Newington's Naomi Harris looks at exit polls, how they work, and how accurate it is likely to be on Thursday evening.
At 10pm on Thursday polling stations across the country will close and the exit poll will be published just moments after. This forecast, drawn from interviews with thousands of voters, will not only become the immediate focus for rumour and conjecture, but will lead to a lot of questions being asked (again) of the opinion pollsters as we wait for the actual results to come in.
With the latest polls ranging from a single digit to a 12-point lead for the Conservatives, from a hung Parliament to a Tory majority of 98, one thing is certain – immediately after the exit poll is published the pollsters will be divided into two different camps: the ‘losers’, who were wildly off-beam in their predictions; and the ‘winners’, who were closest to the exit forecast.
However, the embarrassment and rapture of being in each camp may only be short-lived, as the accuracy of the exit poll will soon be under scrutiny. Everyone in the polling industry knows how high the stakes are and as James Forsyth, political editor of the Spectator and a columnist at the Sun tweeted last week: “There is going to be polling industry blood on the walls on Friday, one way or the other...”
So with all eyes now turning to the exit poll, how does it work and how accurate is it likely to be?
The exit poll will be based on approximately 20,000 interviews in around 140 marginal seats. Variables such as expected turnout and incumbency will be factored in by those doing the number crunching which will be done right up to the wire as most people vote after work.
So far, so good you would think – much more likely to be accurate than the opinion polls.
The fly in the ointment comes in the shape of the postal ballot which is not considered in the exit poll.
The trend in people voting by post has been upward: 4% in 2001, 12.1% in 2005, 15.3% in 2010 and 16.4% in 2015. Missing these voters out could skew the accuracy of the prediction if they do not match the sample interviewed at the polling stations, particularly if postal votes increase again as expected.
You can see that it becomes an even more complex forecast when you throw in for good measure that turnout is higher among postal voters than in person voters (85.8% compared to 63.2% in 2015) and that different parts of the UK have different rates of postal voting which becomes further magnified at a constituency level (in 2015 43.3% voted by post in Houghton and Sunderland compared to 7.4% in Birmingham Perry Bar).
This doesn’t begin to factor in that people will have voted at different stages of the campaign when different events and issues have been at the forefront of peoples’ minds.
Looking to those who visit the polling station – will they tell the truth to the pollsters? The ‘shy Tory factor’ is a term first coined to explain the shock result of 1992 and it was resurgent among political commentators two years ago. Some are now questioning whether the ‘shy Tory factor’ will have the added complexity of ‘shy Corbynites’. This could again have implications that will be unpicked after Friday.
Comparing the results to the predictions, the last three elections have been a mixed bag for the exit poll.
The 2015 exit poll predicted that the Conservatives would be the largest party with 316 seats, followed by Labour at 238, the SNP at 58, the Lib Dems at ten, Plaid Cymru at four, the Greens and UKIP at two a piece, and the others at 19. The poll was correct in the ordering of the parties, in contrast to many of the earlier opinion polls that predicted Labour would be returned with the most seats, but wrong on the numbers. Most importantly it underestimated the number of Tory seats by 14 and overestimated the number of Labour seats by seven. As a result it wrongly predicted a hung parliament, and the prospect of a Conservative majority Government only began to dawn in the wee small hours.
In contrast, the 2010 exit poll was almost spot-on in its prediction with a difference of just six seats. It also correctly predicted the Conservatives would be 19 short of a majority. In 2005 the exit poll was out by 22 seats, as it overestimated the Conservatives’ share by 11 and underestimated the Lib Dems’ and the smaller parties’ share by nine and two respectively. However, it was right in predicting a Labour majority of 66 seats.
The track record of the exit polls means that it is safe to say that the uncertainty about polling accuracy is likely to continue for several hours after the last ballot paper is crossed, and that it is going to be a long and fretful night on Thursday for candidates - and pollsters alike.
For those who are going to be watching from the comfort of a party or an armchair, the Newington team wishes you an enjoyable election night! To receive our analysis of the result on Friday morning, please contact: [email protected].
