Strategic Voter or Strategic Respondent? Survey Mode Effects in Estimating Strategic Voting.

Abstract

With the increasing costs and shrinking response rate, mixed-mode surveys have gained popularity among scholars and practitioners as a less expensive option of data collection. A number of studies have shown considerable modal differences. Particularly when administering web surveys, respondents appear to be more politically knowledgeable, more politically engaged, provide more diffuse, extreme answers, and fewer “Don’t know” responses compared to telephone or face-to-face respondents. In this paper, I examine the presence of survey mode effects in estimated models of strategic voting. I argue that self-administered (web) surveys would yield a higher rate of strategic voters compared to interviewer- assisted (telephone) surveys. Respondents interviewed using web surveys are more likely to express support for a losing party compared to phone survey respondents. Using dual-mode survey data from the 2015 Canadian National Election Study (CNES) and the 2005 British Election Study (BES), I find significant behavioral differences between phone and web respondents. As expected, web surveys estimate a higher rate of strategic voters compared to phone surveys. In addition, strategic voters interviewed via web, contrarily from telephone respondents, appear to be less partisan, more sociotropic voters, and show both portfolio and seat-maximizing behavior. These findings have implications for scholars of voting behavior, as well as election practitioners who aim in identifying and mobilizing likely voters.