The Voter in the 1982 House Elections

Abstract

Comprehensive micro-level models of voter behavior in House elections first became possible with the release of the 1978 National Election Study of the Center for Political Studies. This study generated an enormous volume of scholarly work. The 1980 National Election Study caused less of a stir, although reports of some findings are beginning to be published. Prudence demands that findings from earlier analyses be replicated with later data, both because sampling error can be expected to produce a few pseudo findings in any single study and because voters may behave differently as the national political context changes. Thus this research update uses the 1982 CPS data to estimate two models of voting behavior similar to those we used in our analysis of the 1978 and 1980 elections (Johannes and McAdams, 1981; McAdams and Johannes, 1983). Table 1 provides probit maximum likelihood estimates for each model. The first equation contains dummy variables for each of the seven CPS incumbent contact items, each reflecting a way in which the House member directly or indirectly reached out to the voter. Constituency service is reflected by INFORMATION and CASEWORK, dummies for having received from the incumbent's office either (1) information or (2) help with a problem the constituent was having with the government. Policy-oriented voting is reflected in ISSUES, index of the liberal or conservative policy preferences of voters, and IDEOLOGY, an index of ideological estrangement of the voter from the House voting record of the incumbent. ' Retrospective voting is assessed by variables measuring the respondent's assessment of his or her personal finances and by the respondent's perceptions that the national economy had improved or worsened. Evaluations of Reagan's performance (derived from an average of nine approve-disapprove items), partisanship, and the natural log of challenger spending are also included.