Voting Behavior under Doubts of Ballot Secrecy

Guillem Riambau (Yale-NUS)

5 may 2017, 13:00h, Sala de Juntes

Paper with Kai Ostwald (University of British Columbia)
Ballot secrecy is a cornerstone of electoral democracy, since its real or perceived absence can make voters reluctant to express their true political preferences at the ballot box. While much work has been done on how traceable ballots interact with voter intimidation or vote buying, less is known about how the perceived absence of secret ballots affects voter behavior in contexts without voter intimidation and vote buying. Ballots in the stable city-state of Singapore, which has neither overt voter intimidation nor vote buying, contain a pre-assigned unique and non-transferable ID number. While there is no evidence that votes are traced, the ID number has led to the myth that they are, and that voting for the opposition can have negative consequences at the individual level. This paper uses two measurement approaches to formally estimate and upper and lower bound for (i) how widespread the belief in vote tracing is; and (ii) how this belief affects voting behavior. We find that between 32% and 52% of the electorate believes that authorities track votes in some manner, and that even by conservative estimate, around 8% of the electorate likely changed its voting behavior because of the belief. This magnitude is sufficient to impact electoral outcomes. The paper also makes a methodological contribution: we include an extra statement in a subset of our list experiments that is designed to be false for all respondents. This allows us to estimate the mechanical effect of adding additional statements into list experiments. We find the artificial increase to be significant for a number of subgroups, including elderly respondents and in lower SES. We suggest that future list experiments incorporate this technique in order to control for respondent fatigue.