Hostname: page-component-cd9895bd7-jkksz Total loading time: 0 Render date: 2024-12-28T16:46:33.118Z Has data issue: false hasContentIssue false
  • English
  • Français

Additive and Multiplicative Models of the Voting Universe: The Case of Pennsylvania: 1960–1968

Published online by Cambridge University Press:  01 August 2014

Walter Dean Burnham  and
John Sprague
Walter Dean Burnham
Affiliation:
Washington University, St. Louis
John Sprague
Affiliation:
Washington University, St. Louis
Get access Rights & Permissions [Opens in a new window]

Extract

Of all the fields of political science where quantitative methods have been developed over the past generation, probably the one where scholarly understanding has been most enriched has been that of mass voting behavior. But while we know vastly more about this behavior on the individual and aggregate level than we did a quarter-century ago, there are still large territories on the map which are blank, or in which exploration has only very recently begun. There remain a number of doubtful areas in which issues of methodology and of substantive interpretation are still very much open to systematic inquiry.

One such area is that associated with the interrelation of socio-economic correlates of the vote. That is, there is a real question as to whether such independent or predisposing variables should be conceptualized as making mutually independent or, alternatively, interdependent contributions to the prediction of voting patterns. The normal practice in research involving multiple correlation of aggregate voting behavior with a set of independent variables has been to assume implicitly that the relationship of these variables is additive (i.e., non-interactive) and that the appropriate theoretical representation is of the general form y = b + m1x1 + m2x2 … + mnxn. Such an assumption seems plausible so far as individual voting for American major parties and their candidates is concerned. Thus, for example, the authors of the MIT 1960 simulation study found strong evidence that predispositional factors summate, i.e., are indeed additive in character.

Type
Research Article
Information
American Political Science Review , Volume 64 , Issue 2 , June 1970 , pp. 471 - 490
Copyright
Copyright © American Political Science Association 1970

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 Multiple correlations of predisposing variables over time, of course, are nothing new in voting-behavior studies. See as the locus classicus Gosnell, Harold F., Machine Politics: Chicago Model (Chicago: University of Chicago Press, 1937)Google Scholar; and also his Grass-Roots Politics (Washington: Public Affairs Press, 1942)Google Scholar.

2 Pool, Ithiel de Solaet al., Candidates, Issues and Strategies (Cambridge; M.I.T. Press, 1965), pp. 127–143Google Scholar and especially Table 4.2.

3 See Alker, Hayward Jr., Mathematics and Politics (New York: Macmillan, 1965), pp. 108–111Google Scholar and Blalock, Hubert M. Jr., Causal Inferences in Non-Experimental Research (Chapel Hill: University of North Carolina Press, 1961), pp. 91–93Google Scholar.

4 Soares, Glaucio and Hamblin, Robert, “Socio-Economic Variables and Voting for the Radical Left: Chile, 1952,” this Review, LXI (1967), 1053–1065Google Scholar.

5 Scheuch, E. K., “Cross-National Comparisons Using Aggregate Data,” in Merritt, Richard L. and Rokkan, Stein (eds.), Comparing Nations (New Haven: Yale University Press, 1966), pp. 131–167Google Scholar; see also Diderich, Nils, Empirische Wahlforschung (Köln, 1965), pp. 16–60CrossRefGoogle Scholar.

6 Rokkan, Stein and Valen, Henry, “Regional Contrasts in Norwegian Politics,” in Allardt, Erik and Littunen, Yrjo (eds.), Cleavages, Ideologies and Party Systems (Helsinki, 1964), pp. 162–238Google Scholar; Valen, Henry and Katz, Daniel, Political Parties in Norway (Oslo, 1964), especially pp. 120–186Google Scholar.

7 Burnham, Walter Dean, “American Voting Behavior and the 1964 Election,” Midwest Journal of Political Science, XII (1968), 1–40CrossRefGoogle Scholar.

8 Tingsten, H. L. A., Political Behavior (Stockholm, 1937), pp. 177–180Google Scholar.

9 Berelson, Bernardet al., Voting (Chicago: University of Chicago Press, 1954), pp. 98–106Google Scholar. See also Miller, Warren E., “One-Party Politics and the Voter,” THIS Review, L (1956), 707Google Scholar.

10 We selected here the 1960 county percentages of income receivers earning less than $3,000 per year and of rural-farm population of the county total. Bureau of the Census, County and City Data Book, 1962 ed.

11 A succinct account of the two-variable case is found in Croxton, F. E. and Cowden, D. J., Applied General Statistics, 2nd ed. (Englewood Cliffs: Prentice-Hall, 1955), pp. 508–512Google Scholar.

