The Consequences of Validated and Self-Reported Voting Measures

This paper reports on the results of validation of the self-reportedregistration status and voting behavior of respondents in the1976 and 1978 American National Election Studies. The resultsindicate about one in seven of the respondents misreported theirregistration status or voting behavior. Comparative analysesare conducted using simple regression models to see if differencesin their explanatory power arise using validated and self-reporteddependent variables. The results show that there are no majorchanges in the fundamental nature of basic relationships thathave been observed since the first surveys were conducted. Analysisof the effects of overreported participation on estimates ofthe partisan division of the vote in three sets of subnationalcontests reveals a likely "bandwagan" effect. . Portions of the data utilized in this article were made availableby the Inter-university Consortium for Political and SocialResearch. The data for the 1976 and 1978 American National ElectionStudies were originally collected by the Center for PoliticalStudies of the Institute for Social Research. The Universityof Michigan, under a grant from the National Science Foundation.Neither the original collectors of the data nor the Consortiumbear any responsibility for the analyses or interpretationspresented here. The comments of an anonymous reviewer, whichresulted in a fruitful extension of the analysis, are gratefullyacknowledged.     

Response Validity in Surveysof Voting Behavior

This paper presents the results of a project which validatedthe reported registration and voting behavior of respondentsin a national election study. The accuracy of reported votingbehavior in the 1976 general election is assessed in terms ofthe demographic characteristics of the respondents to the Centerfor Political Studies National Election Study as well as theextent of their participation in a survey panel begun in 1972.Increased levels of registration and turnout are observed inassociation with the number of interviews in which respondentsparticipated, and three alternative social psychological modelsof the effects of preelection interviews are evaluated. Althoughthe interview apparently served as a stimulus to voting, neithera model associated with self-concept theory nor alienation theoryappears to explain the phenomenon adequately. The intervieweffect is significant and appears to be cumulative, indicatingthat researchers usingthe survey method with panel designs shouldbe sitive to the effects of their method on the behavior whichthey are tryingto measure.