ComRes statement on the Report of the Inquiry into the 2015 British General Election Opinion Polls
31st March 2016

ComRes welcomes the report from the BPC / MRS Inquiry team, reviewing the performance of the published polls at the May 2015 General Election and we would like to thank Professor Sturgis and the rest of the panel for their work.

Since the General Election, ComRes has dedicated significant time and resource to analysing our polls and understanding the reasons for the trend in underestimating the Conservative vote share and overestimating the Labour vote share in campaign polls. We are pleased to note therefore that the Inquiry report’s methodological recommendations align with the analysis ComRes has published over the last ten months and the methodological changes we have made.

ComRes has led the way since the General Election in utilising models of the likely electorate in surveys – publishing numerous analyses of what the voting electorate should look like and developing the ComRes Voter Turnout Model in May last year, which has been applied to all voting intention polls since. This model, based on external data such as constituency and ward level turnout and demographics simulates each survey respondent’s likelihood to vote based on their demographics. We therefore welcome that the report recommended reviewing methods for determining turnout probabilities, and to look into developing new quota and weighting variables correlated with propensity to be observed in poll sample and voting intention.

The ComRes model uses multiple demographic factors, such as age and social grade, to model the electorate on cell-by-cell basis, which combined with new quotas and longer fieldwork times in official campaign periods, also echoes the Inquiry’s recommendation to take measures to obtain more representative samples within the weighting cells employed.

Ahead of the 2015 General Election, we used specific questions in the short campaign about whether respondents had already voted by post, and amended their likelihood to vote accordingly. This is something we will continue to include moving forward, and therefore also welcome the Inquiry’s recommendation for other campaign polling to do the same.

The work of the Inquiry Panel is important in understanding cross-industry challenges and communicating them with interested parties. We at ComRes are committed to transparency and increasing accuracy. We therefore have published our methodological experiments, produced written analysis of where the polls went wrong in 2015 and have led the way in developing durable, long-term solutions for the increased accuracy of our research.

The performance of the polls at the last General Election was not good enough – either for our own standards or the public’s. We are therefore committed to constant innovation and improvements of our methods to ensure the polls are as accurate as possible as we move into another election cycle.