¹ The work of Karl Deutsch represents some of the major explorations in the theory of political integration, and his thinking is well represented in his contributions to the following works: International Political Communities—An Anthology (New York: Doubleday Anchor Books, 1966)Google Scholar, and Jacob, Philip and Toscano, James V., eds., The Integration of Political Communities (New York: J. B. Lippincott Co., 1964)Google Scholar. Both of these works include other useful material on the subject. Also useful are: Etzioni, Amitai, Political Unification: A Comparative Study of Leaders and Forces (New York: Holt, Rinehart and Winston, Inc., 1965)Google Scholar; Galtung, Johan, “A Structural Theory of Integration,” Journal of Peace Research, 4 (1968), 375–395CrossRefGoogle Scholar, and Ake, Claude, A Theory of Political Integration (Homewood, Illinois: Dorsey Press, 1967)Google Scholar, which is one of the few recent works specifically concerned with national integration. Also see Binder, Leonard, “National Integration and Political Development,” American Political Science Review, 28 (09, 1964), 622–631CrossRefGoogle Scholar; Weiner, Myron, “Political Integration and Political Development,” The Annals, 358 (03, 1965), 52–64Google Scholar; Goldman, Ralph M., “A Transactional Theory of Political Integration and Arms Control,” American Political Science Review, 63 (09, 1969), 719–733CrossRefGoogle Scholar; and Lijphart, Arend, “Cultural Diversity and Theories of Political Integration,” Canadian Journal of Political Science, 12 (03 1971), 1–14CrossRefGoogle Scholar.

Representative of some of the recent work on political instability are: Bwy, D. P., “Political Instability in Latin America: The Cross-Cultural Test of a Causal Model,” Latin American Research Review, III (1968), 17–66Google Scholar; Davies, James C., “Political Stability and Instability: Some Manifestations and Causes,” Journal of Conflict Resolution, 13 (03, 1969), 1–17CrossRefGoogle Scholar; Duff, Ernest A. and McCamant, John F., “Measuring Social and Political Requirements for System Stability in Latin America,” American Political Science Review, 62 (12, 1968), 1125–1143CrossRefGoogle Scholar; Ivo, K. and Feierabend, Rosalind L., “Aggressive Behavior Within Politics, 1948–1962: A Cross-National Study,” Journal of Conflict Resolution, 10 (09 1966), 249–271Google Scholar; Fossum, Egil, “Factors Influencing the Occurrence of Military Coups d'Etat in Latin America,” Journal of Peace Research, 3 (1967), 209–227Google Scholar; Gurr, Ted, “A Causal Model of Civil Strife: A Comparative Analysis Using New Indices,” American Political Science Review, 63 (12, 1968), 1104–1124CrossRefGoogle Scholar, and Why Men Rebel (Princeton, New Jersey: Princeton University Press, 1970)Google Scholar; Midlarsky, Manus and Tanter, Raymond, “Toward a Theory of Political Instability in Latin America,” Journal of Peace Research, 3 (1967), 209–227CrossRefGoogle Scholar; Nilson, Sten Sparre, “Measurement and Models in the Study of Stability,” World Politics, 11 (10, 1967), 83–110Google Scholar; Tsurutani, Taketsugu, “Stability and Instability: A Note in Comparative Political Analysis,” The Journal of Politics, 30 (11, 1968), 910–933CrossRefGoogle Scholar; and Zolberg, Aristide, 1968, “The Structure of Political Conflict in the New States of Tropical Africa,” American Political Science Review, 62 (03, 1968), 70–87CrossRefGoogle Scholar. The most recent evidence of the interest in the study of instability is in the work generated by the National Commission on the Causes and Prevention of Violence. See especially the comprehensive historical and comparative treatment in The History of Violence in America, ed. Graham, Hugh Davis and Gurr, Ted Robert (New York: Bantam Books, 1969)Google Scholar.

² Because the statistical generalizations on which the analysis is based make the discussion very anonymous, it will be useful to bear in mind that the countries for which data have been gathered are: Botswana, Burundi, Cameroon, Central African Republic, Chad, Congo (Brazzaville), Zaire, Dahomey, Ethiopia, Gabon, Gambia, Ghana, Guinea, Ivory Coast, Kenya, Lesotho, Liberia, Malawi, Mauritania, Mali, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, Somali Republic, Sudan, Tanzania, Togo, Uganda, Upper Volta, Zambia.