12 The issue is whether or not variances are changed proportionately in all variables. Certainly no data will exhibit precise proportionality. What we do in the text below is exhibit a test procedure which assesses the consequences of the error that is (inevitably) introduced by transformation. The test is basically empirical. Paul G. Hoel asserts that these consequences are typically of small magnitude. See Hoel, , Introduction to Mathematical Statistics, 3rd ed. (New York: Wiley, 1962), p. 178Google Scholar. On the other hand, transforming variables to meet distribution assumptions is a common practice engaged in for its statistical benefits. See for example, Johnston, J., Econometric Methods (New York: McGraw-Hill, 1963), especially pp. 44–45Google Scholar.

13 Professor Hamblin has assured us that he does not always, or even usually, obtain dramatic results by resort to a multiplicative mode. Personal communication from R. Hamblin.

14 We report our results only in qualitative terms. The data were from the 1964 S. R. C. election study (courtesy of the Inter-University Consortium for Political Research).

15 As throughout this study, the Democratic percentage is of the three-party vote in 1968, and of the two-party vote in 1960 and 1964. As to the predictor variables, all are taken from the 1960 census, except for percentage Democratic of two-party registration. This is also, however, derived from a 1960 base.

16 Burnham, “American Voting Behavior and the 1964 Election,” op. cit.

17 There is, of course, a longer historical dimension to the story which cannot receive very explicit treatment here. The 1964 variance of 59.51, for example, is the lowest in the electoral history of the state since 1940, although the Eisenhower landslide of 1956 produces an almost comparably low figure of 72.69. All others in the period show variances roughly the same as those of 1960 or 1968. Similarly, if one extends autocorrelations of contiguous pairs of elections back to, say, 1916/1920, evidences of massive realignment emerge between 1920 and 1940 which have no counterpart in the data within Pennsylvania for the 1960's. Thus, the squared coefficients (based on the Democratic percentage of the two-party vote except for 1924) are as follows during this period: 1916/1920: .76; 1920/1924: .77; 1924/1928: .00; 1928/1932: .46; 1932/1936: .61; 1936/1940: .91. Thereafter, except for 1956/1960 and 1960/1964, all pairs yield a squared coefficient of .91 or above. For a discussion of the changing socio-economic and demographic correlations of the Pennsylvania vote during this period, see Burnham, Walter Dean, Critical Elections and the Future of American Politics (New York: Norton, 1970)Google Scholar. The appendix reports two tables relevant to these considerations.

18 It may, unfortunately, be partly an artifact of census data which, while fitting the 1960 election closely, has become more or less obsolescent by 1968. It is more probable than not that the basic relationships among Pennsylvania's counties have not undergone sweeping changes since 1960, but only intercensal interpolation and analysis of 1970 data, once it is available, can settle this question.

19 Patterns for major-party voting in 1964 and 1968 are similar to this.

20 See, e.g., Converse, Philip E., “Religion and Politics: The 1960 Election,” in Campbell, Anguset al., Elections and the Political Order (New York: Wiley, 1966), pp. 96–124Google Scholar.

21 The 1966 Democratic gubernatorial candidate received about 6% of the black vote in Baltimore, running as he did on an explicitly closed-housing platform. This contrasts with an approximately 97% Democratic of the two-party vote in 1964 and an approximately 93% Democratic of the three-party vote in 1968 in Negro precincts of Baltimore City. One of the ironies of contemporary American politics is, thus, that Negroes played a decisive role in the chain of events which was to elevate Spiro T. Agnew to the Vice-Presidency of the United States.

22 For the nineteenth century, see Benson, Lee, The Concept of Jacksonian Democracy (Princeton: Princeton University Press, 1961), pp. 165–207 and 270–287Google Scholar.

23 See, e.g., Philip E. Converse, “The Concept of a Normal Vote,” in Angus Campbell et al., op. cit., pp. 9–39.

24 Wolfinger, Raymond, “Some Consequences of Ethnic Voting,” in Jennings, M. Kent and Ziegler, Harmon (eds.), The Electoral Process (Englewood Cliffs: Prentice-Hall, 1966)Google Scholar.

25 Key, V. O. Jr., and Munger, Frank, “Social Determinism and Electoral Decision: The Case of Indiana,” reprinted in Munger, Frank and Price, H. Douglas (eds.), Readings in Political Parties and Pressure Groups (New York: Crowell, 1964), pp. 366–384Google Scholar.

26 Research which has just been completed by several of our graduate students but not yet verified suggests the existence of similar interactivity in the Wallace vote in Georgia (43.4% of the total) and also in voting in Quebec for Real Caouette's ressentiment-oriented Social Credit movement in 1962. In the latter case, the vote for Progressive Conservative candidates was similarly analyzed as a “control”; the additive form of multiple correlation apparently had slightly greater explanatory power so far as Quebec voting for this Canadian major party was concerned.

