Does mode of administration impact on quality of data? Comparing a traditional survey versus an online survey via a voting advice application

Abstract

This paper compares two modes of administering an election survey: A traditional, door-to-door survey and an identical online version promoted via a Voting Advice Application. Whereas online political surveys are known to suffer from self-selection bias of politically interested respondents, traditional surveys are plagued with socially desirable responding and are susceptible to the effects of satisficing and other fatigue-related effects. Using a propensity score matching methodology, we examine the extent to which such differences exist between the two modes of administration. While we report mixed findings regarding the structure of respondents' answer patterns, significant differences emerged in relation to social desirability bias with the offline group being more "affected" than the online group.