¹ The normative questions of representation are treated in Pitkin, Hanna F., ed., Representation (New York: Atherton, 1969)Google Scholar.

² The most current and comprehensive summary of past findings can be found in Fiorina, Morris P., Representatives, Roll Calls, and Constituencies (Lexington: D.C. Heath, 1974)Google Scholar, especially Ch. 1. See also Eulau, Heinz and Hinckley, Katherine, “Legislative Institutions and Process” in Political Science Annual, Vol. 1, ed. Robinson, James A. (Indianapolis: Bobbs-Merrill, 1966), pp. 85–189Google Scholar, and Huitt, Ralph K. and Peabody, Robert L., Congress: Two Decades of Analysis (New York: Harper & Row, 1969), pp. 3–73Google Scholar.

³ The recurrent finding of weak or nonexistent constituency influence has been challenged on methodological grounds. See, for example, Fiorina, Morris P., “Constituency Influence: A Generalized Model and Its Implications for Statistical Studies of Roll-Call Behavior,” Political Methodology, 2 (1975), 249–66Google Scholar.

⁴ Especially convincing is Miller, Warren E. and Stokes, Donald E., “Constituency Influence in Congress,” American Political Science Review, 57 (1963), 45–56CrossRefGoogle Scholar.

⁵ Although the predominance of party is generally accepted, questions have been raised about its actual influence. Shannon, for example, argues that party position may be determined by the fact that most legislators of a given party face similar demands from constituency majorities. By voting in accordance with these majority preferences, legislators coincidentally establish party position. Thus, party and constituency influences cannot be completely separated in roll-call analysis. Shannon, W. Wayne, Party, Constituency and Congressional Voting (Baton Rouge: Louisiana State University Press, 1968)Google Scholar, Ch. 6. This contention has received empirical support in Kuklinski, James H., “District Competitiveness and Legislative Roll-Call Behavior: A Reassessment of the Marginality Hypothesis,” American Journal of Political Science, 21 (1977), 627–38CrossRefGoogle Scholar.

⁶ This thesis most recently has been explicated and articulated in Mayhew, David R., Congress: The Electoral Connection (New Haven: Yale University Press, 1974)Google Scholar. As Mayhew notes, this response can be manifested in ways other than roll-call voting. One example is the provision of governmental services. See Fiorina, Morris P., “The Case of the Vanishing Marginals: The Bureaucracy Did It,” American Political Science Review, 71 (1977), 177–81CrossRefGoogle Scholar.

⁷ There is no consensus on the “true” nature of the typical voter. For a summary of this continuing debate, see the June 1972 issue of the American Political Science Review, 415–70; Asher, Herbert, Presidential Elections and American Politics: Voters, Candidates and Campaigns Since 1952 (Homewood, Ill.: Dorsey, 1976), especially Pt. IGoogle Scholar; and Niemi, Richard G. and Weisberg, Herbert F., eds., Controversies in American Voting Behavior (San Francisco: W. H. Freeman, 1976), especially Sees. II and IIIGoogle Scholar.

⁸ Miller, and Stokes, , “Constituency Influence,” 48Google Scholar.

⁹ Kingdon, John W., Congressmen's Voting Decisions (New York: Harper and Row, 1973), pp. 59–60Google Scholar.

¹⁰ Mayhew, , Electoral Connection, pp. 13–15Google Scholar. There is some evidence, however, that careerism has begun to decline. See Oppenheimer, Bruce I., “Sub-committee Government and Congressional Reform,” DEA News (Summer 1976), S8–S11Google Scholar.

¹¹ The contradictory findings regarding the marginality hypothesis are aptly summarized in Sullivan, John L. and Uslaner, Eric M., “Congressional Behavior and Electoral Marginality: Some Considerations Based Upon Spatial Models of Electoral Competition” (paper presented at the Annual Meeting of the Midwest Political Science Association in Chicago, 30 April 1976)Google Scholar. The electoral marginality of a district and the simple occurrence of an election are, of course, distinct phenomena. No one has addressed the potentially interesting question of how these two variables interact to affect legislative behavior.