³ The fundamental proposition relating integration to instability is summed up by Mack and Snyder in their review of conflict theory: “The more integrated into the society are the parties to conflict, the less likely will conflict be violent …. Basically, the stability-instability balance will be a resultant of the success or failure of the normative order in regulating conflicts of interest.” Mack, Raymond W. and Snyder, Richard C., “The Analysis of Social Conflict—Toward an Overview and Synthesis,” Journal of Conflict Resolution, 1 (06, 1957), 227Google Scholar.

⁴ See the contrasting definitions of violence in Johnson, Chalmers, Revolutionary Change (Boston: Little, Brown, 1966), p. 8Google Scholar, and Nieburg, H. L., Political Violence (New York: St. Martins Press, 1969), p. 13Google Scholar. It should be noted that structural violence as opposed to interpersonal and physical violence is characteristic of the most stable political systems. See the distinctions made in Galtung, Johan, “Violence, Peace, and Peace Research,” Journal of Peace Research, 3 (1967), 167Google Scholar.

⁵ Quantitative, cross-national research on political development has for the most part ignored African cases, for the very intelligible reason that adequate aggregate data sources for these new nations have not been readily available. The data used in this paper were collected by the African Integration Project at Northwestern University. A complete African data bank is being prepared for circulation on tape, and selections from those data, with discussions of conceptual and methodological issues related to the use of the data, are presented in Morrison, Donald G., Mitchell, Robert C., Paden, John N. and Stevenson, Hugh Michael, Black Africa: A Handbook for Comparative Analysis (New York: The Free Press, 1972)Google Scholar. Sources for the data used in this article are detailed in that work, and most of the indicators used here are presented in rank-ordered tables by nation in the book. Readers interested in additional details relating to the sources of the data in this article are invited to communicate with the authors c/o the Institute for Behavioral Research, York University, Downsview, Ontario.

⁶ See Morrison, Donald G. and Stevenson, Hugh Michael, Conflict and Change in African Politics (New York: The Free Press, forthcoming)Google Scholar.

⁷ “Any attempt to measure the likelihood of revolution should … encompass the social sources of strains and tensions present in the several societies and the social conditions that either encourage or balance extreme acts among the contending forces … the likelihood of revolutions is a function of both positive or conflict-generating factors and negative or conflict-controlling factors.” Feldman, Arnold, “Violence and Volatility: The Likelihood of Revolution,” in Internal War, Problems and Approaches, Eckstein, Harry, ed. (New York: The Free Press, 1964), p. 115Google Scholar.

⁸ Deutsch, Karl W., The Analysis of International Relations (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1968), pp. 332–40Google Scholar.

⁹ Deutsch, Karl W.et al.Political Community and the North Atlantic Area (Princeton: Princeton University Press, 1957), p. 5Google Scholar.

¹⁰ For other useful summaries of dimensions of integration, see Landecker, W. S., “Types of Integration and Their Measurement,” American Journal of Sociology, 56 (01, 1951), 332–40CrossRefGoogle Scholar, and Weiner, “Political Integration and Political Development.”

¹¹ Of course there is the rival hypothesis that increasing interaction increases hostility and conflict unless the interacting parties share similar values relevant to the activity in which they interact. See Coser, Lewis A., The Functions of Social Conflict (London: Routledge and Kegan Paul, 1956), p. 62Google Scholar, and Jacob, Philip and Toscano, James V., eds., The Integration of Political Communities, p. 101Google Scholar.

¹² Deutsch, Karl W., The Nerves of Government (New York: The Free Press, 1963)Google Scholar, and Nationalism and Social Communication (Cambridge, Mass.: MIT Press, 2nd ed., 1966)Google Scholar.

¹³ Deutsch, Karl W., “Transaction Flows as Indicators of Political Cohesion,” in Jacob, and Toscano, Google Scholar.

¹⁴ At the time this paper was written we were able to obtain information on domestic mail flows for only seven of the African countries, internal telephonic communications measures for even fewer, and internal trade patterns for none at all. A particularly troublesome problem, even if the data were available, would be to determine a method of dividing nations into comparable internal units (regions) in order to compare the spatial distribution of transactions within them.

