Despite the growing popularity of VAAs, there is still one pending problem: to what extent are data collected through VAAs a) generalisable and b) comparable to other, traditional types of surveys. Beyond the inherent bias involved in collecting data from self-selected, politically interested individuals, VAAs also differ from traditional political surveys in the mode of data collection. Traditional political surveys usually involve interaction with the person who collects the data, which makes data collection vulnerable to social desirability bias. In order to assess whether data collection through VAAs mitigates social desirability bias in responses, this paper compares data collected from a VAA for the 2013 Presidential elections in Cyprus with data concurrently collected from a door-to-door administration of the same “questionnaire”, where the administrator was instructed to be merely present during completion. The goal is to identify systematic bias connected to the different mode of administering the data collection.