¹² Richard C. Elling, “Patterned Temporal Ideological Instability in the United States Senate: Of Time and Electoral Responsiveness” (unpublished manuscript). See also Brimhall, Dean R. and Otis, Arthur S., “Consistency of Voting by our Congressmen,” Journal of Applied Psychology, 32 (1948), 1–7CrossRefGoogle Scholar.

¹³ McClosky, Herbert, Hoffman, Paul, and O'Hara, Rosemary, “Issue Conflict and Consensus Among Party Leaders and Followers,” American Political Science Review, 54 (1960), 406–72CrossRefGoogle Scholar. This finding has been corroborated in a more recent study of mass and elite opinion: see Shaffer, William, Weber, Ronald, and Montjoy, Robert, “Mass and Political Elite Beliefs about the Policies of the Regime” (paper presented at the Annual Meeting of the American Political Science Association in New Orleans, 7 September 1973)Google Scholar.

¹⁴ Dye, Thomas R., “A Comparison of Constituency Influence in the Upper and Lower Chambers of a State Legislature,” Western Political Quarterly, 14 (1964), 473–80Google Scholar.

¹⁵ Bumham and Sprague provide an excellent general treatment of the additive linear model: Burnham, Walter Dean and Sprague, John, “Additive and Multiplicative Models of the Voting Universe: The Case of Pennsylvania: 1960–1968,” American Political Science Review, 64 (1970), 471–90CrossRefGoogle Scholar. The additive and linear assumptions are questioned in Kuklinski, James H., “Constituency Opinion: A Test of the Surrogate Model,” Public Opinion Quarterly, 41 (1977), 34–40CrossRefGoogle Scholar.

¹⁶ Crane, Wilder W. Jr., “Do Representatives Represent?” Journal of Politics, 22 (1960), 295–99CrossRefGoogle Scholar; Hedlund, Ronald D. and Friesema, H. Paul, “Representatives' Perceptions of Constituency Opinion,” Journal of Politics, 34 (1972), 730–52CrossRefGoogle Scholar; Erikson, Robert S., Luttbeg, Norman R., and Holloway, William V., “Knowing One's District: How Legislators Predict Referendum Voting,” American Journal of Political Science, 19 (1975), 231–46CrossRefGoogle Scholar. An alternative approach to the problem of identifying constituency opinion has been developed as part of Frank Munger's “Two-Party Competition in the American States” project. The technique is described in Weber, Ronald E., Public Policy Preferences in the States (Bloomington: Institute of Public Administration, Indiana University, 1972)Google Scholar, and Weber, Ronald E., Hopkins, Anne H., Mezey, Michael L., and Munger, Frank J., “Computer Simulation of State Electorates,” Public Opinion Quarterly, 36 (1972–1973), 549–65CrossRefGoogle Scholar. This technique is appealing because it is more widely applicable thanreferendum voting. There is some evidence, however, that public opinion estimates have been confused with the socioeconomic factors that define them. See Seidman, David, “Simulation of Public Opinion: A Caveat,” Public Opinion Quarterly, 38 (1975), 331–42CrossRefGoogle Scholar.

¹⁷ The arguments of John E. Jackson have been particularly convincing: see his Constituencies and Leaders in Congress (Cambridge: Harvard University Press, 1974), especially pp. 1–12CrossRefGoogle Scholar. For more detailed discussions of complex models, see Haefele, Edwin T., “A Utility Theory of Representative Government,” American Economic Review, 61 (1972), 350–61Google Scholar; Rothenberg, Jerome, “A Model of Economic and Political Decision-Making,” in Maigolis, Julius, ed., The Public Economy of Urban Communities (Washington, D.C.: Resources for the Future, 1965), pp. 1–38Google Scholar; and Ferejohn, John A., Pork Barrel Politics: Rivers and Harbors Legislation, 1947–1968 (Stanford: Stanford University Press, 1974)Google Scholar.