¹⁵ The operational definitions of communications potential were initially given in terms of 76 individual measures which we conceptually grouped into dimensions of private, commercial and urban communications. Factor analysis of these measures showed a stable structure of four factors which we labelled mass communications, personal communications, commercial communications and aviation communications. The content of the first three of these factors was then analyzed, and high-loading variables within each factor were separately analyzed to produce unidimensional factors, the factor scores for which gave three measures of communications potential. The content of these factors is shown in the appendix to this paper. In addition to these three indices of the scale of communications potential, we also use four static and three dynamic measures which were unique variables in the original factor solution: the percentage of the national population living in cities over 20,000, in 1965; the percentage change in the proportion of the national population living in cities over 20,000, 1955–65; the primacy of the largest city, which is the ratio of the population in the largest city to the population in the second largest city; the number of miles of improved road per 1,000 square miles of land area, c. 1965; the number of miles of railroad track per 1,000 square miles of land area, c. 1965; the percentage change in the number of radios in use, 1960–65, and the percentage change in the number of commercial vehicles, 1958–66. Data and sources for these measures are given in Morrison et al., Black Africa, chapters 5 and 15, and Part II. The use of multiple indicators here, and for all the theoretical dimensions of integration outlined in this paper, is a fundamental methodological procedure for controlling for error in nonexperimental data. See Blalock, Hubert M. Jr., “Estimating Measurement Error Using Multiple Indicators and Several Points in Time,” American Sociological Review, 35 (02, 1970), 101–111CrossRefGoogle Scholar, and Blalock, , “Multiple Indicators and the Causal Approach to Measurement Error,” American Journal of Sociology, 75 (09, 1969), 264–273CrossRefGoogle Scholar. Also see “Introduction,“ in Morrison et al., Black Africa.

¹⁶ See Leonard Binder, “National Integration and Political Development,” and Myron Weiner, “Political Integration and Political Development.”

¹⁷ It is also argued that the integration of the elite itself, rather than mass-elite integration, is of primary importance to national integration. See Ake, A Theory of Political Integration.

¹⁸ Shils, Edward, “Centre and Periphery,” in The Logic of Personal Knowledge: Essays Presented to Michael Polanyi (London: Routledge and Regan Paul, 1961)Google Scholar.

¹⁹ The greater, or more significant, the difference between elite and mass populations in terms of ethnic identity and values, the greater the mass-elite gap, and the less integrated the political system. The measure of disproportionality used for this paper is based on a chi-square statistic measuring the difference between the observed ethnic group proportions in the national population, and the proportions of cabinet members belonging to different ethnic groups. The greater the value of the chi-square statistic the greater the disproportionality, and the greater the mass-elite gap. The chi-square value was then used to calculate a C-coefficient, which normalized the chi-square statistic, which is highly sensitive to the number of observations used in its calculation. The measures used in this paper are based on the composition of the cabinet membership at the time of independence and in either 1967 or the year before a coup in those countries where coups have occurred. A third measure is based on the change in the value of the C-coefficients for those two periods of time. The data for ethnic proportionality in national populations and cabinets is given in Morrison et al., Black Africa, Part II.

²⁰ The measures of mass-elite mobility used in this paper relate to the access to advanced education: (a) the number of students enrolled in institutions of higher learning as a percentage of the relevant age group, 1961–62; (b) secondary-school enrollment as a ratio of the relevant age group, 1957; (c) secondary-school enrollment as a ratio of the relevant age group, 1961, and (d) the average annual rate of increase in the secondary-school enrollment ratio, 1957–61. The data for these variables are taken from The Development of Higher Education in Africa (UNESCO: Report of the Conference on the Development of Higher Education in Africa, 1962), and World Survey of Education, Vol. III, Secondary Education (New York: International Documents Service by arrangement with UNESCO, 1961)Google Scholar. For the importance of higher education as a definition of elite status in Africa, see Lloyd, P. C., ed., The New Elites of Tropical Africa (London: Oxford University Press, 1967)Google Scholar.

²¹ See Jacob, Philip E., “The Influence of Values in Political Integration,” in Jacob, and Toscano, , eds., The Integration of Political Communities, pp. 209–246Google Scholar.

