¹ For example, Ball, M. Margaret, “Bloc Voting in the General Assembly,” International Organization, Vol. 5, No. 1 (February, 1951), pp. 3–31CrossRefGoogle Scholar; Triska, Jan F. and Koch, Howard E. Jr., “Asian-African Coalition and International Organization: Third Force or Collective Impotence?” Review of Politics, Vol. 21, No. 2 (April, 1959), pp. 417–55CrossRefGoogle Scholar; Goodwin, Geoffrey, “The Expanding UN; I-Voting Patterns,” International Affairs, Vol. 36, No. 2 (April, 1960), pp. 174–87CrossRefGoogle Scholar; Rieselbach, Leroy N., “Quantitative Techniques for Studying Voting Behavior in the UN General Assembly,” International Organization, Vol. 14, No. 2 (Spring 1960), pp. 297–304CrossRefGoogle Scholar; Lijphart, Arend, “The Analysis of Bloc Voting in the General Assembly: A Critique and a Proposal,” this Review, Vol. 57, No. 4 (December, 1963), pp. 902–17Google Scholar; Riggs, Robert E., Politics in the United Nations (Urbana, University of Illinois Press, 1958)Google Scholar; Hovet, Thomas Jr., Bloc Politics in the United Nations, (Cambridge, Harvard University Press, 1960)CrossRefGoogle Scholar, and Africa in the United Nations (Evanston, Northwestern University Press, 1963)Google Scholar.

² These distinctions are made by Dahl, Robert, “The Concept of Power,” Behavioral Science, Vol. 2, No. 3 (July, 1957), pp. 201–15CrossRefGoogle Scholar.

³ Africa in the United Nations, pp. 181–85.

⁴ Dahl, Robert, Who Governs? (New Haven, Yale University Press, 1963)Google Scholar.

⁵ See Hovet, Bloc Politics in the United Nations, op. cit., pp. 130–88. Ernst B. Haas has used to advantage a somewhat more specific set of issue-groupings in an analysis of the “coverage” of different conventions ratified by members of the International Labor Organization. See his “System and Process in the International Labor Organization, A Statistical Afterthought,” World Politics, Vol. 14, No. 2 (January, 1962), pp. 339–52Google Scholar.

⁶ Sydney Bailey has presented an interesting discussion of the different tactical approaches to this issue in “The Question of Tibet,” The General Assembly of the United Nations (New York, 1961)Google Scholar, ch. 10.

⁷ Hovet, Bloc. Politics, op. cit., p. 112, feels it “premature” to attempt any formal conclusions about the implications of the total phenomenon. He does suggest, however, that with its mixed blessings the “bloc” phenomenon is likely to increase and should be better understood. Fortunately in Africa in the United Nations considerably more attention is given to the specific issues on which group members disagree and how their votes are related to the American and Soviet positions. This methodology is much more useful than the ad hoc verbal descriptions of group differences that supplemented the bar graphs in Bloc Politics. Arend Lijphart, op. cit., pp. 913–17, has also studied voting cohesion with the positions of other states in mind. His Figure 2 seems quite clearly two-dimensional. The proper dimensionality of UN voting patterns is discussed in detail below.

⁸ Haas, Ernst B., “Dynamic Environment and Static System: Revolutionary Regimes in the United Nations,” in Kaplan, Morton (ed.), The Revolution in World Politics (New York, 1962), p. 278Google Scholar. Italics omitted.

⁹ Neither of these techniques is new to political science. Glendon Schubert has factor-analyzed judicial votes in “The 1960 Term of the Supreme Court: A Psychological Analysis,” this Review, Vol. 56, No. 1 (March, 1962), pp. 90–107Google Scholar; see also MacRae, Duncan Jr., and Meldrum, James A., “Critical Elections in Illinois: 1888–1958,” this Review, Vol. 54, No. 3 (September, 1960), pp. 669–83Google Scholar. Regression is a basic deductive technique in The American Voter. See Campbell, Stokes and Miller, , "“Components of Electoral Decision,” this Review, Vol. 52, No. 2 (June, 1958), pp. 367–87Google Scholar. A readable summary of both techniques may be found in Blalock, Hubert Jr., Social Statistics (New York, 1960)Google Scholar, chs. 17–19 and 21. What is new about these techniques is the ease with which they may be applied. The seventy roll call factor analyses reported here took eleven minutes on the IBM 7090. The regressions took less than a minute each on the IBM 709.

