What the Polls Polled: Towards a Political Economy of British Election Polls

Abstract

Studies of election polls, in Britain and elsewhere, typically focus on the accuracy of the polls. This chapter does something quite new: it takes the 138 polls conducted during the British General Election campaign of 2015 – a number without precedent − and focuses on the wide array of questions the pollsters asked. It shows that while questions about the vote – not only how respondents would vote, but also whether they would vote, why they would vote a particular way, and so on – accounted for about a fifth of the questions that the 11 polling organisations asked, questions about the leaders accounted for more than a quarter of the questions; so did questions about the issues. Taken together, questions about the conduct of the campaign, the parties and the likely outcome accounted for another quarter. It demonstrates that of the questions – nearly 2600 − asked over the 39 days of the campaign, the number and kinds of questions asked by the various polling organisations differed enormously. And it argues not only that academics have failed to treat pollsters as players in their own right, but also that to take pollsters seriously academics will need to connect the polling organisations to their histories and to the media or other interests they serve. ¹;

The data for the individual polls were collated from the polling companies’ websites. A media search was also conducted to identify any polls that may not have been listed on the companies’ websites. Thanks are due to Tom D.C. Roberts for assembling the polls, and to Tom Wynter for constructing a series of databases. Thanks also to Roger Mortimore and Bob Worcester for comments on an earlier draft.