Hostname: page-component-7c8c6479df-995ml Total loading time: 0 Render date: 2024-03-19T08:20:31.947Z Has data issue: false hasContentIssue false

Comment: What Have You Done for Me Lately? Toward An Investment Theory of Voting*

Published online by Cambridge University Press:  01 August 2014

Samuel Popkin
Affiliation:
Cambridge Survey Research andUniversity of California, San Diego
John W. Gorman
Affiliation:
Cambridge Survey Research
Charles Phillips
Affiliation:
University of Texasat Austin
Jeffrey A. Smith
Affiliation:
University of Texasat Austin

Abstract

Image of the first page of this content. For PDF version, please use the ‘Save PDF’ preceeding this image.'
Type
Articles
Copyright
Copyright © American Political Science Association 1976

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Campbell, Angus, Converse, Philip E., Miller, Warren E., and Stokes, Donald E., The American Voter (New York: John Wiley & Sons, 1960)Google Scholar. In citing this and other works from the SRC corpus, we are attempting to outline the central verbal model underlying the work. We are fully aware that alternative quotations that qualify and sometimes even contradict can be found. For a similar effort, see Walter Dean Burnham, “Contributions of the SRC to the Development of Voting Theory” (paper presented at the Fall 1975 APSA convention.) As Burnham notes, “… it is fair to say that there is no single place in any of this immense (SRC) corpus of work where a comprehensive set of theoretical propositions concerning voting behavior is laid down, with each propostion linked with the next in a closed analytical framework. The problem therefore arises that any effort by an outside scholar to provide an approximation to such a summary is likely to be challenged as misleading or incomplete” (Burnham ms. pp. 4–5).

2 Downs, Anthony, An Economic Theory of Democracy (New York: Harper & Row, 1957)Google Scholar. Other examples of the rational-economic approach are: Riker, William H. and Ordeshook, Peter C., “A Theory of the Calculus of VotingAmerican Political Science Review 63 (03 1968) 2543CrossRefGoogle Scholar; Goldberg, Arthur S., “Social Determinism and Rationality as Basis of Party Identification,” American Political Science Review 62 (1969), 525CrossRefGoogle Scholar; and Frohlich, Norman, Oppenheimer, Joe A., Smith, Jeffrey A. and Young, Oran R., “A Test of Downsian Voter Rationality: 1964 Presidential Voting, American Political Science Review (forthcoming)Google Scholar.

3 Campbell, et al. , The American Voter, p. 550Google Scholar.

4 Campbell, Angus, “Voters and Elections: Past and Present,” Journal of Politics, 26 (11 1964), 756CrossRefGoogle Scholar.

5 Campbell, Angus, “Interpreting the Presidential Victory,” in The National Election of 1964, ed. Cummings, Milton C. Jr. (Washington, D.C.: The Brookings Institution, 1966), p. 269Google Scholar.

6 Converse, Philip, Miller, Warren, Rusk, Jerrold, and Wolfe, Arthur, “Continuity and Change in American Politics: Parties and Issues in the 1968 Election,” American Political Science Review 63 (12 1969), 1096CrossRefGoogle Scholar.

7 Cain, Glen G. and Watts, Harold W., “Problems in Making Policy Inferences from the Coleman Report,” American Sociological Review, 35 (04, 1970), 228242CrossRefGoogle Scholar, deal most elegantly with the fallacies and pitfalls of “effect on R2” type measures; see especially their Table 1.

8 Stokes, Donald, Campbell, Angus, and Miller, Warren E., “Components of Electoral Decision,” American Political Science Review 52 (06 1958), 367387CrossRefGoogle Scholar; Campbell, Angus and Stokes, Donald E., “Partisan Attitudes and the Presidential Vote,” in American Voting Behavior, ed. Burdick, Eugene and Brodbeck, Arthur J. (Glencoe, Ill: Free Press, 1959) pp. 353371Google Scholar; Stokes, Donald, “Some Dynamic Elements of Contests for the Presidency,” American Political Science Review 60 (03 1966), 1928CrossRefGoogle Scholar; and Campbell, et al. , The American Voter, pp. 524531Google Scholar.

