Newington's Greg Rosen explores whether the 2017 General Election will be an opportunity for pollsters to regain credibility.

The reputation of polling took a severe knock at the 2015 General Election, again in the Brexit referendum, and a further hit after the election of President Trump.

Politicians, public and pundits were shocked at surprise results that apparently defied pollsters predictions. The more veteran of pundits and politicians recounted folk-memories of the 1992 election at which pollsters had wrongly tipped Labour to win. Lessons were supposed to have been learned, and changes made to methodology – how had pollsters got it wrong again, and could they be trusted?

The short answer is yes and no. They can be trusted but even the best intentioned make mistakes. Lessons were learned – but pollsters themselves have throughout been at pains to point out that they are not the ones claiming complete accuracy for polls. Pollsters have always been clear that opinion polls contain an error margin of some 3%. Although, as the New York Times has reported, some academic research contends that the margin of error can be slightly higher, especially for more local polling, there are important successes for which the polling industry has scarcely had the credit it deserved.

Perhaps one of the most significant was the accuracy of the opinion polls in foretelling the Liberal Democrat annihilation at the 2015 General Election. For several years in the run-up to the election, Lib Dem poll ratings had been staggeringly low, and translating those polls into a swing showed that the Lib Dems would take heavy losses. Against that stood the Lib Dems deserved reputation for skill and tenacity in local campaigns, combined with the inability of many pundits and politicians to conceive of the idea that a political party could go from so many seats in 2010 to so few in 2015. But that is what happened. “I will eat my hat if that’s true”, former Liberal Democrat leader Paddy Ashdown had said when the exit poll was published on election night. His hat apparently survived, but the results were actually slightly more devastating for Lib Dem seats (chopped down to 8) than even the exit poll had predicted (10). The Liberal Democrat result reflected what the opinion polls had signalled, but because so many commentators had disbelieved the implications of the polls for Lib Dem seat numbers, the pollsters got little credit when they were proved right.

The reverse was true of the Labour / Conservative battle. Many observers and politicians had chosen to believe that a hung parliament was inevitable and that the effectiveness of Labour’s “ground war” meant that, despite the margin of error inherent in polls, that Labour would perform on the higher side of the range that the polls suggested and the Conservatives on the lower side. The errors the pollsters did make were a useful scapegoat.

It is clear that the polls were not as accurate as some pollsters were happy to have us believe, but it is all too often the pundits and commentators who place greater weight on opinion polls than they are ever meant to be able to bear. It is well known that pollsters need to weight sampling to secure a more accurate result than the raw data would give – and while this weighting is necessary for accuracy it is impossible to ensure that it does not in itself carry the seeds of potential inaccuracy. Some weighting is to allow for differential turnout, for example amongst different demographics, and this carries a potential risk if there is an unexpected increase in turnout among demographics who don’t usually vote (as happened for the Brexit vote).

Pundits and politicians alike remain reluctant however to talk in the nuanced ranges that most accurately reflect the findings of polls - and far too eager to ascribe a rogue result, a specific and defined shift in the polls, or even just a shift within the margin of error, as representing an ongoing trend. Polls are an approximate snapshot, and just as before photoshop it was said that the camera did not lie, it was nevertheless not the whole picture that the camera necessarily captured.

Good pollsters are here to help explain. Let's use their insights whilst remaining mindful that only Kim Jong-un can genuinely tell us what is going to be in the ballot boxes before the votes are counted.