Details

Autor(en) / Beteiligte

• Whiteley, Paul

Titel

Why Did the Polls Get It Wrong in the 2015 General Election? Evaluating the Inquiry into Pre-Election Polls

Ist Teil von

• The Political quarterly (London. 1930), 2016-07, Vol.87 (3), p.437-442

Ort / Verlag

London: Blackwell Publishing Ltd

Erscheinungsjahr

2016

Quelle

Worldwide Political Science Abstracts

Beschreibungen/Notizen

• The opinion polls got it wrong in the 2015 general election. The Conservatives won a small overall majority in the House of Commons, contrary to almost all predictions except for the exit poll on the night itself. Most previous polls had predicted a hung parliament, with the many indicating that Labour would be the largest party. The average Conservative and Labour vote shares predicted by eight different survey agencies were identical at 33.6 per cent, which understated the actual Conservative vote by 4.2 per cent and overstated the Labour vote by 2.4 per cent in the final pre-election polls. We have to go back to the general election of 1992 to find a comparable set of inaccurate pre-election polls. In contrast the exit poll was spot on, but that had the advantage of interviewing people who actually voted, since the respondents were approached outside the polling stations. The same was clearly not true for participants in the pre-election polls. This outcome prompted the British Polling Council to set up an inquiry into what went wrong, and their preliminary report was published in January of this year. To put this on context, there were a total of 1,942 polls conducted in Britain between 2010 and 2015. This compares with the approximately 3,500 polls in the field between 1945 and 2010. So polling exploded in the years of the Coalition government. The report provides a careful analysis of the various factors which might explain what happened and was written by an experienced team of researchers. They draw attention to a number of factors including unrepresentative samples, problems of weighting the data, biased responses and interviewees who misled pollsters about their voting intentions. Perhaps not surprisingly, they end up concluding that that there is 'no silver bullet [...] the risk of polling misses in the future can be reduced not removed' (p. 54). This article critically evaluates the findings of the report, which by and large are sensible and judicious. But it draws attention to two problems which might give a misleading picture of what went wrong, and so cause misunderstandings about what should be done in the future. The first problem relates to mode comparisons: that is, whether a survey was conducted on the internet, by telephone or by interviewers talking to people in their homes. The second relates to respondents misreporting their voting intentions in these polls, something which is rather neglected in the report.