.3431 0.3590 0.0 0.8130 0.8114 0.8381 0.5353 82 83 83 83 83 S3 83 S3 -0.60245 -0.04156 0.04338 -0.03652 1.00000 -0.05692 0.03483 0.22852 0.0001 0.6361 0.6314 0.8130 0.0 0.6115 0.7775 0.0084 132 132 133 82 133 82 133 132 -0.00886 0.74823 -0.33063 0.02676 -0.05693 1.00000 0.33743 0.16556 0.937O 0.0001 0.0371 0.8114 0.6115 0.0 0.0019 0.1373 83 83 82 82 S3 82 82 82 -0.10390 0.15338 0.09461 -0.02293 0.03483 0.33742 1.00000 0.07845 0.3358 0.0791 0.2805 0.8381 0.7775 0.0019 0.0 0.3713 133 133 133 82 133 82 132 132 -0.18333 -0.03333 -0.15130 -0.07115 0.33853 0.16556 0.07845 1.00000 0.0364 0.7906 0.0834 0.5353 0.0084 0.1372 0.3713 0.0 133 132 133 83 133 83 132 133 a: Pwirion correlation Co.fflcl.nt. / Prob > |R| unrt>r Ho. RlwO / WoaMr o« Obwrv«ton« APPENDIX J VALUES FOR EACH VARIABLE IN THE ASCENDING ORDER 277 OBS DIV I JPAN 0.017>i 57 FRNC 0.2664 113 GHNA 0.5926 2 YMNS 0.0185 58 JMCA 0.2729 114 SRLE 0.5976 3 NRWY 0.0195 59 AUSL 0.2797 115 ZMBA 0.5989 4 IRLD 0.0225 60 MNGL 0.2815 116 GNEA 0.60S9 5 SDAR 0.0295 61 MRTN 0.2818 117 SDAN 0.6263 6 BRZL 0.03S5 62 HNDS 0.2869 118 MALI 0.6283 7 DMNR 0.0381 63 LSTO 0.2985 119 MLYS 0.6299 8 SWDN 0.0415 64 MDGS 0.2999 120 GMBA 0.6324 9 FRG 0.0421 65 RWND 0.3024 121 ANGL 0.6366 10 DNMK 0.0441 66 BRMA 0.3038 122 ETHP 0.6745 11 PRTG 0.0505 67 LAOS 0.3099 123 KNYA 0.6766 12 MLTA 0.0514 68 BLGM 0.3230 124 SAFR 0.6778 13 YMNA 0.0550 69 NZLD 0.3404 125 CHAD 0.6782 14 KORN 0.0564 70 GTML 0.3416 126 ZAIR 0.6964 15 GDR 0.0736 71 USA 0.3425 127 LBRA 0.6969 16 ICLD 0.0745 72 SRLK 0.3426 128 IVCT 0.7086 17 SMLA 0.0772 73 ALBN 0.3456 129 UGND 0.7184 18 PLND 0.0791 74 BOLV 0.3489 130 NGRA 0.7270 19 KORS 0.0819 75 ALGR 0.3494 131 CMRN 0.7568 20 ELSL 0.0830 76 NPAL 0.3495 - 132 TNZN 0.7801 21 CHNA 0.0883 77 SYRA 0.3511 22 HLOV 0.0900 78 PERU 0.3514 23 AUST 0.0921 79 ECDR 0.3649 24 FNLD 0.0991 80 SWTZ 0.3651 25 ITLY 0.1009 81 CYPR 0.3677 26 GRCE 0.1068 82 UK 0.3684 27 LXBG 0.1159 83 LBNN 0.3696 28 BRBD 0.1189 84 SNGP 0.3722 29 FRG? 0.1200 85 TLND 0.3799 30 RMNA 0.1260 86 AFGN 0.4026 31 KWAT 0.1406 87 BTSN 0.4246 32 CLMB 0.1431 88 MRCO 0.4257 33 VNZL 0.1435 89 PKST 0.4368 34 CUBA 0.1465 90 PPNG 0.4601 35 EGPT 0.1495 91 TRNT 0.4665 36 KMPC 0.1584 92 YGSL 0.4670 37 CHLE 0.1600 93 CZCH 0.4694 38 HNGR 0.1694 94 ZIMB 0.5059 39 HATI 0.1729 95 PHLP 0.5108 40 LBYA 0.1888 96 IRAN 0.5139 41 PNHA 0.1900 97 INDS 0.5149 42 NCRG 0.1956 98 HRTS 0.53S3 43 JRDN 0.1974 99 MZBQ 0.5426 44 ISRL 0.2002 100 CNDA 0.5454 45 ARGN 0.2010 101 BNIN 0.5521 46 SPAN 0.2180 102 GBON 0.5540 47 TRKY 0.2184 103 NGER 0.5564 48 NTHL 0.2226 104 CAFR 0.5589 49 IRAQ 0.2289 105 SNGL 0.5590 50 TWAN 0.2314 106 INDA 0.5591 51 URGY 0.2334 107 UPVL 0.5615 52 BLGR 0.2337 108 CNGO 0.5641 53 CRCA 0.2419 109 GYNA 0.5677 54 TNSA 0.2527 110 TOGO 0.5780 55 MXCO 0.2656 111 MLWI 0.5808 56 BRNO 0.2657 112 USSR 0.5869 278 OBS NAME ECO 1 KMPC 90 57 CNGO 510 113 SDAR 4010 2 LAOS 90 58 GYNA 510 114 NZLD 4280 3 MALI 90 . 