¹ Campbell, Angus, Gurin, Gerald and Miller, Warren E., The Voter Decides (Evanston, III.: Row, Peterson, 1954).Google Scholar

² Milbrath, Lester and Goel, M. L., Political Participation, 2nd edn (Chicago: Rand McNally, 1977), pp. 57–74.Google Scholar

³ However, the proposition that participation bolsters efficacy has been a staple of critiques of liberal democracies and accompanying pleas for increased citizen participation. See Pateman, Carole, Participation and Democratic Theory (Cambridge: Cambridge University Press, 1970)CrossRefGoogle Scholar; Macpherson, C. B., The Life and Times of Liberal Democracy (Oxford: Oxford University Press, 1979).Google Scholar

⁴ Relevant studies include, inter alia, Welch, Susan and Clark, Cal, ‘Determinants of Change in Political Efficacy: A Test of Two Hypotheses’, Journal of Political and Military Sociology, 3 (1975), 207–17Google Scholar; Ginsberg, Benjamin and Weissberg, Robert, ‘Elections and the Mobilization of Popular Support’, American Journal of Political Science, 22 (1978), 31–55CrossRefGoogle Scholar; Finkel, Steven E., ‘Reciprocal Effects of Participation and Political Efficacy: A Panel Analysis’, American Journal of Political Science, 29 (1985), 891–913CrossRefGoogle Scholar; Finkel, Stephen E., ‘The Effects of Participation on Political Efficacy and Political Support’, Journal of Politics, 49 (1987), 441–64.CrossRefGoogle Scholar More generally, a substantial body of theory and research suggests that behaviour and attitudes have reciprocal causal linkages. See Bem, D. J., ‘Self Perception Theory’, in Berkowitz, L., ed., Advances in Experimental Social Psychology (New York: Academic Press, 1972)Google Scholar; Jones, E. E. et al. , Attribution: Perceiving the Causes of behavior (Morristown, NJ: General Learning Press, 1972)Google Scholar; Kelman, Herbert C., ‘Attitudes Are Alive and Well and Gainfully Employed in the Sphere of Action’, American Psychologist, 29 (1974), 310–24.CrossRefGoogle Scholar

⁵ Lane, Robert, Political Life (New York: The Free Press, 1959), p. 149.Google Scholar

⁶ See the works cited in Acock, Alan, Clarke, Harold D. and Stewart, Marianne C., ‘A New Model for Old Measures: A Covariance Structure Analysis of Political Efficacy’, Journal of Politics, 47 (1985), 1062–84.CrossRefGoogle Scholar

⁷ Lane, , Political Life, p. 149Google Scholar; Acock, , Clarke, and Stewart, , ‘A New Model’, pp. 1075–7Google Scholar; Acock, Alan and Clarke, Harold D., ‘Alternative Measures of Political Efficacy: Models and Means’, revised version of paper presented at the Annual Meeting of the American Political Science Association, Washington, DC, 09 1986.Google Scholar

⁸ Ginsberg, and Weissberg, , ‘Elections and the Mobization of Popular Support’, passim.Google Scholar

⁹ See, for example, Kelman, , ‘Attitudes are Alive and Well’, passim.Google Scholar

¹⁰ Lane, , Political Life, pp. 147–62Google Scholar; Milbrath, and Goel, , Political Participation, p. 61.Google Scholar

¹¹ Asher, Herbert, ‘The Reliability of the Political Efficacy Items’, Political Methodology, I (1974), 45–72.Google Scholar

¹² Sigelman, Lee, ‘The Nonvoting Voter in Voting Research’, American Journal of Political Science, 26 (1982), 47–56.CrossRefGoogle Scholar

¹³ Center for Political Studies, The American National Series: 1972, 1974 and 1976, Vol. 1 (Ann Arbor, Michigan: Inter-University Consortium for Political and Social Research, 1979), p. XLIV.Google Scholar

¹⁴ The N for the post-election wave is smaller than that for the pre-election wave – 969 v. 2257. This is a result of the CPS's decision to omit the efficacy items from an experimental telephone version of the post-election interview administered to a random half-sample of post-election respondents. See Center for Political Studies, American National Election Study, 1984, Vol. 1 (Ann Arbor, Michigan: Inter-University Consortium for Political and Social Research, 1986), pp. IV–V.Google Scholar

¹⁵ Some analysts use X2 to measure external efficacy but do not present validation analyses to substantiate this classification. See, for example, Miller, Warren E., Miller, Arthur H. and Schneider, Edward J., American National Election Studies Data Sourcebook, 1952–1978 (Cambridge, Mass.: Harvard University Press, 1980).Google Scholar Using the statement to tap internal efficacy reflects both face validity considerations, i.e., the reference to ‘people like me’, and the results of extensive validation tests. See Acock, and Clarke, , ‘Alternative Measures of Political Efficacy’.Google Scholar Re X3 and X4, these have been used in several CPS surveys to measure ‘governmental responsiveness’. Since there is broad agreement that such perceptions constitute the core meaning of external political efficacy, X3 and X4 would seem to be useful indicators of this concept.