¹⁰ Hayward R. Alker, Jr. and Bruce M. Russett, World Politics in the General Assembly (forthcoming). Power measurements are facilitated by the use of both the conflict dimensions presented below and additional information on the sponsorship of resolutions and the intensity of involvement of different groups on these conflicts. Voting cohesion may satisfactorily be measured in terms of mean deviations on the factor scores presented below.

¹¹ Designating plenary meetings as the “Eighth Committee” and the Special Political Committee as the “0th Committee,” the first digit of the identification numbers for each roll call in the Tables corresponds to the Committee in which the vote occurred; the numbers after this digit refer to the order of occurrence of the roll call within the Committee. “S.C.E.A.R.” stands for the Scientific Committee on the Effects of Atomic Radiation; “P.D.R.” symbolizes the People's Democratic Republic of Korea; and “S.C.17” refers to the Special Committee of 17 set up at the Fifteenth Session to implement the Declaration on the Granting of Independence to Colonial Countries and Peoples. “UNCURK” is the United Nations Commission for the Unification and Rehabilitation of Korea, while “CINSGT” is the author's label for the Committee on Information from Non-Self-Governing Territories.

¹² Symbolically the model of voting positions used in factor analysis is

where u _ji is the unexplained voting position of country i on roll call j. For more details on the roll calls and methods described above, see Alker, Hayward R. Jr., Dimensions of Voting in the General Assembly (Ph.D. Dissertation, Department of Political Science, Yale University, 1963)Google Scholar. Harmon, Harry H., Modern Factor Analysis (Chicago, University of Chicago Press, 1960)Google Scholar, ch. 9, derives the “principal component” method of factor analysis used in this article.

¹³ Haas, Ernst, “Regionalism, Functionalism, and Universal International Organization,” World Politics, Vol. 8, No. 2 (January, 1956), pp. 238–63CrossRefGoogle Scholar.

¹⁴ Stoessinger, John G., The Might of Nations (New York, 1961)Google Scholar.

¹⁵ Bloomfield, Lincoln, The United Nations and United States Foreign Policy (Boston, 1960), p. 10Google Scholar.

¹⁶ All uncorrelated factors underlying or “explaining” more than the variance of a single roll-call were extracted and are presented in Table I. For the rationale for stopping factoring at this point, see Harry H. Harmon, op. cit., ch. 14. The variances “explained” by these nine dimensions of conflict are 37.0, 7.8, 4.2, 2.4, 1.8, 1.4, 1.4, 1.1, 1.0, leaving only 17% of all voting “unexplained.”

¹⁷ In matrix notation the equation (s) in footnote 12 can be written as V̂ = AC When A is not a square matrix, component scores are given by the equation C =(A ¹A)⁻¹A′V. See Kaiser, Henry, “Formulas for Component Scores,” Psychometrika, Vol. 27 (March, 1962), pp. 83–8CrossRefGoogle Scholar. As calculated here, these sets of factor scores are very nearly uncorrelated with each other.

¹⁸ See Harmon, loc. cit. A further step, not taken in this article, would be to find the simplest factor structure using correlated (oblique) factors.

¹⁹ Liska, George, International Equilibrium (Cambridge, Harvard University Press, 1957)CrossRefGoogle Scholar.

²⁰ Ernst B. Haas, “Regionalism,” op. cit., esp. pp. 238–41, 260–63.

²¹ O'Brien, Conor Cruise, To Katanga and Back (London, Hutchinson, 1962), p. 18Google Scholar.

²² These rotated factors “explain” the equivalent of 22.7, 8.7, 10.2, 7.6, 1.8, 1.4, 1.6, 2.4, and 1.6 roll calls respectively.

²³ Cited in Padelford, N. J. and Emerson, R., Africa and World Order (New York, 1963), p. 45Google Scholar

²⁴ The West Irian roll calls were all in plenary meetings but are listed in Tables I and II with other colonial questions because they were discussed in terms of the Declaration on the Granting of Independence to Colonial Countries and Peoples.