59 IVCT 540 115 JPAN 4450 4 RWND 100 60 ZIMB 550 116 AUST 4870 5 ETHP 100 61 KORS 560 117 FNLD 5420 6 SMLA 110 62 GTML 570 118 LBYA 5530 7 MLDV 110 63 PRGY 580 119 AUSL 5700 8 BRND 110 64 CLMB 580 120 NTHL 5750 9 BRMA 110 65 ECDR 590 121 ICLD 5930 10 NPAL 110 66 GHNA 590 122 FRNC 5950 11 UPVL 110 67 MRTS 610 123 LXBG 6020 12 CHAD 120 68 NCRG 700 124 BLGM 6270 13 BNIN 130 69 DMNR 720 125 FRG 6670 14 NGER 130 70 SYRA 720 126 NRWY 6760 15 MLWI 130 71 TNSA 730 127 DNMK 6810 16 GNEA 130 72 PERU 760 128 CNDA 6930 17 INDA 140 73 MLYS 760 129 USA 7120 18 ZAIR 140 74 CUBA 800 130 SWDN 8150 19 AFGN 150 75 MNGL 860 131 SWTZ 8410 20 LSTO 160 76 ALGR 870 132 KWAT 15190 21 PKST 160 77 TRKY 900 22 TNZN 170 78 TWAN 930 23 MZBQ 180 79 CRCA 960 24 GMBA 180 80 CHLE 990 25 HATI 190 81 BRZL 1030 26 SRLK 190 82 MXCO 1050 27 YMNS 200 83 LBNN 1070 28 MDGS 200 84 JKCA 1110 29 SRLE 200 85 RMNA 1240 30 INDS 220 86 CYPR 1240 31 CAFR 220 87 IRAQ 1250 32 KNYA 220 88 SAFR 1270 33 UGND 230 89 PNMA 1290 34 YMNA 250 90 URGY 1300 35 TOGO 250 91 MLTA 1390 36 EGPT 260 92 BRBD 1410 37 SDAN 270 93 ARGN 1550 38 CMRN 280 94 YGSL 1550 39 MRTN 320 95 PRTG 1570 40 NGRA 340 96 IRAN 1660 41 TLND 350 97 TRNT 2000 42 BTSN 350 98 BLGR 2110 43 HNDS 360 99 HNGR 2150 44 BOLV 360 100 VNZL 2280 45 SNGL 360 101 GRCE 2340 46 ANGL 370 102 IRLD 2390 47 CHNA 380 103 SNGP 2450 48 PHLP 380 104 GBON 2540 49 LBRA 410 105 USSR 2550 50 ZMBA 420 106 PLND 2600 51 KORN 450 107 SPAN 2750 52 ELSL 460 108 ITLY 2810 53 JRDN 460 109 CZCH 3610 54 MRCO 470 110 UK 3780 55 PPNG 470 111 ISRL 3790 56 ALBN 510 112 GDR 3910 OBS NAME DISQ 1 KMPC 57 ZMBA 0.000 113 YGSL 0.260 2 LAOS 58 MRCO 0.000 114 BRZL 0.280 3 MALI 59 KORS 0.000 115 PNMA 0.280 4 RWND 60 GHNA 0.000 116 VNZL 0.300 5 ETHP 61 TNSA 0.000 117 NCRG 0.350 6 SMLA 62 CUBA 0.000 118 SRLK 0.380 7 MLDV 63 MXCO 0.000 119 TWAN 0.430 8 BRND 64 URGY 0.000 120 ISRL 0.430 9 NPAL 65 ARGN 0.000 121 CLMB 0.500 10 UPVL 66 PRTG 0.000 122 UGND 0.760 11 CHAD 67 IRAN 0.000 123 BOLV 0.900 12 BNIN 68 HNGR 0.000 124 PERU 0.920 13 NGER 69 GRCE 0.000 125 MLYS 0.990 14 MLWI 70 IRLD 0.000 126 PRGY 1.000 15 GNEA 71 SNGP 0.000 127 JMCA 1.155 16 AFGN 72 PLND 0.000 128 DMNR 1.215 17 LSTO 73 SPAN 0.000 129 GTML 1.400 18 MZBQ 74 ITLY 0.000 130 ECDR 1.830 19 