RATIONALIZATION AND DERIVATION PROCESSES IN SURVEY STUDIES OF POLITICAL CANDIDATE EVALUATION

In order to assess the determinants of voters' candidate preferences, some analysts have examined responses to open-ended questions that ask citizens what might make them vote for or against a particular candid ate. However, psychological theory and research suggest that the succe ss of these reports in predicting voting may be because they reflect r ationalizations of preferences rather than the reasons that give rise to them. And indeed, using data from a panel survey conducted during t he 1990 elections in Ohio, we found that voters' reports of the reason s for their preferences were principally rationalizations. Rationaliza tion was especially strong among politically involved voters and those with little exposure to the media. Derivation of preferences from lik es and dislikes was most pronounced among voters who made up their min ds late in the campaign. These findings support the on-line model of v oter decision making and suggest that open-ended questions asking vote rs about their likes and dislikes are not well suited to revealing the real reasons for their preferences.