Self-deception and the trustworthiness of political polling

This paper studies to what extent self-deception is related to someone’s reported voting intentions. Specifically, it studies whether extreme party voters are more prone to self-deceptive behaviour than non-extreme party voters. This finding could serve as an explanation for the large discrepancy found in election polling. Especially as mixed-method pre-election polling is becoming more and more used, it is likely the discrepancy can be explained by the behaviour of people rather than the method of polling. To test this, an online experiment is conducted. The experiment that is used is for a great extent based on the experiment of Mijović-Prelec and Prelec (2010), who use it to find evidence for self-deception under students. The experiment makes use of Korean characters, which participants must classify as either male or female. Using a logit model to analyse the answer patterns, there is no significant evidence to conclude that being an extreme-party voter influences the likelihood of showing self-deceptive behaviour. However, overall a large percentage of self-deceptive answers is found, showing that this psychological tendency is indeed present.