¹⁸ Clausen, for example, argues that a representative's “real” constituency consists of those who voted for him in the last election. Yet, his analysis is ultimately based on the legally specified constituency. Clausen, Aage R., How Congressmen Decide: A Policy Focus (New York: St. Martin's, 1973), Ch. 6Google Scholar.

¹⁹ Key, V. O., Public Opinion and American Democracy (New York: Knopf, 1967), p. 8Google Scholar.

²⁰ Ibid., Ch. 16.

²¹ Clausen, , How Congressmen Decide, pp. 16–17Google Scholar.

²² There were a total of 51 propositions for the three elections. Unfortunately, incomplete county returns for 18 propositions precluded their use in this study.

²³ The factor analysis was done with unities as the estimates of communalities and an orthogonal (varimax) rotation was used to obtain the matrix of factor loadings. The orthogonal rotation was selected as optimal.

²⁴ Two dimensions were discarded because of the lack of high loadings.

²⁵ Erikson, Robert S. and Luttbeg, Norman R., American Public Opinion: Its Origins, Content, and Impact (New York: Wiley, 1973), p. 54Google Scholar.

²⁶ Rostvold, Gerhard N., Financing California Government (Belmont, Calif.: Dickenson Press, 1967), p. xiGoogle Scholar.

²⁷ This point is made to distinguish government administration and government management. For a full discussion of the latter, see Clausen, How Congressmen Decide.

²⁸ Spearman, Charles, “General Intelligence, Objectively Determined and Measured,” American Journal of Psychology, 15 (1904), 201–93CrossRefGoogle Scholar. We used Spearman's two-factor solution, as opposed to more common techniques, for several reasons. First, the two-factor solution is particularly appropriate when the existence of a single factor is a viable hypothesis. Since our initial lists of bills were determined by the previously identified constituency policy dimensions, the existence of a single factor was assumed. Second, if Guttman scaling, the usual scaling technique in political science, had been used, a large number of bills falling on the various dimensions would have been excluded from subsequent analysis. Our objective was to ensure unidimensionality yet retain as many bills as possible; Spearman's technique met this objective. Finally, most of our attention in the following pages will be focused on the contemporary liberalism dimension. As Weisberg has noted, the Guttman scale may not be appropriate when the direction of social change is at issue. See Weisberg, Herbert F., “Scaling Models for Legislative Roll-Call Analysis,” American Political Science Review, 66 (1972), 1306–15CrossRefGoogle Scholar. It is not our intention to discuss the relative merits of the order-relations and proximity-relations models. Such questions are adequately raised and handled in the Weisberg article.

Particularly helpful references to the technical and theoretical applications of Spearman's test of unidimensionality include: Sir Thomson, Godfrey, The Factorial Analysis of Human Ability (Boston: Houghton-Mifflin, 1939)Google Scholar; Fruchter, B., Introduction to Factor Analysis (New York: Van Rostand, 1954)Google Scholar; and Harman, Harry H., Modern Factor Analysis, 3rd ed. (Chicago: University of Chicago Press, 1976)Google Scholar.

The problem of absentee voting was also given considerable attention. Our ultimate decision was to exclude absences in the scoring. The basic assumption underlying this decision is that the roll-call votes on which no absences are observed, with respect to each legislator, are a representative sample of the set of roll calls in the scale. A careful review of the absences gave us no reason to believe that the remaining roll calls were unrepresentative of the whole set for any legislator. In addition, legislators having an absentee rate of 50 percent or more on a particular dimension were excluded from further analyses. Our final Ns were: 1970 senate = 39; 1971–1973 senate = 40; 1970 assembly = 73; 1971 and 1973 assembly = 75; 1972 assembly = 76.