9 This short description of the scales is from Stokes, et al. , “Components of Electoral Decision,” p. 380Google Scholar. Although the exact coding methods used by the SRC in their works have never been published, the assignment of responses in most cases is indicated by their code book. Using our best judgments on appropriate grouping procedures we were able to replicate all published results accurately enough to ensure that the original coding scheme would not produce any non-trivial changes in our results. There was, however, a change in format in the 1972 survey which limited respondents to three rather than five evaluative comments for and against each party and candidate. Therefore, in these tables we have employed only the first three responses for each question in our estimation of the six-factor models for past elections. This means that these tables do not correspond directly with tables using all five responses. Kagay, Michael R. and Caldeira, Greg A., (“I Like the Looks of his Face: Elements of Electoral Choice, 1952–1972,” paper delivered at the 1975 Annual Meeting of the American Political Science Association, San Francisco, California, Sept. 2–5, 1975)Google Scholar have performed a remarkably similar analysis with slightly different coding schemata (based on three different coding schemata supplied them by the SRC) and using all five responses for the earlier years. Although their results do not accord exactly with those reported here, this is largely due to our use of three responses in earlier years. Neither paper's coding schema, and neither five- nor three-answer procedures supports the assertions of “Majority Party.” We wish to acknowledge collegial assistance of the highest order from Kagay and Caldeira.

10 While the truncated nature of the dependent variable means that probit analysis is preferable to regression analysis, we have adhered to the original SRC method. Kagay and Caldeira (Ibid.) have used probit analysis as a check and found no serious changes from the regression results.

11 The SRC's 1968 election survey provides a rich set of data which has been underutilized by political scientists because of the very element which makes it unique: the presence of a viable third-party candidate. The SRC's six-factor model is explicitly based on two-way comparisons, because three-way elections raise the issue of strategic voting so we have concentrated on the two-way elections.

12 “The 1972 election … exposed the pervasive consequences that follow from dislodging party loyalty and candidate appeal from positions of dominance in the mass electoral process and complementing them with a wide-spread concern over policies intended to resolve very real national problems” (p. 753).

13 The large candidate weights for 1972 might be explained away by arguing that respondents were actually reacting to the candidate solely in terms of his issue positions, so that the candidate is merely a surrogate for issues. The correlations between the issue factors and the candidate factors, however, are less than .2 and are among the lowest in the correlation matrix.

14 In describing group-oriented concerns as a second level, lesser articulation of interests, Campbell et al. argue that “… there is little comprehension of long-range plans for social betterment, or of basic philosophies rooted in postures toward change or abstract conception of social and economic structure of causation. The party or candidate is simply endorsed as being ‘for’ a group with which the subject is identified or as being above the selfish demands of groups within the population. Exactly how the candidate or party might see fit to implement or avoid group interests is a moot point, left unrelated to broader ideological concerns. But the party or candidate is ‘located’ in some affective relationship toward a group or groups, and the individual metes out trust on this basis.” Campbell, et al. , The American Voter, pp. 234235Google Scholar. In addition, see the discussion of mass political concepts, pp. 218–225.

15 Two recently published studies using the same data sets in somewhat different contexts have produced similar conclusions. Declercq et al. report a four-factor model including party identification, issue orientation, candidate image and party image predicting vote for five presidential elections (1956–72). Candidate image produces the largest unstandardized regression coefficient for 1972, with issue-orientation second. Both of these variables have increased in this measure of importance over the five elections while the two party variables have decreased. Declercq, Eugene, Hurley, Thomas L., and Luttbeg, Norman R., “Voting in American Presidential Elections, 1956–1972,” American Politics Quarterly Vol. 3 (07, 1975), 222246CrossRefGoogle Scholar. Kirkpatrick et al., employ total candidate and party attitude scores in a regression on vote for 1952–72. Only the standardized regression coefficients are recorded, with the weight of candidate attitudes monotonically increasing over this period, and party monotonically decreasing. A three-factor model, including party identification, shows similar results, with candidate declining only for 1968. Kirkpatrick, Samuel A., Lyons, William, Fitzgerald, Michael R., “Candidates, Parties and Issues in the American Electorate: Two Decades of Change,” American Politics Quarterly, 3, (07, 1975), 247283CrossRefGoogle Scholar.

