¹ MacRae, D. Jr., “A Method for Identifying Issues and Factions from Legislative Votes,” this Review, 59 (12, 1965), 909–926Google Scholar; H.R. Alker, Jr., “Dimensions of Conflict in the General Assembly,” ibid., 58 (September, 1965), 642–657; Grumm, J. G., “A Factor Analysis of Legislative Behavior,” Midwest Journal of Political Science, 7 (11, 1963), 336–356CrossRefGoogle Scholar. Earlier work in this area includes MacRae, D. Jr., Dimensions of Congressional Voting (Berkeley, Calif.: The University of California Press, 1958)Google Scholar; Farris, C. D., “A Scale Analysis of Ideological Factors in Congressional Voting,” in Wahlke, J.C. and Eulau, H. (Eds.), Legislative Behavior (Glencoe, Ill: The Free Press, 1959), 399–413Google Scholar; and Harris, C. W., “A Factor Analysis of Selected Senate Roll-Call Votes, 80th Congress,” Educational and Psychological Measurement, (1948), 582–591Google Scholar.

² For a more complete discussion and critique of the existing literature see Marwell, G., Dimensions of Legislative Cleavage (Ph.D. Dissertation, Department of Sociology, New York University, 1962)Google Scholar. This work also contains more complete descriptions of all the procedures used in this paper and of the bills which comprise the various factors.

³ G. M. Belknap, “Scaling Legislative Behavior,” Wahlke and Eulau, op. cit., 388–398.

⁴ In recent factor-analytic works by political scientists the merits of oblique rather than orthogonal factors have been strongly advanced: cf. MacRae, “A Method of Identifying Issues and Factions from Legislative Votes,” op. cit.; and Alker, H. R. Jr., and Russett, B. M., World Politics in the General Assembly (New Haven: Yale University Press, 1965)Google Scholar. This is not the place for extended discussion of the merits of the two approaches, but it may be noted that differences between results of orthogonal and oblique rotations tend to be important where there are a large number of factors retained for rotation, and where the factors are poorly defined. Since the analysis below essentially involves the rotation of only three very well defined factors, and since a dominant-cluster technique for the generation of factor scores is used, results from the two approaches should be very similar. A look at figures 1–3 and the accompanying discussion should illustrate this. Note also that many analysts argue the theoretical advantages of orthogonal rotations, especially for domains with nearly general first factors. For a detailed discussion of factor analysis, see Harman, H. H., Modern Factor Analysis (Chicago: The University of Chicago Press, 1960)Google Scholar.

⁵ The “Total Variance” of the data is equal to the number of items, since factor analysis conventionally standardizes each item to unit variance. Thus, for the seventy-item matrices the figure is seventy, and “proportion of total variance” indicates the amount of variance accounted for by the factor, divided by seventy. “Predictable Variance,” sometimes called “common variance,” is the total amount of variance shared by the items of the matrix, or the sum of the communalities of the items in the matrix. Any item's communality is the amount of its variance which is common to other items in the matrix, and which may thus be predicted from these other items.

⁶ Several additional factors were also extracted in this and the following two factor analyses, but are of no interest here. Most had just one item loaded above .4, precluding interpretation of common characteristics. The rest had two clearly duplicative items defining the content, and no other substantive content.

⁷ Congressional Quarterly Almanac, Vols. V–X (Washington, D.C.; Congressional Quarterly News Features, 1949–54). The five-digit number should be interpreted as follows: the first digit gives the volume number, with 0 standing for volume ten; the next three digits give the page number; the last digit gives the number on the page given to the bill. More complete descriptions of the contents of relevant bills may be fou nd by looking in the Almanac.

⁸ For a complete discussion see Harman, op. cit., 337–361.

⁹ It might also be noted that the use of dominant clusters tends to produce scores much more similar to those which would have been derived using the results of an oblique rotation. The use of unweighted scores within clusters is obvious. However, the generally questionable value of weighting is pointed out by Aiken, L. R. in “An other Look at Weighting Test Items,” Journal of Educational Measurement (1966), 183–185CrossRefGoogle Scholar.

¹⁰ Grassmuck, G., Sectional Biases in Congress on Foreign Policy (Baltimore: The Johns Hopkins Press, 1950)Google Scholar. The definition of “Midwestern” districts used in this analysis combines Grassmuck's “Lake” and “Great Plains” states.

¹¹ Holcombe, A. N., The New Party Politics (New York: Norton, 1933)Google Scholar; Key, V.O. Jr., Public Opinion and American Democracy (New York: Knopf, 1961)Google Scholar; Alford, R., Party and Society (Chicago: Rand-McNally, 1963)Google Scholar; Grassmuck, op. cit.; Turner, J., Party and Constituency: Pressures on Congress (Baltimore: The Johns Hopkins Press, 1951)Google Scholar; and D. MacRae, Jr., Dimensions of Congressional Voting, op. cit.

¹² In comparing foreign policy votes in the 81st (Truman), 86th (Eisenhower) and 87th (Kennedy) Congresses, Kesselman documents this point. Significant proportions of Democrats moved from internationalist to isolationist positions after Eisenhower's election while a similar number of Republicans changed in the opposite direction. After Kennedy's election the process was reversed, several Republicans becoming isolationist and several Democrats changing to internationalist positions: Kesselman, M., “Presidential Leadership in Congress on Foreign Policy, A Replication of a Hypothesis,” Midwest Journal of Political Science, 9 (11, 1965), 401–406CrossRefGoogle Scholar.