Do perceived poll results affect party preferences, or do party preferences affect perceived poll results?

It is well established in the literature that a party’s perceived standing in the polls affects voters’ probability to vote for that party. However, do voters perceive a parties’ poll performance in accurate ways, or is there a nonrandom error to poll perceptions? In this paper, we argue that poll perceptions are systematically biased. On the basis of data from a voter survey conducted in four countries (N=22,504) we find for most parties an interplay of poll perceptions and probabilities to vote. Indeed, we more often find probabilities to vote influencing poll perceptions than vice versa. This bias tends to be larger among the lower-educated, and among the less knowledgeable. We conclude by setting our findings in wider perspective and discussing the relevance to the research field, and society more generally.