²² These values structure identity and loyalty in the political system. As Deutsch argues: “The population of different territories might easily profess verbal attachment to the same set of values without having a sense of political community that leads to political integration. The kind of sense of community that is relevant for integration … turned out to be rather a matter of mutual sympathy and loyalties; of ‘we-feeling,’ trust and mutual consideration; of partial identification in terms of self-images and interests; of mutually successful predictions of behavior and of cooperative action in accordance with it …. Peaceful change could not be assured without this kind of relationship.” Deutsch, Karl W.et al., Political Community in the North Atlantic Area, p. 36Google Scholar.

²³ The importance of congruence in the authority values in different subsystems of society is the basis of the argument in Eckstein, Harry, A Theory of Stable Democracy (Princeton: Princeton University, 1961)Google Scholar.

²⁴ “A society is pluralistic to the extent that it is structurally segmented and culturally diverse.” van den Berghe, Pierre, “Towards a Sociology for Africa,” Social Forces, 43 (10, 1964), 11–18CrossRefGoogle Scholar.

²⁵ Our measures of linguistic pluralism are: a) the number of different languages spoken in each country; b) the percentage of the national population who speak the major vernacular language, and c) a measure of linguistic homogeneity indicating the ratio of speakers of the dominant vernacular to second vernacular speakers, c. 1968. Religious pluralism is measured by: a) the proportion of the population that identifies with Islam, b) the proportion of the population that identifies with Christianity, c) the ratio of the population identifying with the predominant universal religion (Islam or Christianity) to the population identifying with the other major religion, and d) a measure of religious frationalization calculated as

where PR_i is the proportion of the population identified with Christianity, Islam, Independent Churches, and Traditional religions. The data used in the construction of these variables is given and documented in Morrison et al., Black Africa, Chap. 2.

²⁶ We have elsewhere defined and described ethnic units in each of our 32 African nations. See “National Integration and Ethnicity,” in Morrison et al., Black Africa. Each of the ethnic units has been coded on a set of variables, ordinally scaled and adapted from the codes in Murdock, George Peter, The Ethnographic Atlas (Pittsburgh: University of Pittsburgh Press, 1967)Google Scholar. A summary description of 24 such variables is given in Morrison et al., op. cit. The national variance across ethnic units has then been calculated for each of these variables, giving a number of measures of cultural pluralism. See Morrison et al., Black Africa, chapter 16. Factor analyzing these measures we obtained a solution showing three major dimensions of cultural pluralism in these nations, and these factors were then separately factor analyzed to give unidimensional factors indicating the national variance in the ethnic units' patterns of ecological adaptation, authority and stratification, and social organization. See the appendix to this paper for the content of these unidimensional factors, the factor scores for which are used as three measures of cultural pluralism in our analysis.

²⁷ Cf. the definition of the state given by Weber, Max in The Theory of Social and Economic Organization, (Glencoe, Illinois: The Free Press, 1947)Google Scholar.

²⁸ This has to do with what Deutsch calls the range of integration—“the range of rewards and deprivations for the component units, by which an integrated relationship is maintained among them.” Deutsch, , The Analysis of International Relations, p. 159Google Scholar.

²⁹ Compare the argument of Etzioni: “A political community is a community that possesses three kinds of integration: (a) it has an effective control over the use of the means of violence … (b) it has a center of decision-making that is able to affect significantly the allocation of resources and rewards throughout the community; and (c) it is the dominant focus of political identification for the large majority of politically aware citizens.” Etzioni, , Political Unification, p. 4Google Scholar.

³⁰ “The measures we use are: a) legislative fractionalization at independence, calculated as

where PR_i is the proportion of legislative seats held by the j'th of a set of different parties; b) the number of political parties established from the date of a nation's independence to 1969; c) percentage of the wage labour force employed in the public sector; d) rate of growth in the ratio of government spending to GNP, 1963–65; e) rate of growth in government revenue as per cent of GNP, 1963–65; f) rate of growth in central government revenue, 1963–66; and g) central government revenue, c. 1963. The data for these variables are given in Morrison et al., Black Africa, chaps. 7, 8, and 9.