¹⁶ Confirmatory factor analysis (CFA) differs from conventional exploratory factor analysis (EFA). CFA enables one to perform statistical tests of the goodness-of-fit of a measurement model in which the number of factors and patterns of factor loadings for the indicator variables are hypothesized a priori. Also, unique factors may be hypothesized to be inter-correlated, and correlations between common factors are parameter estimates rather than arbitrary values as, for example, with oblique rotations of exploratory factor solutions (Long, J. Scott, Confirmatory Factor Analysis: A Preface to LISREL (Beverly Hills, Calif.: Sage Publications, 1983)CrossRefGoogle Scholar; Joreskog, Karl J. and Sorbom, Dag, LISREL VI: User's Guide, 3rd ed (Mooresville, Ind.: Scientific Software Inc., 1984)).Google Scholar

¹⁷ This is in sharp contrast to a single-construct measurement model that does not distinguish between internal and external efficacy. For the pre- and post-election waves of the 1984 survey the single-construct model yields X ²₁s of 203.13, p < 0.001 and 126.99, p < 0.001, respectively.

¹⁸ Such comparisons are important for establishing the general applicability of a measurement model. The fit of a measurement model for various groups cannot be assumed, but must be established empirically, with groups being selected in terms of theoretical or substantive rationales relating to particular political contexts. In the United States, comparisons of racial, gender and educational groups are arguably important, although others would be appropriate as well.

¹⁹ The utility of personal competence and political trust in construct validity tests is substantiated by analyses of the CPS efficacy measures. See Acock, and Clarke, , ‘Alternative Measures of Political Efficacy’.Google Scholar

²⁰ The 106 respondents who claim to have voted and three who claimed not to have voted are eliminated. See Center for Political Studies, American National Election Study, 1984, 2nd ICPSR edn, Vol. 1 (Ann Arbor, Michigan: Inter-University Consortium for Political and Social Research, 1986), pp. 579–80.Google Scholar Regarding the properties of the resulting sample, men, blacks, 22–35 year olds, persons with university educations at the bachelor's level and weak Democrats are all slightly more likely than others not to have their self-reports substantiated. A probit analysis confirms the weakness of these correlates – R ² = 0.08, and only party identification is statistically significant – weak Democrats are less likely to be in the validated group (t = – 2.13). Also, there are only very small differences between the validated and non-validated voters' responses to the eight pre- and postelection efficacy items. A probit analysis of the validated vote screening variable using the eight efficacy variables as predictors produces an R ² = 0.03 and none of the predictors are statistically significant. This finding differs from those reported in an analysis of the 1964, 1976 and 1980 data. See Silver, Brian, Anderson, Barbara A. and Abramson, Paul R., ‘Who Overreports Voting’, American Political Science Review, 80 (1986), 613–24.CrossRefGoogle Scholar

²¹ Voting turnout and campaign activity constitute separate factors. In an initial CFA of the participation items a two-factor model fits the data reasonably well – X ²₅ = 12.86, p = 0.025, AGFI = 0.979. Modification indices indicated that the fit could be improved if 0_εs for two of the campaign activity items, ‘display button or sticker’ and ‘work for candidate or party’ were correlated. This produces an excellent fit – X ²₄ = 6.08, p = 0.193, AGFI = 0.988, and these 0_εs are correlated in all the analyses reported below.

²² Ehrenhalt, Alan, ed., Politics in America: Members of Congress in Washington and at Home (Washington, DC: Congressional Quarterly Press, 1985).Google Scholar

²³ To permit error terms for indicators in the pre- and post-election waves of interviewing to be correlated, the latent exogenous variables, i.e., pre-election internal efficacy, voting and campaign activity, are treated as endogenous variables in the LISREL specification. Accordingly, these constructs and the coefficients for the causal linkages between them and post-election internal and external efficacy are labelled as ηs and βs, respectively. Since pre-election internal and external efficacy, voting and campaign activity are conceptually exogenous, causal linkages between them are not specified. These variables are allowed to be correlated by freeing elements in the ψ matrix. Similarly, no causal linkage is specified between post-election internal and external efficacy and, accordingly, the appropriate element in this matrix is freed. Also, since the candidate support constructs in the LISREL models are measured as single indicators, their variance in the ψ matrix is fixed at the value of the variance of their indicators.

²⁴ Long, , Confirmatory Factor Analysis, p. 29.Google Scholar

²⁵ Including linkages between voting and campaign activity, on the one hand, and internal efficacy, on the other, does not effect the results i.e., X ²₅₂ = 69.01, p = 0.057, AGFI = 0.986. The voting-internal efficacy and campaign activity – internal efficacy coefficients are not statistically significant, and the significance of other coefficients in the model remains unaltered.

²⁶ Wolfinger, Raymond E. and Rosenstone, Steven J., Who Votes? (New Haven, Conn.: Yale University Press, 1980).Google Scholar