16 The combined term, however, has often been confused in presentation with the meaning more properly associated with the unstandardized coefficient itself. This distinction can be clarified by consideration of a pair of polar cases. If a particular factor had, for a given election, a very large regression coefficient—indicating that it was an important highly salient factor in voters' decisions— but had a mean of zero—indicating widespread disagreement on the issue within the electorate—the for the factor would be zero, indicating that its net effect on the election was nil. On the other hand, widespread agreement in a particular area would produce a very large mean, and consequently a large , ceteris paribus. This accounts in large part for the great deal of importance attached to the electorate's candidate evaluations during the fifties, when there were few negative comments recorded about Eisenhower, especially in 1956. In fact, though, the candidate weights (B) themselves were not nearly as dominant in the fifties as the focus on their net effect () suggested at the time.

17 In The American Voter widely disbursed issue positions within the parties were described as lack of congruence or lack of concensus and were used as inferential evidence that issues were not important. Now widely disbursed issue positions within the party are described as “poliarization.” See Campbell, et al. , The American Voter, pp. 185187Google Scholar.

18 If is the difference on factor i between two individuals (j, k) (i.e., XijXik) the difference between them in the probability of a Republican vote due to their difference on factor i will be: . Over the entire population (for all j, k) the average value of this measure will always be zero. Therefore a mean square measure is employed so that over the population we have

for j, k or

which is equal to

For convenience has been dropped, Kagay and Caldera, op. vit also investigated several other ingenious and noteworthy measures of partisan polarization.

19 Campbell, , “Voters and Elections: Past and Present,” p. 746Google Scholar. See also Converse, Philip E., “The Concept of the Normal Vote,” in Elections and the Political Order, ed. Campbell, Angus, Converse, Philip, Miller, Warren, and Stokes, Donald (New York: Wiley, 1966), pp. 939Google Scholar.

20 Campbell, et al. , The American Voter, p. 172173Google Scholar.

21 Campbell, , “Voters and Elections. …” p. 752Google Scholar.

22 Converse, Philip E., “Comment: The Status of Nonattitudes,” American Political Science Review 68 (06 1974), 650Google Scholar.

23 Converse, Philip E., “Change in the American Electorate,” in The Human Meaning of Social Change, ed. Campbell, Angus and Converse, Philip (New York, Russell Sage Foundation, 1972), p. 324Google Scholar.

24 This possibility was recognized but not pursued in 1968: “Indeed, this irony is compounded when the role of the Vietnam issue is jointly taken into account. We have indicated above that the public was deeply impatient with the Johnson Administration, in part, because of the handling of the war. Blacks stood out as the major demographic grouping most exercised about the entanglement in Vietnam. They were more likely than whites to opine that the government should never have undertaken the military commitment there. They also were more likely to feel that American troops should be brought home immediately, a position not generally associated with the Johnson Administration. None the less. … Negro enthusiasm not only for Hubert Humphrey, but for Lyndon Jphnson as well, remained high to the very end. It seems quite evident that when black citizens were making decisions about their vote Vietnam attitudes paled into relative insignificance by contrast with attitudes toward progress on civil rights within the country …” Converse, Philip E. et al. “Continuity and Change in American Politics” (p. 1085)Google Scholar.

25 Converse, Philip, “The Nature of Belief Systems in Mass Publics,” in Ideology and Discontent, ed. Apter, David (Glencoe, Ill: Free Press, 1964), p. 245Google Scholar; note also that when Converse screened for “self-starting” issue concern, turnover correlations approached the level of stability shown for party identification!

26 Shapiro, Michael, “Rational Political Man: A Synthesis of Economic and Social-Psychological PerspectivesAmerican Political Science Review, 63 (12 1969), 11061119CrossRefGoogle Scholar.

27 Converse, Philip E., “Information Flow and the Stability of Partisan Attitudes,” in Elections and the Political Order, ed. Campbell, et al. , pp. 138139Google Scholar.

28 Campbell, et al. , The American Voter, pp. 102103Google Scholar.

29 Campbell, et al. , The American Voter, p. 279Google Scholar.

30 Campbell, , et al. , The American Voter, p. 121, emphasis addedGoogle Scholar.

31 Downs, , An Economic Theory of Democracy, p. 26Google Scholar. It must be noted that Downs is speaking of an “ideal-team” party system, which speaks clearly with one well-defined voice, a point which obviously has crucial implications for any attempt to apply his theory to the U.S.

32 These conclusions are elaborated later in this paper. See also the excellent discussion by Stokes, Donald, “Spatial Models of Party Competition,” in Elections and the Political Order, pp. 161179Google Scholar.