³¹ A large number of measures relating to military and security forces in African nations were factor analyzed, and from this analysis we constructed three unidimensional factors indicating the size of military and security systems, the budgetary allocations to such systems, and the amount of foreign military aid in these nations. The content of these factors is shown in the appendix to this paper. In addition to the factor scores derived from these three scales, we use three variables that were unique to the original factor solution as measures of coercive potential: the percentage change in the national defence budget as a proportion of GNP, 1964–67; the number of political parties banned through the period independence to 1969, and the number of persons arrested for political offences in the period independence through 1969. For the sources of military data and a presentation of selected indicators, see Morrison et al., Black Africa, chap. 12. The data on parties banned is taken from Part II of that book, and the data on political arrests have been coded from the sources used in the collection of data on political instability reported in Chapter 11.

³² This conceptualization of political instability, and the theory presented here, should be sharply distinguished from more common approaches to the analysis of political instability, which are based on a conceptualization of political instability as aggression directed by individuals against the political system, and which explain such aggression by variation in indices of frustration, or relative deprivation, such as the ratio of social want satisfaction to social want formation. See Ivo, K. and Feierabend, Rosalind L., “Aggressive Behavior Within Politics 1948–1962: A Cross-National Study,” Journal of Conflict Resolution, 10 (09, 1966)Google Scholar; Davies, James, “Toward a Theory of Revolution,” American Sociological Review, 37 (1962)Google Scholar; Raymond Tanter and Manus Midlarsky, “A Theory of Revolution”; D. P. Bwy, “Political Instability in Latin America …” and Ted Gurr, “A Causal Model of Civil Strife …” The theoretical basis for these approaches is frustration-aggression theory, which is well summarized in Berkowitz, Leonard, Aggression: A Social Psychological Analysis (New York: McGraw-Hill, 1962)Google Scholar, and reviewed by Ted Gurr in “Psychological Factors in Civil Violence,” World Politics, 20 (01, 1968), 245–278CrossRefGoogle Scholar, and Why Men Rebel, Chaps. 2 and 3. Our approach should also be contrasted with that in recent research based on minimal theoretical focus, in which the data have been allowed to “speak for themselves” providing factor solutions of indicators of conflict behaviour, which are then labeled as types of political instability—e.g., turmoil and anomic violence. See, for example, Rummell, Rudolph J., “Dimensions of Conflict Behavior Within Nations, 1946— 59, Journal of Conflict Resolution, 10 (03, 1966), 65–73CrossRefGoogle Scholar, and Tanter, Raymond, “Dimensions of Conflict Behavior Within and Between Nations 1958–60” JCR, 10 (03, 1966), 41–64Google Scholar. For details of our own conceptualization of political instability, and a review of related work, see Morrison, and Stevenson, “Political Instability in Independent Black Africa: More Dimensions of Conflict Behavior Within Nations,” The Journal of Conflict Resolution, 15 (09, 1972), 347–368CrossRefGoogle Scholar.

³³ Ted Gurr, “Psychological Factors in Civil Violence.”

³⁴ Feierabend, Ivo K., Feierabend, Rosalind L., and Nesvold, Betty A., “Social Change and Political Violence: Cross-National Patterns,” in The History of Violence in America, Davies, and Gurr, , eds., pp. 632–687Google Scholar.

³⁵ Rummell, “Dimensions of Conflict Behavior Within Nations, 1946–59.”

³⁶ Gurr, Ted, “A Causal Model of Civil Strife: A Comparative Analysis Using New Indices,” American Political Science Review, 67 (12, 1968) 1104–1124CrossRefGoogle Scholar.

³⁷ See the discussion in Eckstein, Harry, “On the Etiology of Internal War,” History and Theory, 4 (1965), especially p. 135CrossRefGoogle Scholar. Factor analysis has been widely used as an inductive approach to typologies of instability. Factor solutions of data on political instability are variously entitled revolution, subversion, and turmoil (Rummell Dimensions … 46–59”); internal war and turmoil (Tanter, “Dimensions … 58–60”); or organized and anomic violence (Bwy, “Political Instability in Latin America”). Such has been the impact of these empirical findings that direct coding is now organized in terms of a factor-based typology of internal war, conspiracy, and anomic violence (Gurr, “A Causal Model of Civil Strife”). Ironically, the complexities of these typologies are for the most part ignored in analysis of the conditions giving rise to political instability, in which the dependent variable is most often an aggregate measure of civil strife or instability.