GMBA 75 GDR 0.000 131 ZIMB 2.375 20 HATI 76 JPAN 0.000 132 SAFR 2.885 21 YMNS 77 FNLD 0.000 22 SRLE 78 AUSL 0.000 23 CAFR 79 BLGM 0.000 24 YMNA 80 NRWY 0.000 25 TOGO 81 DNMK 0.000 26 MRTN 82 SWDN 0.000 27 BTSN 83 JRDN 0.035 28 HNDS 84 ALGR 0.035 29 SNGL 85 IRAQ 0.040 30 ANGL 86 BLGR 0.040 31 CHNA 87 INDS 0.045 32 LBRA 88 TRKY 0.045 33 KORN 89 SYRA 0.050 34 PPNG 90 RMNA 0.060- 35 ALBN 91 EGPT 0.070 36 CNGO 92 PKST 0.075 37 GYNA 93 CMRN 0.075 38 IVCT 94 UK 0.080 39 MRTS 95 PHLP 0.090 40 MNGL 96 AUST 0.105 41 CYPR 97 TLND 0.120 42 MLTA 98 USA 0.120 43 BRBD 99 CRCA 0.125 44 TRNT 100 CHLE 0.125 45 GBON 101 CNDA 0.135 46 SDAR 102 FRG 0.140 47 LBYA 103 SWTZ 0.140 48 ICLD 104 NZLD 0.160 49 LXBG 105 SDAN 0.170 50 KWAT 106 ELSL 0.200 51 BRMA () 107 CZCH 0.205 52 INDA () 108 NTHL 0.205 53 ZAIR (1 109 KNYA 0.215 54 TNZN (1 110 LBNN 0.235 55 MDGS () Ul FRNC 0.245 56 NGRA () 112 USSR 0.245 280 OBS NAME COER 1 BRBD .0000000 57 EGPT 0.006490 113 PNMA 0.03372 2 ICLD .0000000 58 GNEA 0.006862 114 PLND 0.03613 3 LXBG .0000000 59 PRTG 0.007124 115 CNGO 0.03790 4 RMNA .0000741 60 GDR 0.007150 116 JRDN 0.04082 5 ALBN .0004417 61 PKST 0.007154 117 NCRG 0.04176 6 CHNA .0005945 62 ALGR 0.007162 118 KWAT 0.04377 7 JPAN .0006707 63 ZAIR 0.007184 119 FRNC 0.04615 8 CMRN .0006852 64 TRNT 0.007303 120 DNMK 0.04786 9 KORN .0008634 65 CYPR 0.007510 121 ZIMB 0.05444 10 BLGR .0008888 66 SRLK 0.007563 122 BLGM 0.05732 11 FNLD .0009942 67 TNSA 0.007563 123 LBNN 0.06089 12 INDA .0010029 68 PERU 0.007670 124 YMNA 0.06275 13 MXCO .0010330 69 BRND 0.007731 125 ISRL 0.07488 14 SWTZ .0011243 70 CRCA 0.008599 126 ARGN 0.07527 15 URGY .0012249 71 IRLD 0.008710 127 SPAN 0.09796 16 RWND .0013581 72 SDAN 0.008754 128 GYNA 0.09800 17 PPNG .0016563 73 SNGL 0.008878 129 USSR 0.14374 18 MDGS .0017296 74 SMLA 0.009634 130 UK 0.27319 19 NZLD .0017715 75 TOGO 0.009647 131 ITLY 0.58471 20 AUST .0022223 76 MRTS 0.009662 132 GRCE 0.75219 21 ETHP .0022907 77 HNDS 0.009745 22 BRZL .0023606 78 ELSL 0.009905 23 PRGY .0024519 79 IRAN 0.010000 24 TLND .0025426 80 JMCA 0.010045 25 UPVL .0025969 81 KORS 0.010277 26 AUSL .0027051 82 YMNS 0.010350 27 MALI .0027701 83 KNYA 0.01052S 28 AFGN .0028232 84 UGND 0.010651 29 INDS .0028503 85 VNZL 0.010926 30 SDAR .0029672 86 BTSN 0.011281 31 BRMA .0031305 87 CUBA 0.011608 32 MZBQ .0033948 88 GTML 0.012008 33 NRWY .0034430 89 CAFR 0.012331 34 NPAL .0034662 90 ECDR 0.012523 35 HNGR .0036199 91 GHNA 0.013012 36 MLTA .0036816 92 MLYS 0.013571 37 CNDA .0038699 93 MRTN 0.013675 38 NGRA .0040772 94 BNIN 0.014048 39 TWAN .0041950 95 LSTO 0.014279 40 NGER .0042241 96 YGSL 