²⁹ The analysis is based on a tetrachoric r matrix. The applicability of tetrachoric r to dimensional analysis is discussed in MacRae, Duncan Jr., Issues and Parties in Legislative Voting (New York: Harper & Row, 1970), Ch. 3Google Scholar, and Weisberg, Herbert F., “Dimensional Analysis of Legislative Roll Calls” (Ph.D. dissertation, University of Michigan, 1968)Google Scholar. In the strictest sense, factor analysis is based on a Pearson correlation matrix. The major consequence of relaxing this assumption was the elimination of a few bills on each original list which otherwise might have been retained. For more general and lucid discussions of considerations involved in selecting appropriate measures of statistical relationship, see MacRae, , Issues and Parties, esp. pp. 98–110Google Scholar, and Weisberg, Herbert F., “Models of Statistical Relationship,” American Political Science Review, 68 (1975), 1638–55CrossRefGoogle Scholar. Their discussions of the sensitivity of the various models to the marginals is particularly helpful.

³⁰ Accordingly, our measure of policy agreement is the correlation of the appropriate factor scores. It is these conelations which are reported throughout the remainder of this paper.

³¹ These two issues received widespread coverage during the 1972 campaign.

³² Interviews were conducted with assemblymen during August 1974.

³³ MacRae, Duncan Jr., Dimensions of Congressional Voting (Berkeley: University of California Press, 1958)Google Scholar, Appendix C. For an economic argument which leads to the same conclusion reached by MacRae, see Breton, Albert, The Economic Theory of Representstion (Chicago: Aldine, 1974)Google Scholar.

³⁴ We have chosen the contemporary liberalism dimension because of the higher level of representation on it and also because of the lack of any discernible pattern on the other two dimensions.

³⁵ This correlation is particularly impressive when it is remembered that both sets of data, constituency and legislative, were previously factor analyzed, thus removing some of the original variance in our data.

It should be noted that the differences reported in this section are not an artifact of different variances of the variables resulting from partition of the data.

³⁶ Figure 1 also suggests that representatives do respond. Eulau and Karps argue recently that researchers have not provided evidence of response, as opposed to simple congruence. See Eulau, Heinz and Karps, Paul K., “The Puzzle of Representation: Specifying Components of Responsiveness,” Legislative Studies Quarterly, 2 (1977), 233–54CrossRefGoogle Scholar. Stone also has used a longitudinal analysis to provide evidence of response: Stone, Walter J., “A Panel Analysis of Representation in Congress: A Preliminary Report” (paper presented at the annual meeting of the American Political Science Association, Washington, D.C., 4 September 1977)Google Scholar.

³⁷ This statement was made by former Hughes, Iowa Governor Harold, Cedar Rapids Gazette, 20 April 1964Google Scholar.

³⁸ It might be argued that the greater overall representativeness by assemblymen can be attributed to district size. The high levels of representation by senators during election years indicate that the difference cannot be explained by this factor.

³⁹ Cited in The Federalist Papers, No. 53.

⁴⁰ The Federalist Papers, No. 52. James Madiso the author of both numbers 52 and 53, himself favored a longer term of office.

⁴¹ This theme is echoed throughout Key, V. O., American State Politics: An Introduction (New York: Knopf, 1966)Google Scholar.

⁴² Malcolm E. Jewell and Samuel C. Patterson provide one of the better summaries of the impact of presidential elections on constituency voting for state legislators. See their The Legislative Process in the United States, 2nd ed. (New York: Random House, 1973), pp. 124–31Google Scholar.

⁴³ Key, , American State Politics, p. 32Google Scholar.

⁴⁴ The findings also may say something about the limits of representativeness, at least as political scientists have measured it.

⁴⁵ In fact, they were slightly less representative. Correlations for Democrats and Republicans were .41 and .44, respectively.

⁴⁶ See footnote 4 for a discussion of the basic problem. An adequate test for constituency versus party influence would be to observe the relationship between party-roll-call voting correlations and constituency-roll-call voting correlations across the four years in each chamber. However, given the high correlations of party and roll-call voting, including the lack of variance thereof, we do not have the requisite data base.