‘Unrepresentative sample’ caused failure of UK opinion polls
Inquiry explains incorrect predictions ahead of Conservatives’ 2015 general election win
An electoral worker counts ballots as polls close in Britain’s general election, at a counting centre in Sheffield, in this May 7th, 2015 file photo. Photograph: Andrew Yates/Reuters
The failure of opinion polls to predict the result of last year’s general election in the UK was due to unrepresentative polling samples, an inquiry has found.
And the inquiry said it could not rule out the possibility of “herding” caused by pollsters designing their surveys and weighting responses in such a way that their results were closely in line with those of rival organisations.
The independent inquiry, chaired by Southampton University professor of research methodology Patrick Sturgis, was commissioned by the British Polling Council (BPC) and the Market Research Society (MRS) after polls suggesting a neck-and-neck race were confounded by the Conservatives’ comfortable 6.5 per cent victory margin on May 7th, 2015.
Some Labour supporters claim that inaccurate polling could have swayed the result of the election, by ensuring that attention was focused on the possibility of a Labour coalition with the Scottish National Party, rather than the agenda of a Conservative-only government.
Prof Sturgis’s findings chime with a separate study released last week by polling expert Prof John Curtice, which concluded that sampling methods may have resulted in too many Labour supporters being questioned.
The inquiry’s preliminary report, being released at the Royal Statistical Society in London, has found that unrepresentative samples were the “primary” cause of the polls’ inaccuracy.
Methods of sample recruitment used by the polling organisations resulted in “systematic over-representation of Labour voters and under-representation of Conservative voters”, which was not adequately mitigated through weighting of responses.
Other potential issues, such as mis-reporting of voter turnout, question wording or the treatment of overseas voters, postal voters and unregistered voters were likely to have made “at most a modest contribution” to skewing results.
Evidence of a late swing to the Tories was inconsistent, with some polls finding evidence of voters changing their minds at the last minute, while others did not. But researchers found that this would have accounted for at most “a small amount of the total polling error”.
Prof Sturgis said: “There have been many theories and speculations about what went wrong in 2015 but, having considered the available evidence, the inquiry panel has concluded that the ways in which polling samples are constructed was the primary cause of the polling miss.”
He said that the lack of variation in Labour and Conservative vote shares in polls over the course of the campaign was “surprising”.
The inquiry was “unable to rule out the possibility that ‘herding’ — whereby pollsters made design decisions that caused their polls to vary less than expected given their sample sizes — played a part in forming the statistical consensus”. But he stressed that evidence of herding behaviour need not imply malpractice by pollsters.
The inquiry will make recommendations to the BPC and MRS when it publishes its final report in March.