Why did the election polls get it wrong…or did they?
By Helen Lester, Sagitta Research Consultant and former MORI Pollster

Following the UK’s general election on Thursday 7 May 2015, the media has been awash with news stories about how the opinion polls ‘got it completely wrong’, but did they in actual fact?

The Ipsos MORI / GFK NOP exit poll for the BBC / ITV News / Sky News was far from ‘completely wrong’, correctly predicting both the Liberal Democrats’ collapse, as well as the contrasting SNP cataclysm. The exit polls also showed that the Conservatives would win the most seats (albeit suggesting they would be 10 seats short of a majority, rather than wining outright as they did). In reality, the results from the polling station exit interviews were very close to what actually happened, especially when you take into account margin of error. As the chart below demonstrates, the predicted number of Conservative seats was only incorrect by 2% and the Labour ones by -1%, with all other parties’ seats being accurate in percentage terms, albeit the individual seat numbers being slightly different. Of course, when it comes to parliamentary seats, 2% can (and did) make the significant difference between an outright win or a hung parliament.

I accept that the pre-election opinion polls did not so accurately forecast the end result (saying it was too close to call between Labour and Conservatives), but did they reflect the opinions of the nation at that point in time? In other words, were they really incorrect? After all, a market research survey is not a crystal ball, predicting future actions. It’s a snapshot of peoples’ opinions at the time they were interviewed. Yes, researchers can use modelling techniques to try and extrapolate the likely result. However, modelling is ultimately using the information provided by respondents and is reliant on much interpretation. What I am more interested in is whether the respondents’ results at the heart of the survey, the core research, were true. Moreover, the repeated ‘quick and dirty’ polls often commissioned by newspapers ahead of the election are only after attention-grabbing headline results. The media are not necessarily willing to pay (or wait) for sophisticated modelling analysis – in fact, do journalists even want that, is it not perhaps more exciting to say “it’s too close to call!”?!

The British Polling Council is undertaking an independent enquiry to investigate why the final pre-election polls, published on the eve of the election, were inaccurate. The main polling companies have already offered their own varied explanations about what the media would call a ‘debacle’. Sampling error is one possibility of course, as is a late swing to the right (voters deciding on the day to vote Tory). Indeed, Ipsos MORI’s final poll for the Evening Standard on 5-6 May reported that one fifth (21%) may change their mind on which party they would vote for on 8th May. However, whilst both of these could be true, I think there is another major factor at play – shame, or the ‘closet voters effect’ as Nigel Farage stated in reference to his own party’s supporters. In other words, I think pre-election, many people were too embarrassed or even ashamed, to say they would be voting Tory, even if they had already decided they would. If this is true, why? Why would a Conservative voter be less likely to outwardly support their party than a Labour supporter? As widely reported, Labour supporters were much more vocal on Twitter than their right-wing counterparts. Although Twitter users are not representative of the voting population, it’s interesting to think about the consequence this may have as more and more tweets were shared during the campaign and ultimately on polling day. Among my own Facebook group of friends and family (which clearly is a small sample and not necessarily representative of the British public), I have noted that the left-wingers have been very vocal in sharing their political views, both leading up to the election, on polling day and afterwards. On the contrary, the right-wingers have appeared more reserved and perhaps even ‘ashamed’ about voicing their political stance. If this is the case, and Conservative voters were just too embarrassed to admit they were going to vote Tory in the pre-election opinion polls, what changed on the day of the election to make the exit polls more accurate? I think something changes when someone leaves a polling station. Call it ‘polling pride’ perhaps. In other words, I think that when you ask someone coming out of a polling station who they have just voted for, there is a temporary euphoria and positivity surrounding their vote, they are proud to have taken part in shaping our country’s future. They have conviction about where they put their “X” and are happy to divulge it. These closet Conservatives believed a blue government would be the more secure option for our country. However, fast forward 24 hours, and as they return to their Facebook and Twitter accounts, television and online news reports, work, pub etc, they, like me, may have witnessed an onslaught of anti-conservatism and retreated back to their closets!