¹ Shanks, J. Merrill and Miller, Warren E., ‘Policy Direction and Performance Evaluation: Complementary Interpretations of the Reagan Elections’, British Journal of Political Science, 20 (1990), 143–235CrossRefGoogle Scholar. For the first essay in this series, see Miller, Warren E. and Shanks, J. Merrill, ‘Policy Decisions and Presidential Leadership: Alternative Interpretations of the 1980 Presidential Election’, British Journal of Political Science, 12 (1982), 299–355.CrossRefGoogle Scholar

² Weisberg, Herbert, ‘Some Perspectives on the 1988 Presidential Election’ (paper presented at the 1989 meeting of the American Political Science Association, Atlanta, Ga.).Google Scholar

³ Shanks, and Miller, , ‘Policy Direction and Performance Evaluation’, pp. 225–7.Google Scholar

⁴ For a contrasting interpretation of the 1988 NES data, see Kinder, Donald P., Mendelberg, Tali, Dawson, Michael et al. , ‘Race and the 1988 American Presidential Election’ (paper presented at the 1989 meeting of the American Political Science Association, Atlanta, Ga.).Google Scholar

⁵ Shanks, and Miller, , ‘Policy Direction and Performance Evaluation’, see pp. 149–56, 169–79, 210–17.Google Scholar

⁶ The following works have had the greatest influence on our continuing use of this key distinction: Stokes, Donald E., Campbell, Angus and Miller, Warren E., ‘Components of Electoral Decisions’, American Political Science Review, 52 (1958), 367–87CrossRefGoogle Scholar; Miller, Warren E. and Levitin, Teresa E., Leadership and Change: Presidential Elections 1952 to 1976 (Cambridge, Mass.: Winthrop Publishers, 1978)Google Scholar; and Achen, Chris, Interpreting and Using Regression (Beverley Hills, Calif.: Sage, 1982).CrossRefGoogle Scholar

⁷ To provide a brief glimpse of recent political history we shall present data for voters in the five most recent presidential elections. The definition of voters – as opposed to non-voters – in 1980, 1984 and 1988 is provided by a combination of the respondents' self-report of voting or not voting and the NES staff's determination that such reports are valid or invalid. The validation of respondents' report of voting is accomplished through a detailed on-site inspection of local official administrative records. (For 1972 and 1976, our measure of turnout relies entirely on respondents' self-reports.) From one reasonable analytical perspective, one could agree that we should emphasize inter-election variations in the extent to which individuals with given attributes are more or less likely to vote. In this phase of our analysis we have chosen, instead, to compare the active electorates of different years, deliberately begging the question of whether voters and non-voters change attributes rather than given attributes being variously associated with voting or non-voting.

⁸ Black, Earl and Black, Merle, Politics and Society in the South (Cambridge, Mass.: Harvard University Press, 1987), Ch. 11, p. 237Google Scholar, Table 11.1.

⁹ Miller, Warren, ‘Party Identification and the Reagan Legacy’, Electoral Studies, 5 (1986), 101–21.CrossRefGoogle Scholar

¹⁰ Miller, , ‘Party Identification and the Reagan Legacy’.Google Scholar

¹¹ As a momentary aside, it is worth noting that neither the overall changes in the partisan and ideological predispositions of the non-voters, nor their more detailed patterns of possibly interrelated change, match what we have seen among voters. For example, between 1984 and 1988, non-voters became marginally more pro-Democratic in their party identifications and less conservative in their ideology.

¹² For two of the most influential simultaneous equation models, see Jackson, John E., ‘Issues, Party Choices, and Presidential Votes’, American Journal of Political Science, 19 (1975), 161–85CrossRefGoogle Scholar, and Fiorina, Morris P., Retrospective Voting in American National Elections (New Haven, Conn.: Yale University Press, 1981)Google Scholar. Some of our uncertainties about the usefulness of their perspectives are elaborated on pp. 162–3 below, but a complete discussion of those issues is beyond the scope of the current article.

¹³ See Green, Donald and Palmquist, Bradley, ‘Of Artifacts and Partisan Instability’, American Journal of Political Science, 34 (1990), 872–901CrossRefGoogle Scholar, for a comprehensive discussion of the ‘endogeneity’ and measurement error associated with partisan identification.

¹⁴ Since we regard all causes of partisanship as potential causes of the vote through some other mediating factor, we have not been able to remove any remaining biases in our estimate of the total effect of partisanship by moving to a two-stage analysis which uses a ‘predicted’ instead of the measured version of partisanship based on the reduced form equation for that variable. To be sure, any relationship between party identification and other omitted causes of the vote will produce an upward bias in our estimate of the total effect for party identification, just as measurement error in party identification will produce an attenuated estimate of that coefficient. Two-stage procedures are not available to remove those biases, however, if all the causes of partisanship must also be included in the structural equation for the vote.

¹⁵ While it can be correctly inferred in Table 14 that our measure of policy-related predispositions concerning racial discrimination and race itself were clear antecedents and partial explanations of attitudes towards the death penalty, it is interesting to note that those variables appeared to play only a minor role in affecting evaluations of the candidate's effectiveness in dealing with crime.

¹⁶ Shanks, and Miller, , ‘Policy Direction and Performance Evaluation’, pp. 212–14.Google Scholar

¹⁷ Such values are identical to those produced by first constructing the residual for each of our explanatory variables – i.e., by constructing a variable that contains only that portion of the variation of each variable not explained by the variables which preceded it – and then calculating the standardized regression coefficient for that variable when it is included along with all the variables at the same and previous stages.

¹⁸ Shanks, and Miller, , ‘Policy Direction and Performance Evaluation’, pp. 212–17.Google Scholar