33 Campbell, Angus, Gurin, Gerald, and Miller, Warren, The Voter Decides (Evanston, Ill.: Row, Peterson, 1954). p. 107. Emphasis addedGoogle Scholar.

34 See for example Campbell, Angus and Cooper, H. C., Group Differences in Attitudes and Votes: A Study of the 1954 Congressional Election (SRC, 1956), p. 95Google Scholar, as cited in Natchez, Peter B., “Images of Voting: The Social Psychologists,” Public Policy 17 (Summer 1970) 564Google Scholar. This paper was aided by Natchez's stimulating analysis.

35 Campbell, et al. , The Voter Decides, pp. 9596Google Scholar.

36 Campbell, et al. , The American Voter, p. 430Google Scholar.

37 Ibid., p. 430.

38 Ibid., pp. 185–186.

39 SRC 1956-58-60 panel reported by Pierce, John C. and Rose, Douglas D., “Non-attitudes and American Public Opinion: The Examination of a Thesis,” APSR 68 (06 1974), 626650, esp. p. 632CrossRefGoogle Scholar.

40 Converse, , “The Concept of a Normal Vote,” p. 21Google Scholar.

41 Converse, , “Religion and Politics: 1960” p. 101Google Scholar in Elections and the Political Order. Turnout as a separate variable was ignored in the conception of the normal vote on the grounds that the “frequency of ‘dynamic’ nonvoting is negligible in high turnout presidential elections …” (Converse, , “The Concept of a Normal Vote,” p. 24, fn. 18.Google Scholar) Further, the actual relations between turnout and closeness to the normal vote estimate have been the opposite of those which were predicted. By SRC reckoning, the smaller the turnout, the closer to the normal vote. (Campbell, , “Voters and Elections,” p. 748Google Scholar, Converse, , “Stability and Change in 1960” in Elections and the Political Order. p. 91Google Scholar.) See also Richard A. Brody, “Change and Stability in Partisan Identification: a Note of Caution” (unpublished ms.).

42 Stokes, et al. , “Components of Electoral Decision,” p. 373Google Scholar.

43 Ibid., p. 375.

44 Converse, et al. , “Continuity and Change in American Politics. …” p. 1099(emphasisadded)Google Scholar.

45 An extremely stimulating account of the influence on Congress can be found in Mayhew, R., Congress: The Electoral Connection (New Haven: Yale University Press, 1974)Google Scholar.

46 While Cambridge Survey Research (CSR) conducted numerous statewide studies for McGovern for President Incorporated, this reference and all of the analysis presented in the third section of this paper are based on the two principal election studies. The first of these, CSR 159-147, was an 11,000 interview national survey which provided stratified statewide readings of 11 states. An underlying national sample of about 1,300 interviews was obtained, through the use of weighting. This survey was conducted in person the last week of June 1972. Sampling, interviewing, and data processing were all conducted using the professional standards commonly employed, although these standards are necessarily less rigorous than those employed by the Survey Research Center.

In the third week of September 1972, a panel of 1,000 respondents from the June survey were recontacted by telephone (labeled by CSR as survey 159-180). This panel survey constitutes the basis for most of the analysis presented in section three of this paper. The data collected by CS R for McGovern for President Incorporated remains through contractual arrangements and other agreements the joint property of the company and the client. The authors regret that because of the expressed wishes of the client, no copies of the data set can be distributed for the time being. Luckily the Survey Research Center data are available to test all the assertions made in this paper.

47 The context of public opinion within which the election is held must play a role in the instrumental investor's evaluation of the candidates as well. For example, when a voter's position on key issues could be described as Utopian (e.g., absolute equalization of income or nuclear war with Russia) he might very well not support a candidate with similar positions, but instead choose a candidate more likely to be able to deliver at least some of his important desired outputs. Of course, an instrumental voter in this situation who regarded the expected output of the alternative candidates to be trivial or negative, or who believed that he could affect the output of the political system in the long run by casting a protest vote in a particular election, would be behaving consistently with our model by voting for the Utopian candidate. A similar calculation would be appropriate where the agreement occurred on less Utopian issue positions.

48 We are indebted to Professor Joe Oppenheimer for pointing out a paper that lends weight to the idea that the instrumental investor uses competence measures rather than simply proximity measures. Bruner, Jerome S. and Korchin, Sheldon J. argued in “The Boss and the Vote: A Case Study in City Politics,” Public Opinion Quarterly, 10 (Spring 1964), 123Google Scholar, that many Boston voters chose to vote for James Michael Curley despite the fact that they felt closer to other candidates on the issues, because they believed that he could deliver more to them than the other candidates.