³⁸ By the intensity of violence we mean the number of persons involved in unpredictable or physically destructive behavior; the duration of such behavior; and the extent of the alterations in the established political structure brought about by such behavior.

³⁹ A review of the relevant writings on the subject is given by Rapoport, David C., “Coup d'Etat: The View of the Men Firing Pistols,” in Nomos VIII: Revolution, Friedrich, Carl J., ed. (New York: Atherton Press, 1966)Google Scholar.

⁴⁰ These data, and all measures of political instability used in this paper, are summarized in chapter 11, “Political Instability,” in Morrison et al., Black Africa. Part II of that book contains abbreviated verbal descriptions of the events coded. The sources used for the collection of these data are also detailed there.

⁴¹ Since we are dealing with the postcolonial history of Sub-Saharan Africa, we are not concerned with anticolonial revolts or with the liberation struggles in the Portuguese territories and white-controlled Southern Africa.

⁴² Elite instability is measured as follows: Successful coups d'état are given the numerical weight 5; attempted but unsuccessful coups are weighted 3; and plots, 1. The index for each country is the sum of all such events so weighted in the period from independence to 1970. Communal instability is mea-sured as follows: Civil wars are given the weight 4; rebellions, 3; irredentism, 2; and ethnic violence, 1. The index for each country is calculated by summing the numerical weights for each event multiplied by the number of years in which such an event was reported. It should be noted that the individual measures used in the calculation of these indices are highly intercorrelated, and that they represent independent dimensions in factor analytic results based on a much larger set of variables. Those results indicate support for our theoretical distinctions between elite, communal, and mass instability, and they also indicate additional evidence for the existence of an independent dimension measuring “turmoil,” which is based on indicators of strikes, riots, demonstrations, numbers killed and arrested, and terrorism. Although this dimension has been regularly reported in other studies, we do not use it as a dependent variable in this study, because we believe that the events involved are the least reliably reported and can only very loosely be interpreted as attempts to change authority relationships in the political system. See Morrison and Stevenson, “Political Instability in Independent Black Africa …”

⁴³ The notion of process as applied to the discussion so far may be confusing, but the distinction between static and dynamic models of integration is in part artificial. We see process as continuous interrelationships among characteristics of a system over time—i.e., as a series of state descriptions over time. What is important about the distinction is that it sensitizes the researcher to the consideration of historical, or time-series, relationships. Although we are not investigating the major questions of what dimensions of integration precede others in points of time in the development process, and although we cannot in this paper explore the many possibilities of lagged relationships between aspects of integration and instability, we do intend to explore the differences that result from testing our model of integration and instability with static, cross-sectional data, as opposed to tests based on variables measuring rates of change over time. Our expectation is that the results will be quite different, with important implications for the analysis of integration and instability.

⁴⁴ Lipset, Seymour Martin, Political Man (New York: Anchor Books, 1963), p. 64Google Scholar.

⁴⁵ See Deutsch, Nationalism and Social Communication.

⁴⁶ Etzioni, , Political Unification, p. 94Google Scholar.

⁴⁷ Gurr, , “A Causal Model of Civil Strife,” 1116Google Scholar, and Bwy, , “Political Instability in Latin America,” 49–50Google Scholar.

⁴⁸ Guetzkow, Harold, Multiple Loyalties: Theoretical Approach to a Problem in International Organization (Princeton: Center for Research on World Political Institutions, Princeton University, 1955), p. 33Google Scholar. Guetzkow argues for the greater effectiveness of positively reinforced rather than punitively induced loyalty to political systems.

⁴⁹ See Midlarsky and Tanter, “Toward a Theory of Political Instability in Latin America.”

⁵⁰ See the introduction to Morrison et al., Black Africa, for a more detailed discussion.