27 This table's fourth column includes the following Pennsylvania urban-metropolitan counties: Allegheny, Beaver, Berks, Bucks, Cambria, Cumberland, Dauphin, Delaware, Erie, Lackawanna, Lancaster, Lehigh, Luzerne, Montgomery, Northampton, Philadelphia, Washington, Westmoreland and York. The other three columns are based on minor-civil-decision data for the countries concerned.

28 If one executes a simple linear correlation program with two independent variables—percentage professional-managerial and percentage skilled and semi-skilled in the 1960 male labor force—it can at once be seen that (a) very significant differences exist in the regression equations, the partials, the beta weights and the R2's for four New Jersey counties at this level; and (b) an aggregation of the minor-civil division units of these four counties, taken together, yields an R2 which accounts for only three-fifths of the variance. The arrays look like this:

29 The Wallace vote in these metropolitan counties ranges from 3.1%, 4.1%, 4.2%, and 4.8% in Lackawanna, Lehigh, Northampton and Luzerne counties, respectively, to 9.8%, 10.6%, 10.7% and 115% in Beaver, Washington, Bucks and Allegheny counties, respectively.

Thus, while it is clear that a positive relationship exists between the magnitude of the Wallace vote and the percentage of blue-collar workers in New Jersey minor-civil-divisions, there is considerable variability in this relationship between the divisions in one county and another. The amount of variance in the Wallace vote which is accounted for by these two variables ranges from leas than one-third to more than four-fifths.

30 The Wallace percentages in these cities were: Fall Eiver, 2.9%; Lawrence, 3.9%; New Bedford, 4.1%; Revere, 4.8%; Somerville, 4.3%. Impressionistically, it appears that the same general pattern of relatively greater Wallace penetration in blue-collar areas than in middle-class areas obtains in Massachusetts as in Pennsylvania and New Jersey. But both the intercepts and the slopes probably vary considerably from those for either Delaware or Luzerne counties.

31 For fuller elaboration of this argument, see Walter Dean Burnham, Critical Elections and the Future of American Politics, op. cit., Chapter 7.

32 Lowi, Theodore J., “Party, Policy and Constitution,” in Chambers, William N. and Burnham, Walter Dean (eds.), The American Party Systems (New York: Oxford, 1967), p. 263Google Scholar.

33 Lipset, S. M. and Rokkan, S., Party Systems and Voter Alignments (New York: The Free Press, 1967), pp. 1–64Google Scholar.

34 Lowi, op. cit.

35 Partitioning of the country between the “Greater South” (the eleven ex-Confederate states plus Oklahoma and Kentucky) and the remaining parts of the country reveals the following patterns in major third-party years since 1892:

Not only does the Wallace movement represent the largest Southern third-party insurrection since 1860; it is surpassed in the non-Southern states since that date only by the 1912 and 1924 Progressives.

36 The partial correlation of % Wallace with % Socialist in 1914 is +.316, while it is +.618 between % Wallace and mean % Democratic, 1932–1948. The latter two correlate at +531. The relative significance of the former in correlations with the Wallace vote is clearly very small compared with the latter: b1 = −.017, while b2 = +.627. R2 for both predictors = .382. Neither the partials nor the R2 are quite as large as might have been supposed a priori. This corresponds to the fact that the Wallace vote was much more uniformly distributed across the state's 77 counties than were either the 1914 Socialist vote or the 1932–1948 Democratic vote. Thus, the variance for the Wallace vote was 43.52 and the standard deviation 6.60; for the 1914 Socialist vote (with a mean almost identical with the Wallace mean in 1968) the variance was 97.73 and the standard deviation 959; and for the 1932–1948 mean Democratic percentage the two figures were 119.69 and 10.94 respectively. The data for the latter measure were taken from Benson, Oliveret al., Oklahoma Votes (Norman, Oklahoma: Bureau of Government Research, University of Oklahoma, 1964), p. 18Google Scholar.

37 See, e.g., Schattschneider, E. E., The Semi-Sovereign People (New York: Holt, Rinehart & Winston, 1960), pp. 97–113Google Scholar; Burnham, Walter Dean, “The Changing Shape of the American Political Universe,” this Review, LIX (1965), 7–28Google Scholar.

38 Johnson, Chalmers, Revolutionary Change (Boston: Little, Brown, 1966), pp. 27–33Google Scholar.

39 Liepelt, Klaus and Mitscherlich, Alexander, Thesen zur Wahlerfluktuation (Frankfurt a/M, 1968), pp. 32–36Google Scholar.

40 MacRae, Duncan Jr., and Meldrum, James A., “Critical Elections in Illinois: 1888–1958,” this Review, LIV (1960), 669–683Google Scholar.

Submit a response

Comments

No Comments have been published for this article.