49 Trust, a generalized sense that a candidate has your interests at heart and is concerned about people like you, clearly is a second major dimension of candidate evaluation and consideration. The simultaneous relations between trust and competence will be explored in future work. See also, Benjamin I. Page, Choices and Echoes in Presidential Elections (forthcoming).

50 Downs, , An Economic Theory of Democracy, p. 107Google Scholar.

51 The authors of The Voter Decides recognized this about the 1952 election: “Eisenhower, because of his background and experience seems to have been perceived as a person peculiarly able to cope with the nation's international problems.” Campbell, et al. , The Voter Decides, p. 58Google Scholar.

52 The transcripts show that the Nixon campaign recognized this explicitly:

Haldeman: … So little is known about McGovern, you'll have a better chance of changing people's mind about him …

Haldeman: … to start with, you got 40 per cent of the people who will vote for you no matter what happens …

President: … I agree.

Haldeman: … and you got 40 per cent of the people who will vote against you no matter what happens, so you have got 20 per cent of the people left in the middle who may vote for you or may not and that 20 per cent is what you've gotta work on. His argument is that you're so well known, your pluses are clear, clear as well as your minuses; that getting one of those 20 who is an undecided type to vote for you on the basis of your positive points is much less likely than getting them to vote against McGovern by scaring them to death about McGovern; and that, that's the area that we ought to be playing.

Published transcript of a recording of a meeting between the President and H. R. Haldeman on June 23, 1972 from 2:20 to 2:45 PM.

53 So many demographic traits are good behavioral predictors that social scientists routinely scrutinize attitudinal data for the “relevant” demographic breaks; when a question occasionally does not fit into the usual pattern, it may even be taken as a scientific breakthrough: “The point is that in absence of the ordinary guidelines of group position that would allow the citizen to adopt the position of those like him or the guideline of group-relatedness that would allow the citizen to adopt a position he perceives as relevant for the groups with which he identifies, cognitive and affective relationship to the events themselves may structure attitudes.” From Verba, Sidney, Brody, Richard A., Parker, Edwin B., Nie, Norman H., Polsby, Nelson W., Ekman, Paul, and Black, Gordon S., “Public Opinion and the War in Vietnam,” American Political Science Review 61 (06 1967), 331CrossRefGoogle Scholar. There has been a change in attitudes toward the relevance of demographic variables in assessing candidates. In the 1960 election, Kennedy's Catholicism was considered an irrelevant, illegitimate topic by voting commentators. Yet by 1974, there was considerable serious discussion about whether an individual as personally wealthy as Nelson Rockefeller was qualified to be a vice-president. Should an Arab-American or a Jewish-American be President?

54 Stokes, Donald E. and Miller, Warren E., “Party Government and the Saliency of Congress,” in Elections and the Political Order, ed. Campbell, et al. , p. 290Google Scholar.

55 The strategies appropriate for voters under conditions of uncertainty have recently been the subject of some analysis. See, for example, Page, Benjamin I., “The Theory of Political Ambiguity,” American Political Science Review (forthcoming)Google Scholar. The concept of a risky candidate is related to—but distinct from—the concept of candidate competence. See Shepsle, Kenneth A., “The Strategy of Ambiguity: Uncertainty and Electoral Competition,” American Political Science Review 66 (06 1972), 555568CrossRefGoogle Scholar.

* The arguments in this section are, for the most part, distilled from recent work by Joe A. Oppenheimer and Norman Frohlich. See Joe A. Oppenheimer, “Relating Coalitions of Minorities to the Voters Paradox, or Putting the Fly in the Democratic Pie,” paper given at the 1972 meeting of the Southwest Political Science Association, San Antonio, Texas, March 30–April 1, 1972; Oppenheimer, Joe A., “Some Political Implications of Vote Trading and Voting Paradox: A Proof of Logical Equivalence: A Comment, “American Political Science Review 69 (09 1975)CrossRefGoogle Scholar; Frohlich, Norman and Oppenheimer, Joe A., Modern Political Economy (Englewood Cliffs, N.J.: Prentice Hall, forthcoming). Chapter 7Google Scholar.