0.014900 41 GMBA .0043104 97 KMPC 0.015740 42 IVCT .0043987 98 SRLE 0.016132 43 CLMB .0045649 99 LBRA 0.016282 44 PHLP .0049870 100 IRAQ 0.016326 45 CZCH .0051394 101 ZMBA 0.017998 46 NTHL .00S3808 102 MLDV 0.018019 47 USA .0055974 103 FRG 0.018471 48 SWDN .0056585 104 SYRA 0.020910 49 ANGL .0058032 105 LAOS 0.021046 50 TRKY .0058581 106 SAFR 0.022626 51 TNZN .0060092 107 SNGP 0.023811 52 CHAD .0060258 108 LBYA 0.026478 53 MRCO .0061295 109 GBON 0.027615 54 CHLE .0061709 110 BOLV 0.029071 55 MNGL .0063898 111 DMNR 0.029718 56 MLWI .0064015 112 HATI 0.032066 281 OBS NAME ELT 1 BRBD 57 AUST 0.00 113 MDGS 0.60 2 ICLD 58 BRZL 0.00 114 ALGR 0.60 3 LXBG 59 PRGY 0.00 115 INDA 0.72 4 ALBN 60 AUSL 0.00 116 IRAQ 0.72 5 CHNA 61 NRWY 0.00 117 KNYA 0.86 6 KORN 62 HNGR 0.00 118 MLYS 0.92 7 RWND 63 TWAN 0.00 119 USSR 0.94 8 PPNG 64 CLMB 0.00 120 SAFR 0.99 9 ETHP 65 NTHL 0.00 121 SDAN 1.04 10 UPVL 66 USA 0.00 122 SPAN 1.04 11 MALI 67 SWDN 0.00. 123 IRAN 1.08 12 AFGN 68 CHLE 0.00 124 GHNA 1.17 13 SDAR 69 EGPT 0.00 125 BRMA 1.20 14 MZBQ 70 GDR 0.00 126 INDS 1.44 15 NPAL 71 TNSA 0.00 127 YGSL 1.48 16 MLTA 72 PERU 0.00 128 JRDN 1.60 17 NGER 73 CRCA 0.00 129 LBNN 1.80 18 GMBA 74 IRLD 0.00 130 UGND 2.00 19 IVCT 75 ELSL 0.00 131 ZAIR 2.24 20 ANGL 76 JMCA 0.00 132 NGRA 3.76 21 CHAD 77 KORS 0.00 22 MNGL 78 VNZL 0.00 23 MLWI 79 CUBA 0.00 24 GNEA 80 GTML 0.00 25 TRNT 81 ECDR 0.00 26 CYPR 82 FRG 0.00 27 BRNO 83 SNGP 0.00 28 SNGL i 84 BOLV 0.00 29 SMLA 85 DMNR 0.00 30 TOGO 86 PNMA 0.00 31 MRTS 87 PLND 0.00 32 HNDS 88 NCRG 0.00 33 YMNS 89 DNMK 0.00 34 BTSN 90 ISRL 0.00 35 CAFR 91 ARGN 0.00 36 MRTN 92 ITLY 0.00 37 BNIN 93 GRCE 0.00 38 LSTO 94 TNZN 0.03 39 KMPC 95 FRNC 0.06 40 SRLE 96 PRTG 0.07 41 LBRA 97 BLGR 0.08 42 MLDV 98 TLND 0.09 43 LAOS 99 SYRA 0.12 44 LBYA 100 PKST 0.15 45 GBON 101 PHLP 0.16 46 HATI 102 TRKY 0.18 47 CNGO 103 SRLK 0.20 48 KWAT 104 ZMBA 0.20 49 YMNA 105 CMRN 0.20 50 GYNA 106 RMNA 0.27 51 JPAN 0 107 CZCH 0.32 52 FNLD 0 108 BLGM 0.37 53 MXCO 0 109 ZIMB 0.51 54 SWTZ 0 110 UK 0.51 55 URGY 0 111 MRCO 0.52 56 NZLD 0 112 CNDA 0.56 282 OBS NAME GEO 1 BRBD 0 57 VNZL 0 113 MDGS 2 ICLD 0 58 CUBA 0 114 ALGR 3 LXBG 0 59 SNGP 0 115 INDA 4 ALBN 0 60 DMNR 0 116 IRAQ 5 KORN 0 61 PNMA 0 117 KNYA 6 RWND 0 62 PLND 0 118 MLYS 7 UPVL 0 63 NCRG 0 119 USSR 8 MALI 0 64 ARGN 0 120 SAFR 9 SDAR 0 65 GRCE 0 121 SDAN 10 MZBQ 0 66 CHNA 1 122 SPAN 11 MLTA 0 67 PPNG 1 123 IRAN 12 NGER 0 68 ETHP 1 124 GHNA 13 GMBA 0 69 AFGN 1 125 BRMA 14 IVCT 0 70 NPAL 1 126 INDS 15 MNGL 0 71 ANGL 1 127 YGSL 16 MLWI 0 72 CHAD 1 128 JRDN 17 GNEA 0 73 CYPR 1 129 LBNN 18 TRNT 0 74 KMPC 1 130 UGND 19 