Hostname: page-component-7479d7b7d-rvbq7 Total loading time: 0 Render date: 2024-07-09T14:20:11.946Z Has data issue: false hasContentIssue false
  • English
  • Français

A Note on Redistributive Politics*

Published online by Cambridge University Press:  01 August 2014

John L. Sullivan
John L. Sullivan*
Affiliation:
Indiana University
Get access Rights & Permissions [Opens in a new window]

Abstract

Fry and Winters, in a recent article in this journal, concluded that political variables are more important than socioeconomic variables in terms of redistributive policies. They based their analysis, however, on twelve political but only six socioeconomic variables. This research note re-examines these relationships, utilizing twelve political and twelve socioeconomic variables. The findings are strikingly reversed, whether one considers all twenty-four variables or the best five political and the best five socioeconomic variables. However, these findings reflect a shotgun approach, simply more and more variables added to a regression equation. To reduce and clarify the analysis, two criteria are suggested for selecting independent variables: the size of the zero-order correlations, and the degree of multicolinearity among the independent variables. When three political and three socioeconomic variables are compared using these criteria, the results are once again inconsistent with those reported by Fry and Winters.

Type
Articles
Information
American Political Science Review , Volume 66 , Issue 4 , December 1972 , pp. 1301 - 1305
Copyright
Copyright © American Political Science Association 1972

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.)

Footnotes

*

I would like to thank Yale University and NIMH for aid in the form of a postdoctoral fellowship.

References

1 Fry, Brian R. and Winters, Richard F., “The Politics of Redistribution,” American Political Science Review, 64 (06, 1970), 521CrossRefGoogle Scholar.

2 Taken from Congressional District Data Book, 1963.

3 Hofferbert, Richard, “Socioeconomic Dimensions of the American States: 1890–1960,” Midwest Journal of Political Science, 12 (1968), 410–413CrossRefGoogle Scholar.

4 It is Aw, as developed by Lieberson. See Lieberson, Stanley, “Measuring Population Diversity,” American Sociological Review, 34 (12, 1969), 850–862CrossRefGoogle Scholar.

5 Sullivan, John L., “Political Correlates of Social, Economic, and Religious Diversity in the American States,” Journal of PoliticsGoogle Scholar, forthcoming.

6 Rae, Douglas W. and Taylor, Michael, The Analysis of Political Cleavages (New Haven, Conn.: Yale University Press, 1970), pp. 6–8Google Scholar.

7 U.S. Department of Commerce, Bureau of the Census, Statistical Abstract of the United States, 1961 (Washington, D.C.: U.S. Government Printing Office, 1961), p. 518 and p. 40Google Scholar.

8 Dye, Thomas, “Inequality and Civil-Rights Policy in the States,” Journal of Politics, 31 (11, 1969), 1082–1083CrossRefGoogle Scholar.

9 Fry and Winters, p. 522.

10 See Communications, American Political Science Review, 64 (12, 1970), 1249–1251CrossRefGoogle Scholar.

11 See, among others, Blalock, Hubert M. Jr., “Correlated Independent Variables: The Problem of Multicolinearity,” Social Forces, 42 (12, 1963), 233–237CrossRefGoogle Scholar.

12 The size of the within-set correlations is not relevant unless we attempt to compare the effects of within-set variables. Within-set multicolinearity will, however, affect the redundancy of the effects of that set on the dependent variable.

13 Traditionally, econometricians have established the criterion that correlations of .80 and above indicate “harmful multicolinearity.” This criterion can be misleading, for it has been suggested that it is not simply the size of the correlations among the independent variables that matters, but their size relative to the multiple correlation of independent variables with the dependent variable. Therefore multicolinearity is considered “harmful” if rij ≥ Ry, where rij represents the correlation between two independent variables, and Ry stands for the multiple correlation between the independent variables and the dependent variable. See Klein, Lawrence R., An Introduction to Econometrics (Englewood Cliffs, N.J.: Prentice-Hall, 1962)Google Scholar. In the present case, the multiple correlation between these six independent variables (the three political variables and the three socioeconomic variables with the lowest between-set correlations, Table 3) and the dependent variable is .66. It can be seen in Table 3 that none of the nine correlations between these three political and three socioeconomic variables exceeds this limit.

14 It can be noted, however, that random measurement error in the dependent variable does not attenuate the slope coefficients in a regression equation, whereas random measurement error in the independent variables does attenuate them. Therefore, measurement of the independent variables is, in fact, more important in regression than is measurement of the dependent variable, although careful measurement of both kinds of variables is a desirable goal. If we can assume accurate measurement of the independent variables discussed in this paper and can further assume that the measurement error in the redistributive ratio is random, the regressions are accurate. See Johnston, John, Econometric Methods (New York: McGraw Hill, 1963)Google Scholar.

Submit a response

Comments

No Comments have been published for this article.