²⁵ An important argument for both this fourfold substantive classification and the two-dimensional geopolitical interpretation offered earlier is that similar analyses seem appropriate for the Second, Seventh and Twelfth General Assemblies. Cf. Alker, op. cit., or Alker and Russett, op. cit. (forthcoming).

²⁶ Analyses of the East-West conflict in simple behavioral terms often run into the same confussion that Hovet notes regarding similar Soviet and Afro-Asian voting behavior: they do not distinguish Cold War from self-determination issues. Empirically, the “substantive” and “behavioral” viewpoints are complementary. Each roll call can be checked as to both its rotated and unrotated components. Correlating factor loadings or factor scores makes this relationship explicit. The East-West and North-South correlations of self-determination factor scores are 0.74 and 0.43. The same numbers for the Cold War alignment are 0.50 and −0.20; for Moslem questions they are 0.39 and −0.12; and for supranationalism factor scores −0.25 and −0.86.

²⁷ Alger, Chadwick, “Non-Resolutional Consequences of the United Nations and Their Effect on International Conflict,” Journal of Conflict Resolution, Vol. 5, No. 2 (June, 1961), pp. 128–47CrossRefGoogle Scholar.

²⁸ Early Soviet support of Security Council resolutions on the Congo is sometimes overlooked. See Dallin, Alexander, The Soviet Union at the United Nations (New York, 1962)Google Scholar, ch. 10.

²⁹ The chart on page 126 summarizes membership in both kinds of groupings in 1959.

³⁰ Ernst Haas, “System and Process in the International Labor Organization: A Statistical Afterthought,” loc. cit.

³¹ Russett, Bruce M., “The Calculus of Deterrence,” The Journal of Conflict Resolution, Vol. 7, No. 2 (June, 1963), pp. 97–109CrossRefGoogle Scholar, and Community and Contention: Britain and America in the Twentieth Century (Cambridge, 1963)Google Scholar, ch. IV.

³² Arend Lijphart, op. cit, has correctly pointed out the limitations of arbitrarily chosen groups in the study of Assembly politics. The groupings of states used in Table III represent combinations of regular and ad hoc caucusing units and regional interest groups. They serve as a convenient compromise between inductively found groupings (as in Figure 1) and Hovet's list of regular caucusing groups (Africa in the United Nations, p. 74.). Old Europeans are a frequently united group of West Europeans, Old Commonwealth members, the United States and South Africa; Mongolia is considered only a member of the Soviet bloc; Turkey and Cyprus are considered Asian states. Arabs, Africans and Asians have been exclusively defined with respect to each other, Old Europeans and the Soviet bloc. Yugoslavia, Israel and Nationalist China have not been listed with any group.

³³ The R²s for equations (1)–(4) below are exceptionally high ones in social research. The multiple correlation coefficient of the East-West model is 0.88. The equation explains 78% of the major voting alignment in the General Assembly. Similarly, the North-South Model explains 70%, the Cold War and supranationalism models 48% and 44% respectively. Even for factor scores in their present largely uncorrelated form (see note to Table III), high correlations with the rotated self-determination conflict indicate that influences affecting these policy positions differ from those which determine Cold War membership alignments.

Using regional and caucusing-group variables that are much more immediate to the voting (see Table III), the corresponding R²s are roughly similar: 0.79, 0.78, 0.30 and 0.72. Cold War membership alignments can be explained better by more distant environmental variables, like aid and trade, than by caucusing-group memberships. Supranationalism and self-determination (with group membership regressions explaining 72% and 85% respectively) definitely reflect a good deal of group cohesion. Group cohesion by itself, however, is a poor predictor of what a group has agreed on, for which additional explanations, like equations (1)–(4), are required.

³⁴ In the case of a standardized dependent variable, standardized β-coefficients are obtained by multiplying a concrete b-coefficient times the standard deviation of its independent variable, β-coefficients are comparable; b-weights are not.

³⁵ In these equations b-coefficients are given with the independent variables, β-weights are given below them in parentheses. The concrete units for dichotomous variables (e.g., alliances) are 1 or 0; for aid figures they are $100 per capita; for per capita G.N.P. $1,000; and for all percentage figures, 100%. It should be noted that the communist “independent” variables are highly intercorrelated, enough so that further analysis using a single index of communist ties might be desirable.