Although the political world is just absorbing the results of the US election, we are still only a few months past one of the most remarkable UK General Election results in history. On November 25th, the Royal Statistical Society Social Statistics Group hosted a fascinating event examining how the polls performed in the UK General Election. The panel featured Professor Sir John Curtice, Dr Patrick English from YouGov, and Luke Tryl from More in Common, who shared their thoughts on the successes and challenges faced by pollsters during the campaign.
The discussion covered key issues such as the repeated overestimation of Labour’s support, the increasing use of MRP (Multilevel Regression and Poststratification) models, and the difficulties in predicting turnout and ensuring accurate samples. They also explored the persistent challenges in representing certain demographics, such as younger and ethnic minority voters, and the impact of these factors on polling accuracy. The speakers discussed how late swings in voter intentions, a historically low turnout, and the efficient distribution of Labour votes complicated efforts to predict outcomes accurately. Constituency-level predictions and the extent of SNP losses also emerged as significant topics, highlighting the ongoing challenges with election and political polling.