Over-Reporting in Referendum? Effects of Referendum Results on Survey Responses

• In Event: Understanding Mass Behavior in Taiwan in a New Era

Abstract

Overreporting of voting behavior is a well-known problem in survey research. That is, respondents systematically overreport their participation in elections. Besides public office elections, referendums are also important institutional arrangement by which the electorate can vote on a particular proposal of a policy or specific law. However, referendums are different from elections in two aspects. On the one hand, participating in referendums indicate an expression of preferences over specific issues rather than candidates. Expressing preferences is not such socially desirable behavior as choosing government officials and representatives. On the other hand, abstaining in a referendum might imply a tendency towards the status quo. Voters might not bother to go to the voting station and cast the ballot. If that is the case, we might not expect to observe overreporting of voting in referendums. In this paper, we analyze the data collected by the Taiwan Institute for Governance and Communication Research (TIGCR) in 2019 after the referendum held in 2018. We first examine whether or not overreporting exists in referendum voting. We then apply a multilevel item response theory (IRT) model to investigating the effects of referendum results on respondents’ responses, besides individual attributes.

Authors

• Tsung-han Tsai, Academia Sinica

• Chuan-Hsien Chang, Academia Sinica