Hostname: page-component-76fb5796d-9pm4c Total loading time: 0 Render date: 2024-04-26T22:34:55.024Z Has data issue: false hasContentIssue false
  • English
  • Français

The Relative Importance of Socioeconomic and Political Variables for Public Policy*

Published online by Cambridge University Press:  01 August 2014

Michael S. Lewis-Beck
Michael S. Lewis-Beck*
Affiliation:
University of Iowa
Get access Rights & Permissions [Opens in a new window]

Abstract

Since Dawson and Robinson, a dominant issue in the quantitative study of public policy has been the relative importance of socioeconomic and political variables for determining policy outcomes. It is argued here that past efforts to resolve this issue have been unsatisfactory, largely because they relied on inadequate statistical techniques, i.e., simple correlation, partial correlation, or multiple regression. Coefficients from these techniques are irrelevant for all but the most peculiar models of public policy. In general, if the researcher wishes to assess the relative importance of independent variables, it will be necessary to resort to path analysis of a formally constructed causal model. The comparison of “effects coefficients,” derived from path analysis, is offered as the preferred means of evaluating independent variables, superior to comparisons of coefficients from simple correlation, partial correlation, or multiple regression. When the effects coefficients are actually calculated for a popular model of welfare policy, socioeconomic variables appear much more important than political variables, contrary to interpretations coming from the more traditional statistical techniques.

Type
Articles
Information
American Political Science Review , Volume 71 , Issue 2 , June 1977 , pp. 559 - 566
Copyright
Copyright © American Political Science Association 1977

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 especially to thank Donald J. McCrone and Lawrence B. Mohr for their contributions to the development of this paper.

References

1 Dawson, Richard E. and Robinson, James A., “Inter-party Competition, Economic Variables, and Welfare Policies in the American States,” Journal of Politics, 25 (May, 1963), 265–289 CrossRefGoogle Scholar.

2 Key, V. O. Jr., Southern Politics (New York: Vintage Books, 1949), p. 307 Google Scholar; Lockard, Duane, New England State Politics (Princeton, New Jersey: Princeton University Press, 1959), pp. 320–340 CrossRefGoogle Scholar.

3 Cnudde, Charles F. and McCrone, Donald J., “Party Competition and Welfare Policies in the American States,” American Political Science Review, 63 (September, 1969), 858–866 CrossRefGoogle Scholar; Dye, Thomas R., Politics, Economics, and the Public: Policy Outcomes in the American States (Chicago: Rand McNally, 1966)Google Scholar; Hofferbert, Richard I., “The Relation Between Public Policy and Some Structural and Environmental Variables in the American States,” American Political Science Review, 60 (March, 1966), 73–82 CrossRefGoogle Scholar; Sharkansky, Ira and Hofferbert, Richard I., “Dimensions of State Politics, Economics, and Public Policy,” American Political Science Review, 63 (September, 1969), 867–879 CrossRefGoogle Scholar.

4 Fry, Brian R. and Winters, Richard F., “The Politics of Redistribution,” American Political Science Review, 64 (June, 1970), 508–522 CrossRefGoogle Scholar.

5 Fry and Winters, p. 521.

6 Booms, Bernard H. and Halldorson, James R., “The Politics of Redistribution: A Reformulation,” American Political Science Review, 67 (September, 1973), 924–933 CrossRefGoogle Scholar.

7 Uslaner, Eric M. and Weber, Ronald E., “The ‘Politics’ of Redistribution: Toward a Model of the Policy-making Process in the American States,” American Politics Quarterly, 3 (April, 1975), 130–170 CrossRefGoogle Scholar.

8 Tompkins, Gary L., “A Causal Model of State Welfare Expenditures,” Journal of Politics, 37 (August, 1975), 392–416 CrossRefGoogle Scholar.

9 Duncan, Otis D., “Partials, Partitions, and Paths,” in Sociological Methodology 1970, ed. Borgatta, E. and Bohrnstedt, G. (San Francisco: Jossey-Bass, 1970), pp. 38–47 Google Scholar; Linn, R. L. and Wirts, C. E., “Assumptions in Making Causal Inferences from Part Correlations, Partial Correlations, and Partial Regression Coefficients,” Psychological Bulletin, 12 (November, 1969), 307–310 CrossRefGoogle Scholar.

10 Lewis-Beck, Michael S. and Mohr, Lawrence B., “Evaluating Effects of Independent Variables,” Political Methodology, 3 (February, 1976), 27–47 Google Scholar.

11 Dawson and Robinson.

12 Hofferbert.

13 Dawson and Robinson.

14 Cnudde and McCrone, p. 860; Dye, , Politics, Economics, and the Public, p. 285 Google Scholar; Lockard, , New England State Politics, pp. 336–337 Google Scholar.

15 Dawson and Robinson; Dye, Politics, Economics, and the Public.

16 Fry and Winters, p. 519.

17 Cnudde and McCrone.

18 Dawson and Robinson.

19 Dye, , Politics, Economics, and the Public, pp. 293–295 Google Scholar.

20 Sharkansky and Hofferbert, p. 877.

21 Linn and Werts, p. 308; Duncan, 1970, pp. 39–40.

22 If the partial correlation is calculated and found to be zero or not significant statistically, then the conclusion of spuriousness is legitimate, granting this causal structure. However, if the partial correlation is found to be statistically significant, then it is not necessarily proper to infer, as is frequently done, that the independent variable in question, e.g., political structure, does have an impact on policy. This caution is understandable when it is recalled that the partial correlation is merely a correlation among residuals, e.g., ru 2 u 3. Thus, even if this correlation is significant, it may “simply reflect some third variable other than socioeconomic conditions, e.g., geographic region, which is operating to produce spuriousness between political structure and public policy.