56 Downs argued that under conditions of uncertainty “… many a voter finds party ideologies useful because they remove the necessity of his relating every issue to his own philosophy. Ideologies help him focus attention on the differences between the parties; therefore they can be used as samples of ail the differentiating stands. With this short cut a voter can save himself the cost of being informed upon a wider range of issues.” Downs, , An Economic Theory of Democracy, p. 98Google Scholar.

57 Converse, , “The Nature of Belief Systems in Mass Publics,” p. 227Google Scholar.

58 Stokes, “Spatial Models of Party Competition,”

59 Campbell, et al. , The American Voter, p. 550Google Scholar.

60 For the sake of simplicity, the authors will always assume two parties or two candidates in this analysis.

61 See, for example, Converse, “The Nature of Belief Systems in Mass Publics,” or Norman H. Nie with Andersen, Kristi, “Mass Belief Systems Revisited: Political Change and Attitude Structure,” The Journal of Politics, 36 (08 1974), 540591Google Scholar.

62 Of course, there is another model under which a single dimension, such as the pro-Democratic to pro-Republican continuum described in The Voter Decides, would arise. That is the “loyalist” model, where the electorate is incapable of ordering issues in any way which is not defined by the party but follows the “party line” slavishly.

63 Although our disagreement here is with the logical basis of the assumption that hawks for McCarthy were obviously and clearly misguided, this example has assumed so much importance that its empirical basis needs to be reviewed. When Phillip Converse talks of the widely noted paradox that “many voters across the county had moved from a ‘leftist’ McCarthy vote in the primaries to a ‘rightist’ Wallace vote in the general election” or when Converse talks of voters for whom McCarthy was the “wrong”candidate (the wrong McCarthy?) he is referring to the SRC's 1968 election analysis and its reference to “a major McCarthy to Wallace transfer.” The empirical basis offered for these statements is fourteen (14) voters. See Converse, et al. , “Continuity and Change in American Politics …”, pp. 10931095Google Scholar; Converse, Philip, “Public Opinion and Voting Behavior,” Handbook of Political Science, ed. Polsby, Nelson and Greenstein, Fred (Reading, Mass.: Addison Wesley, 1975), pp. 81, 100, 120, 124Google Scholar.

64 Cambridge Survey Research 159-147 showed 6.2 per cent of respondents volunteered this preference order in June 1972 when it was not offered to them by interviewers who presented only the common scale.

65 Downs, , An Economic Theory of Democracy, pp. 6062Google Scholar.

66 No justification is presented for the assumption, implied several times by the authors of “Majority Party.” that voters in “extreme” positions are most intense about issues, or that such voters “intensely favored,” “intensely preferred” their positions.

This is the familiar myth of the moderate centrist and the intense extremist. However, it is simply not the case that people in the middle of a scale are necessarily less intense about their position than are those on the extreme. Consider, for example, a scale with “no tenure” for university professors at one end, “status quo” in the center, and “student determination of tenure” at the other extreme. Would the people with the most intense feelings be the extremists? Miller, et al. , A Majority Party …,” (pp. 763, 764)Google Scholar.

67 See for example: Riker, William H. and Ordeshook, Peter C., An Introduction to Positive Political Theory (Englewood Cliffs, N.J.: Prentice-Hall, 1973), p. 107Google Scholar.

68 Hotelling, Harold, “Stability in Competition,” The Economic Journal 39 (1929), 4157CrossRefGoogle Scholar; and Smithies, Arthur, “Optimum Location in Spatial Competition,” The Journal of Political Economy, 49 (1941), 423439CrossRefGoogle Scholar.

69 Scammon, R. M. and Wattenberg, B. J., The Real Majority (New York: G. P. Putnam & Sons, 1970)Google Scholar.

70 Evidence of this is that unprecedented ticket-splitting took place. If one is to believe “Majority Party,” millions of people simultaneously rejected George McGovern's issue positions, then turned around to vote for candidates holding basically the same positions. The victory of Jim Abourezk in South Dakota as McGovern lost his home state and the victories of Dick Clark, Joe Biden, or Walter Mondale must raise some questions concerning pure proximity measures.

71 Other surveys in which we repeated the policy questions verify for us that McGovern was losing the “liberal” constituencies and not that the “liberal” constituencies were shrinking. The effects observed were clearly not regression to the mean or any other statistical artifact of the panel format. Nixon simply held his t 1 issue and demo-graphic constituencies in every single case far better than did McGovern.