⁵¹ Campbell, Donald T. and Fiske, Donald W., “Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix,” Psychological Bulletin, 56 (03, 1959), 81–105CrossRefGoogle ScholarPubMed; citation is from p. 83. It should be noted that our own use of the multitrait, multimethod logic is not strictly tied to the argument and method illustrated in this seminal publication. Where Campbell and Fiske are concerned primarily with experimental research, our own work is based on nonexperimental data. Multitrait operationalization in our work, therefore, cannot approximate the standards of strict replication of experimental methods of data collection as a basis for assessing reliability, and the different methods of data collection we use are not strictly comparable to the alterations of treatment and experimental design discussed by Campbell and Fiske. We are also interested in an elaboration of the multimethod framework to include a multi-unit dimension for nonexperimental research: that is, we feel that data measured on different entities, or units of analysis (such as individuals, ethnic groups, and elite groups), should be compared for converging validation of conceptual variables.

⁵² In particular we utilize data collected on ethnic groups and members of cabinets, in addition to the more general kinds of statistics aggregating the characteristics of individuals or goods in a nation.

⁵³ When these procedures are not adopted, factor scores often include the effects of variables which individually load low on the factor, but may be an important influence on the total factor score.

⁵⁴ See Wright, Sewall, “The Method of Path Coefficients,” Annals of Mathematical Statistics, 5 (1934), 161–215CrossRefGoogle Scholar; and Wright, “Path Coefficients and Path Regressions: Alternate or Complimentary Concepts,” Biometrics, 16 (06, 1960), 189–202CrossRefGoogle Scholar; Duncan, Otis Dudley Jr., “Path Analysis: Sociological Examples,” American Journal of Sociology, 72 (07, 1966), 1–16CrossRefGoogle Scholar; Land, Kenneth, “Principles of Path Analysis,” in Sociological Methodology 1969 (San Francisco: Jossey-Boss, 1969)Google Scholar; Boyle, Richard, “Path Analysis and Ordinal Data,” American Journal of Sociology, 75 (01, 1970) 461–480CrossRefGoogle Scholar. A computer program for calculating path coefficients, standardized partial regression coefficients, and the Z-scores of the regression coefficients has been written by Louis Goodman, of the Sociology Department at Yale University, and Donald Morrison. This program was used for computing the results analyzed in this paper.

⁵⁵ These results are corroborated by the results of a stepwise multiple discriminant analysis in which our indicators of integration are used as independent variables, but nominal dependent variables are used to indicate whether or not a country has experienced a coup d'état or civil war, rebellion, and irredentism. This technique calculates a discriminant function, a linear combination of the independent variables, which maximizes differentiation among the groups of cases typologized according to the dependent variables. Each nation used in this analysis is then assigned a score derived from the discriminant function, and that score indicates whether the nation should be classified in one or another of the dependent variable groups. Using this technique, we are able to classify correctly all African nations as having experience or no experience of communal instability, and more than 80 per cent of African nations as having or not having experienced coups d'état, on the basis only of knowledge of their values on indicators of integration. For a discussion of this technique, see Morrison, Donald G. and Art, Richard Jr., A Fortran Program for Stepwise Multiple Discriminant Analysis, 04 1967, Vogelback Computing Center, Northwestern UniversityGoogle Scholar, and Behavioral Science, 13 (01, 1968), 82–83Google Scholar.

⁵⁶ Compare the argument in Zolberg, “… Political Conflict in the New States of Tropical Africa.”

⁵⁷ Gurr, , Why Men Rebel, p. 24Google Scholar.

⁵⁸ For more pessimistic conclusions, see Feit, Edward, “Military Coups and Political Development: Some Lessons from Ghana and Nigeria,” World Politics, 20 (01, 1968), 179–193CrossRefGoogle Scholar, and O'Connell, J., “The Inevitability of Instability,” Journal of Modern African Studies, 5 (1967), 181–191CrossRefGoogle Scholar.

⁵⁹ Some of the important considerations left out of this analysis are the effects of modernization on the relationship between cultural pluralism and political instability. We have investigated that problem in our “Cultural Pluralism, Modernization and Conflict: Sources of Political Instability in Black Africa,” Canadian Journal of Political Science (03, 1972)Google Scholar.

⁶⁰ van den Berghe, Pierre, “Dialectic and Functionalism: Toward a Theoretical Synthesis,” American Sociological Review, 28 (10, 1963), 696–697CrossRefGoogle Scholar.

⁶¹ See Black, C. E., The Dynamics of Modernization: A Study in Comparative History (New York: Harper and Row, 1966)Google Scholar.

⁶² Deutsch, , The Analysis of International Relations, p. 198Google Scholar.