23 Dye, , Politics, Economics, and the Public, pp. 295–297 Google Scholar; Fry and Winters.

24 Fry and Winters, p. 521.

25 Cnudde and McCrone.

26 Blalock, H. M. Jr., “Causal Inferences, Closed Populations, and Measures of Association,” American Political Science Review, 61 (March, 1967), 130–136 CrossRefGoogle Scholar; Tukey, J. W., “Causation, Regression, and Path Analysis,” in Statistics and Mathematics in Biology, ed. Kempthorne, O., Bancroft, T. S., Gowen, J. W., and Lush, J. L. (Ames, Iowa: Iowa State College Press, 1954), pp. 35–66 Google Scholar; Turner, M. E. and Stevens, C. D., “The Regression Analysis of Causal Paths,” Biometrics, 15 (June, 1959), 236–258 CrossRefGoogle Scholar; Wright, Sewell, “Path Coefficients and Path Regression: Alternative or Complementary Concepts?” Biometrics, 16 (June, 1960), 189–202 CrossRefGoogle Scholar.

27 Boomsand Halldorson, p. 932.

28 Dye, , Politics, Economics, and the Public, p. 285 Google Scholar; Lockard, , New England State Politics, pp. 336–337 Google Scholar; Cnudde and McCrone, p. 860.

29 Duncan, 1970, p. 40.

30 Uslaner and Weber; Tompkins.

31 Uslaner and Weber, pp. 138, 156.

32 Cnudde and McCrone.

33 Wonnacott, Ronald I. and Wonnacott, Thomas H., Econometrics (New York: Wiley, 1970), pp. 193–195 Google Scholar.

34 Goldberger, Arthur S., “On Boudon's Method of Linear Causal Analysis,” American Sociological Review, 35 (February, 1970), 97–101 CrossRefGoogle Scholar; Land, Kenneth C., “Identification, Parameter Estimation, and Hypothesis Testing in Recursive Sociological Models,” in Structural Equation Models in the Social Sciences, ed. Goldberger, A. S. and Duncan, O. D. (New York: Seminar Press, 1973), pp. 19–49 Google Scholar.

35 See Lewis-Beck and Mohr, 1976, for a complete explication of the effects coefficient. Basically, the effects coefficient is an extension of earlier attempts to assess “total effects” in a causal system (for earlier treatments, see Alwin, Duane G. and Hauser, Robert M., “The Decomposition of Effects in Path Analysis,” American Sociological Review, 40 (February, 1975), 37–47 CrossRefGoogle Scholar; Duncan, Otis D., “Path Analysis: Sociological Examples,” in Causal Models in the Social Sciences, ed. Blalock, H. M. (Chicago: Aldine, 1971), pp. 137–138 Google Scholar; Finney, John M., “Indirect Effects in Path Analysis,” Sociological Methods and Research, 2 (November, 1972), 175–186 CrossRefGoogle Scholar; Land, Kenneth C., “Principles of Path Analysis,” in Sociological Methodology 1969, ed. Bongatta, E. F. (San Francisco: Jossey-Bass, 1969), pp. 16–17 Google Scholar; Lewis-Beck, Michael S., “Determining the Importance of an Independent Variable: A Path Analytic Solution,” Social Science Research, 3 (June, 1974), 95–107)CrossRefGoogle Scholar. However, the effects coefficient is more satisfactory than prior efforts for a number of reasons. First, it has generalized applicability. That is, it may be used to assess the impact of any independent variable in the system, endogenous as well as exogenous (on this distinction, see Wonnacott and Wonnacott, pp. 155–156). And, it is applicable to any linear additive causal structure, whether it be recursive or nonrecursive, just-identified or overidentified (on these differences, see Wonnacott and Wonnacott, pp. 193–195, 172–189). (Of course, for nonrecursive systems, estimation techniques such as twostage least squares must be used, rather than ordinary least squares; see Johnston, J., Econometric Methods (New York: McGraw-Hill, 1972), pp. 380–384)Google Scholar. Further, the effects coefficient is based on a more precise and comprehensive breakdown of relationships in the causal system, dividing the possible relations between two variables into direct effect (DE), indirect effect (IE), spurious relation (S) and unanalyzed relation (U).

36 Sharkansky and Hofferbert, pp. 876–877.

37 Sharkansky and Hofferbert, p. 877.

38 When totally different causal structures underlie the statistics employed for evaluation, comparable results cannot reasonably be expected. In an analysis of two distinct data sets, not only were tlie distances between the effects of variables altered considerably depending on whether simple correlation, partial correlation, standardized partial regression, or effects coefficients were used, but there was also an occasional discrepancy in sign, and tlie rank ordering of the variables in terms of their effects was changed in a great many instances (Michael S. Lewis-Beck and Lawrence Mohr, B., “Evaluating Effects of Independent Variables: A Path Analytic Approach,” Institute of Public Policy Studies (Ann Arbor: The University of Michigan, Discussion Paper #59)Google Scholar.