BRND 0 75 SRLE 1 131 ZAIR 20 SNGL 0 76 LAOS 1 132 NGRA 21 SMLA 0 77 CNGO .22 TOGO 0 78 FNLD 23 MRTS 0 79 MXCO 24 HNDS 0 80 SWTZ 25 YMNS 0 81 AUST 26 BTSN 0 82 NRWY 27 CAFR 0 83 NTHL 28 MRTN 0 84 USA 29 BNIN 0 85 GDR 30 LSTO 0 86 PERU 31 LBRA 0 87 GTML 32 MLDV 0 88 ECDR 33 LBYA 0 89 FRG 34 GBON 0 90 BOLV 35 HATI 0 91 DNMK 36 KWAT 0 92 ISRL 37 YMNA 0 93 ITLY 38 GYNA 0 94 TNZN 39 JPAN 0 95 FRNC 40 URGY 0 96 PRTG 41 NZLD 0 97 BLGR 42 BRZL 0 98 TLND 43 PRGY 0 99 SYRA 3 44 AUSL 0 100 PKST 3 45 HNGR 0 101 PHLP : 46 TWAN 0 102 TRKY ] 47 CLMB 0 103 SRLK 3 48 SWDN 0 104 ZMBA 3 49 CHLE 0 105 CMRN : 50 EGPT 0 106 RMNA 3 51 TNSA 0 107 CZCH 3 52 CRCA 0 108 BLGM 3 53 IRLD 0 109 ZIMB 3 54 ELSL 0 110 UK 3 55 JMCA 0 111 MRCO 3 56 KORS 0 112 CNDA 3 283 OBS NAME EXT 1 ICLD 0 57 MALI 2 113 I.BYA 4 2 LXBG 0 58 NGER 2 114 HATI 4 3 MLTA 0 59. GMBA 2 115 KWAT 4 4 HNDS 0 60 IVCT 2 116 YMNA 4 5 MLDV 0 61 GNEA 2 117 TWAN 4 6 JPAN 0 62 SNGL 2 118 EGPT 4 7 URGY 0 63 CAFR 2 119 TNSA 4 8 BRZL 0 64 LBRA 2 120 JMCA 4 9 PRGY 0 65 GBON 2 121 KORS 4 10 HNGR 0 66 ETHP 2 122 GRCE 4 11 CLMB 0 67 SRLE 2 123 CHNA 4 12 SWDN 0 68 PERU 2 124 LAOS 4 13 CHLE 0 69 ECDR 2 125 GDR 4 14 CRCA 0 70 BOLV 2 126 FRG 4 15 ELSL 0 71 TNZN 2 127 ITLY 4 16 VNZL 0 72 PKST 2 128 FRNC 4 17 CUBA 0 73 PHLP 2 129 SYRA 4 18 DMNR 0 74 CMRN 2 130 ALGR 4 19 PLND 0 75 KNYA 2 131 JRDN 4 20 NCRG 0 76 SAFR 2 132 LBNN 4 21 FNLD 0 77 IRAN 2 22 AUST 0 78 INDS 2 23 NRWY 0 79 YGSL 2 24 NTHL 0 80 NGRA 2 25 DNMK 0 81 MZBQ 3 26 PRTG 0 82 MLWI 3 27 MDGS 0 83 TRNT 3 28 ALBN 84 TOGO 3 29 MNGL 85 MRTS 3 30 BRND 86 YMNS 3 31 SMLA 87 MRTN 3 32 BTSN 88 BNIN 3 33 LSTO 89 GYNA 3 34 NZLD 90 SNGP 3 35 AUSL 91 ANGL 3 36 IRLD 92 CHAD 3 37 PNMA 93 CYPR 3 38 ARGN 94 CNGO 3 39 PPNG 95 SWTZ 3 40 AFGN 96 TRKY 3 41 NPAL 97 ZMBA 3 42 KMPC 98 MRCO 3 43 MXCO 99 CNDA 3 44 USA 100 INDA 3 45 GTML 101 IRAQ 3 46 ISRL 102 MLYS 3 47 BLGR 103 SDAN 3 48 TLND 104 SPAN 3 49 SRLK 105 GHNA 3 50 RMNA 106 UGND 3 51 CZCH 107 ZAIR 3 52 BLGM 108 BRBD 4 53 ZIMB 109 KORN 4 54 UK HP RWND 4 55 USSR 111 UPVL 4 56 BRMA 112 SDAR 4 284 NAME TCLD .0000000 57 KWAT .0013158 113 BOLV 0.01046 LXBG .0000000 58 MRCO .0013222 114 ZIMB 0.01168 FNLD .0000000 59 BRMA .0014055 115 JRDN 0.01360 ALBN .0000000 60 MDGS .0015868 116 DMNR 0.01451 BTSN .0000000 61 SWDN .0018055 117 FRNC 0.01484 RMNA .0000000 62 SRLK .0018379 118 CNGO 0.01763 NGER .0000000 63 SNGP .0019277 119 GBON 0.01800 GMBA .0000000 64 HNDS .0019585 120 LBNN 0.02228 BRBD .0000000 65 KMPC .0019830 121 NCRG 0.02386 KORN .0000000 66 ELSL .0019909 122 ISRL 0.03144 RWND .0000000 67 NPAL .0020477 123 SPAN 0.03176 GDR .0000616 68 CLMB .0020838 124 ARGN 0.03350 HNGR .0000697 69 TRNT .0020942 125 YMNA 0.03552 AUST .0000967 70 TRKY .0021004 126 PNMA 0.03704 BLGR .0000988 71 SMLA .0021513 127 BLGM 0.04130 PRGY .0001067 72 GHNA .0024339 128 MLDV 0.04545 CUBA .0001168 73 GTML .0024538 129 GYNA 0.05924 TNZN .0001507 74 ANGL .0024691 130 UK 0.13439 CMRN .0001714 75 MLWI .0025229 131 GRCE 0.16746 CHNA .0001762 76 TOGO .0025510 132 ITLY 0.47546 UPVL .0001857 77 PRTG .0026359 MALI .0001981 78 LAOS .0027009 NRWY .0002029 79 NGRA .0027600 IVCT .0002320 80 MLTA .0027663 SWTZ .0002368 81 VNZL .0029359 GNEA .0002550 82 SDAN .0030583 TWAN .0003563 83 CHLE .0032003 PPNG .0004144 84 CYPR .0034158 BRZL .0004412 85 MRTN .0034423 TLND .0004476 86 PERU .0036232 HATI .0004723 87 USA .0038071 URGY .0004867 88 PKST .0039207 SOAR .0005168 89 FRG .0039278 INDS .0005273 90 KORS .0040172 JPAN .0005463 91 TNSA .0044773 CZCH .0005601 92 KNYA .0045345 AUSL .0006373 93 SRLE .0045386 ALGR .0006978 94 DNMK .0049020 NZLD .0007092 95 CHAD .0049451 MZBQ .0007287 96 LBYA .0051600 IRAQ .0007482 97 BNIN .0052122 MNGL .0003013 98 SYRA .0052825 INDA .0008230 99 ECDR .C054V1V AFGN .0008835 100 SNGL .0056051 NTHL .0009338 101 CRCA .0057571 ETHP .0009656 102 PLND .0059084 IRLD .0009975 103 MRTS .0060680 EGPT .0010201 104 IRAN .0062767 CNDA .0010745 105 MLYS .0064017 MXCO .0010932 106 UGND .0075464 PHLP .0011701 107 USSR .0078159 BRND .0011940 108 SAFR .0078605 YMNS .0012138 109 LSTO .0086290 YGSL .0012510 110 ZMBA .0093132 ZAIR .0012940 111 JMCA .0095643 LBRA .0013132 112 CAFR .0099256 285 OBS NAME ATT 1 ICLD .0000000 57 BTSN 0.003241 113 DMNR 0.04950 2 LXBG .0000000 58 CNDA 0.003317 114 GTML 0.05398 3 FNLD .0000000 59 MRTN 0.003442 115 ZIMB 0.05972 4 BRBD .0000000 60 NTHL 0.003528 116 JRDN 0.06491 5 MNGL .0000000 61 USA 0.003597 117 SPAN 0.06954 6 LBRA .0000000 62 PPNG 0.003730 118 CYPR 0.07008 7 RMNA .0000041 63 GMBA 0.004320 119 BLGM 0.07670 8 ALBN .0000491 64 RWND 0.004349 120 ELSL 0.07708 9 HNGR .0000697 65 SMLA 0.004661 121 ISRL 0.08790 10 BLGR .0000988 66 TRKY 0.004768 122 NCRG 0.09746 11 GDR .0001232 67 DNMK 0.004902 123 ARGN 0.12521 12 KORN .0001440 68 BRND 0.005075 124 LAOS 0.14450 13 SNGL .0002548 69 ALGR 0.005234 125 KMPC 0.17054 14 GNEA .0002550 70 TLND 0.005259 126 GYNA 0.21298 15 NZLD .0003546 71 ECDR 0.005638 127 YMNA 0.30014 16 MALI .0003963 72 SAFR 0.005860 128 GRCE 0.33491 17 JPAN .0004121 73 BRMA 0.006163 129 FRNC 0.34849 18 IVCT .0004640 74 USSR 0.006730 130 LBNN 0.87768 19 CMRN .0005141 75 MLYS 0.007262 131 ITLY 1.62270 20 CZCH .0005414 76 MLWI 0.007339 132 UK 4.19939 21 SWTZ .0005625 