Research Note: Partisan Priming and Subjective Evaluations of the Economy: Evidence from a Survey Experiment

Abstract

Issues and controversies connected to problems of endogeneity plague many topics of interest in political science, perhaps none more so than in the field of economic voting where in recent years a lively debate has developed over the potential endogeneity of subjective economic evaluations to partisan preferences. Although a great deal of attention has focused on dealing with these problems at the analysis stage rather less attention has been paid to dealing with the problem at source – that is at the measurement stage. In this study we use a question order experiment to assess whether partisan priming influences subjective evaluations of the economy. If it does then endogeneity bias might be reduced by making questions easier for people to comprehend and answer and by taking steps to minimize the need for information shortcuts.

Acknowledgements

We are grateful to Bobby Duffy and the team at Ipsos MORI for their support with data collection. We also thank Michael Lewis-Beck for constructive comments on earlier drafts. A preliminary version of this article was presented at the EPOP annual conference at the University of Edinburgh from 12 to 14 September 2014. We thank participants for their helpful comments and feedback.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1. In order to have confidence in the internal validity of our experiment, it is important that the questions that precede our experiment are not politically sensitive, and so do not act as an inadvertent prime. The only questions that came before our experiment were the standard module on demographics (excluding income and ethnicity which are asked at the very end of the interview) and a module on the subject of stamp collecting. There is thus little risk that our experiment is contaminated by the questions that come before it.

2. As the dependent variable is a 5-point scale, we have replicated these regression analyses with ordered probit models, with almost identical results.