Pre-election polls and voters' behaviour

Understanding a citizen's decision to vote has been object of academic interest for long, and among the most empirically contested topics is what role pre-election polls could effectively play in encouraging or discouraging turnout. Conceptually, the smaller the predicted margin of victory, the higher voters' participation should be. A simple mechanism possibly generating this is that the more competitive the electoral race, the higher will be an individual's perception of the importance of her voting decision. Anecdotal evidence supports this statement: in the UK general election of 2001, an expected high margin of victory for Labour resulted in very low turnout. For that election, a clear majority of abstainers believed there was no point in voting as it was obvious that Labour would win anyway. Yet, for example the constituency of Arundel and South Down witnessed the victory of a conservative candidate and a well above the average turnout, raising concerns on the existence of heterogeneity in the effect of opinion polls. Through what mechanism do pre-election polls affect voters' behaviour? Do media outlets play any role in (re)shaping this effect? And finally, has the impact of pre-election polls been changing over time? I intend to address these questions in the UK electoral context through different empirical designs and with the aid of various data sources.