Keywords: Political behavior, participation and culture

A persistent concern amongst political actors and media commentators is that the publication of opinion polls of vote intention during election campaigns may themselves influence vote intentions. For instance, voters might prefer parties that are performing more strongly in the polls if they prefer to be on the winning side, the so-called ‘bandwagon effect’. Alternatively, they might decide to vote for a poorly performing party to prevent it dropping below a seat threshold in proportional systems, the ‘underdog effect’. It is because of fears that voters will be influenced by opinion polls that their publication is banned in many jurisdictions in the days leading up to an election. A particular concern in this regard is that the polls might be presenting an inaccurate picture of the vote intentions of the wider electorate. When the polls are wrong, as they undoubtedly sometimes are, voters would be basing their electoral choices, whether to vote and who to vote for, on incorrect information. In this paper, we present the results of a study which considers this question in the context of the 2015 and 2017 UK General Elections. These elections are interesting case studies for assessing this question because, in 2015, the polls showed hardly any movement throughout the campaign, whereas in 2017, there was a notable late surge in support for the Labour party. Our empirical strategy is to link contemporary polling data to five waves of the British Election Panel Study, which interviewed a large panel of respondents over 13 waves between February 2014 and June 2017. We use random and fixed effects models to estimate the causal effect of published polls on stated vote intentions.