Pollsters and the EU Referendum: Tough Job, Fair Shout
June 24 2016
A substantial sideshow to the debate that has dominated British media for the past two months has been speculation over how opinion pollsters would fare. Some even suggested that if they were badly wrong again, they might give up. Given that pressure, we think they did OK.
Several factors made this a hard one to call: the unprecedented nature of the choice (please don't compare it to 1975!); the lack of a recent previous vote on which to base 'swings' or comparisons; the complete crossing of normal party-political lines which produced two sides of strange bedfellows (think David Cameron and TU leaders on a platform, or Nigel Farage and Dennis Skinner nominally at least on the same side); and the unprecedented line-up of international figures and celebrities happy to say their piece, particularly for Remain. We're not sure what this will do for the celebrity endorsements business, but probably nothing good.
In the light of which, pollsters were not too far out - they said it was too close to call, with the possible exception of Populus yesterday (http://blogs.spectator.co.uk/2016/06/andrew-cooper-bizarrely-sets-bar-cameron-win-referendum-10-points ), and their predictions for the most part hovered around a mark not too far from what eventually transpired. You'll be able to read plenty of detailed analyses elsewhere, no doubt, so we'll just flag up a few things that struck us:
- SurveyMonkey made much of predicting the general election result fairly accurately, and seemed to have a good new approach in setting their basic voting intention question in context with other attitudes - and this time they were pretty accurate in giving Leave a one-point lead with a couple of days to go. Then... a late swing to Remain was reported, giving it a three-point lead. So not what it might have been, but we'll say the jury is still out on that one
- Opinium produced an eve-of-poll estimate putting Leave ahead by 2 points, or 7 percent if turnout likelihood was factored in. The final margin was 3.8 percent, putting the Opinium figures nearer than any of the other late-breakers
- Kudos also to Ben Page of Ipsos MORI (pictured) for one of the most interesting tweets about turnout. It was supposed to be common knowledge that a high turnout would imply lots of young voters putting in an appearance, therefore favouring Remain - but Page picked up the theory that a *very* high turnout would actually favour Brexit, representing a lot of normally-disillusioned voters making the effort - likely to be anti-establishment and more likely to be voting Leave. The final turnout was extremely high at 72.2%, fitting nicely with the theory.
- We still need to come up with an explanation for the gap between most final polls and the actual vote, and perhaps 'late swing' - which TNS announced they were probing with a two-wave qual investigation yesterday - is a bit of a red herring. Did many people really decide in favour of Brexit on the very last day? The 'shy tory' theory fits a little better, to our minds, but it needs adapting, especially as this wasn't a party political split, as above. Perhaps there's a more general 'quiet man / woman' factor, representing a group who - try as we might - are less likely to end up giving us their opinions, and in this case they were more likely to be Leavers.
Anyway, 6 or 7 out of ten for the pollsters in general - it was a tight one, and a tough one, but there's no substitute for what they produce and of course no silver bullet. We may be biased, but we'd say they can keep their jobs for now.
All articles 2006-23 written and edited by Mel Crowther and/or Nick Thomas, 2024- by Nick Thomas, unless otherwise stated.