77 UGND 0.007342 22 NRWY .0006086 78 KNYA 0.007646 23 URGY .0006128 79 SDAN 0.007773 24 CHNA .0006219 80 IRLD 0.007827 25 TNSA .0007787 81 PKST 0.008966 26 EGPT .0007801 82 MLDV 0.009091 27 AUST .0008706 83 CUBA 0.009107 28 BRZL .0009138 84 KWAT 0.009211 29 MLTA .0009221 85 ETHP 0.009817 30 TWAN .0009263 86 IRAN 0.009873 31 UPVI. .0009287 87 PERU 0.010115 32 PRGY .0009606 88 LBYA 0.010320 33 TOGO .0010204 89 CRCA 0.010363 34 SRLE .0011346 90 PHLP 0.010478 35 YGSL .0011373 91 MRTS 0.010922 36 MDGS .0011541 92 NGRA 0.011294 37 CAFR .0012407 93 LSTO 0.011505 38 MRCO .0012561 94 PNMA 0.013032 39 SWDN .0012897 95 HNDS 0.013318 40 NPAL .0014245 96 SYRA 0.013767 41 TNZN .0015068 97 GBON 0.014000 42 SDAR .0015504 98 AFGN 0.014195 43 AUSL .0016730 99 JMCA 0.015940 44 KORS .0017535 100 MZBQ 0.017974 45 INDS .0017913 101 HATI 0.019126 46 NGER .0019920 102 CHAD 0.019231 47 PRTG .0019934 103 ZAIR 0.020473 48 MXCO .0022658 104 CLMB 0.020612 49 GHNA .0023180 105 CNGO 0.022670 50 CHLE .0024002 106 ZMBA 0.024447 51 SNGP .0024096 107 IRAQ 0.024690 52 SRLK .0025571 108 YMNS 0.025837 S3 PLND .0028199 109 BOLV 0.029707 54 INDA .0028943 110 VNZL 0.032200 55 BNIN .0029784 111 FRG 0.033826 56 TRNT .0031414 112 ANGL 0.044974 286 OBS NAME ASS 1 ICLD .000000000 57 SAFR .0000465 113 SMLA 0.00108 2 LXBG .000000000 58 CNDA .0000467 114 SYRA 0.00112 3 FNLD .000000000 59 INDA .0000571 115 IRAN 0.00116 4 BRBD .000000000 60 MXCO .0000596 116 FRNC 0.00128 5 MNGL .000000000 61 MRCO .0000661 117 KWAT 0.00132 6 LBRA .000000000 62 PERU .0000755 118 YMNS 0.00156 7 RMNA .000000000 63 AUSL .0000797 119 PNMA 0.00206 8 ALBN .000000000 64 SRLK .0000799 120 SPAN 0.00209 9 HNGR .000000000 65 TLND .0000839 121 LAOS 0.00236 10 BLGR .000000000 66 EGPT .0000900 122 NCRG 0.00254 11 GDR .000000000 67 AUST .0000967 123 GYNA 0.00423 12 KORN .000000000 68 NTHL .0001033 124 ELSL 0.00569 13 NZLD .000000000 69 URGY .0001081 125 GTML 0.00661 14 MALI .000000000 70 YGSL .0001137 126 CNGO 0.00672 15 JPAN .000000000 71 NGRA .0001271 127 YMNA 0.00696 16 IVCT .000000000 72 SDAR .0001292 128 GRCE 0.00790 17 CMRN .000000000 73 KMPC .0001416 129 LBNN 0.01053 18 CZCH .000000000 74 MDGS .0001443 130 UK 0.01794 19 SWTZ .000000000 75 ECDR .0001658 131 ARGN 0.02267 20 NRWY .000000000 76 MLYS .0001911 132 ITLY 0.14417 21 CHNA .000000000 77 TNSA .0001947 22 MLTA .000000000 78 ALGR .0002094 23 TWAN .000000000 79 PRGY .0002135 24 UPVL .000000000 80 USSR .0002171 25 SRLE .000000000 81 CLMB .0002265 26 CAFR .000000000 82 MLWI .0002294 27 NPAL .000000000 83 ZMBA .0002328 28 INDS .000000000 84 CUBA .0002335 29 GHNA .000000000 85 FRG .0002345 30 SNGP .000000000 86 AFGN .0002356 31 PLND .000000000 87 HATI .0002361 32 TRNT .000000000 88 CHLE .0002527 33 BTSN .000000000 89 SWDN .0002579 34 MRTN .000000000 90 PHLP .0002925 35 PPNG .000000000 91 TNZN .0003014 36 GMBA .000000000 92 UGND .0003059 37 RWND .000000000 93 ISRL .0003381 38 DNMK .000000000 94 KNYA .0003557 39 SDAN .000000000 95 BNIN .0003723 40 MLDV .000000000 96 IRLD .0003837 41 LBYA .000000000 97 PKST .0004136 42 CRCA .000000000 98 TRKY .0004825 43 MRTS .000000000 99 SNGL .0005096 44 HNDS .000000000 100 GNEA .0005101 45 GBON .000000000 101 ETHP .0005230 46 MZBQ .000000000 102 JMCA .0005313 47 CHAD .000000000 103 ZIMB .0005652 48 ANGL .000000000 104 VNZL .0005682 49 JRDN .000000000 105 BRND .0005970 50 BLGM .000000000 106 IRAQ .0006413 51 PRTG .000016474 107 NGER .0007470 52 KORS .000031883 108 CYPR .0008245 53 BRMA .000036039 109 DMNR .0009210 54 BRZL .000042015 110 LSTO .0009588 55 USA .000043928 111 TOGO .0010204 56 ZAIR .000046215 112 BOLV .0010460 287 OBS NAME DEATH 1 ICLD 0 57 FRNC 55 113 DMNR 4269 2 LXBG 0 58 PNMA 56 114 INDA 4280 3 FNLD 0 59 URGY 61 115 SDAN 4411 4 BRBD 0 60 NPAL 63 116 ARGN 4586 5 MNGL 0 61 GHNA 65 117 MZBQ 4631 6 RMNA 0 62 GNEA 66 118 IRAQ 5231 7 NZLD 0 63 BLGM 74 119 SRLK 5246 8 SWTZ 0 64 ECDR 87 120 YMNS 7381 9 UPVL 0 65 MLWI 90 121 ANGL 9673 10 MRTN 0 66 LBYA 99 122 ZAIR 9978 11 GMBA 0 67 MRCO 105 123 LAOS 22457 12 DNMK 0 68 USSR 105 124 RWND 26333 13 MLDV 0 69 SDAR 113 125 UGND 30259 14 CRCA 0 70 BRZL 120 126 ETHP 33623 15 AUSL 0 71 ELSL 122 127 KMPC 56575 16 HNGR 1 72 EGPT 153 128 LBNN 60721 17 IVCT 1 73 ITLY 161 129 BRND 81979 18 NRWY 1 74 SPAN 172 130 PKST 310980 19 MLTA 1 75 MDGS 183 131 INDS 575938 20 BTSN 1 76 ALGR 194 132 NGRA 1995235 21 TOGO 1 77 KORS 200 22 KWAT 1 78 ISRL 216 23 BLGR 2 79 VNZL 218 24 TRNT 2 80 NCRG 225 25 MALI 3 81 TRKY 229 26 SNGP 3 82 GTML 247 27 ALBN 5 83 GYNA 337 28 SWDN 6 84 KNYA 363 29 PPNG 7 85 YGSL 379 30 AUST 7 86 USA 392 31 CMRN 9 87 MXCO 440 32 CNDA 10 88 MLYS 476 33 CAFR 11 89 TWAN 478 34 GDR 12 90 CNGO 537 35 KORN 13 91 HATI 568 36 SNGL 15 92 PERU 737 37 NTHL 16 93 CUBA 745 38 TNSA 16 94 BOLV 768 39 PLND 23 95 SAFR 795 40 LSTO 24 96 JMCA 916 41 SRLE 27 97 CLMB 1062 42 CZCH 28 98 CHLE 1207 43 MRTS 28 99 ZMBA 1295 44 SMLA 30 100 CYPR 1365 45 JPAN 32 101 TLND 1433 46 LBRA 35 102 UK 1455 47 PRGY 35 103 YMNA 1492 48 TNZN 38 104 SYRA 1804 49 NGER 38 105 BRMA 1837 50 GBON 39 106 JRDN 2125 51 GRCE 42 107 CHAD 2264 52 PRTG 43 108 IRAN 2957 53 BNIN 46 109 CHNA 3481 54 HNDS 53 110 AFGN 3549 55 IRLD 53 111 ZIMB 3853 56 FRG 55 112 PHLP 4193