The final published ComRes poll of the campaign showed an eight-point lead for Remain, in contrast to the final result which was 51.9% for Leave, 48.1% for Remain. Of course we are disappointed that the final poll was not closer to the result and , as ever, we will be conducting a thorough review of our approach to see what further improvements can be made.
This was an unprecedented referendum with very different challenges from those of a General Election, and in which the drivers of voting behaviour yesterday were unique. Regional and demographic differences also played a powerful role, and it was clear in the final days that a large proportion of voters were still moving between dramatically different binary choices with considerable fluidity.
While we begin our internal analysis of what happened, our early thinking is that the shifting UK political landscape may require a new approach to predictive political polling, in order to take greater account of the regional demographic and new political landscapes. There is good evidence for this from 2014 when we correctly predicted support for Scottish independence in southern Scotland and the Borders, from 2015 when our polling was spot-on in predicting the Lib Dem wipeout in South West England, and from May 2016 when we got the final London Mayoral result right. Yet the whole industry finds it harder than ever to forecast accurately aggregated national vote shares. The regional and demographic shifts which were evident in the EU Referendum show no signs of abating and so we will need to find a way to manage this new dynamic.
Our internal deliberations into the contrast between our final Referendum poll and the result will continue until completion, but in the meantime we remain 100% committed to our goal of delivering deep insights of the highest quality for our clients.