The 2024 UK election campaign was dominated by discussion of the polls, from start to finish. This was partially because of the sheer volume of polls being published. We had more MRP (multi-level regression post-stratification) polls than ever before, many giving quite different pictures of the size of Labour’s lead.
The chart below shows the average performance of 27 polls which predicted vote shares in the contest just prior to the election on July 4. The polling predictions are on the left and the actual vote shares are on the right for each of the five UK-wide political parties.
As a standard industry approximation, if the results differ from the outcomes by more than 3%, there is a statistically significant difference between the polling and the outcome. In other words, the pollsters got it wrong.
Final poll predictions and actual vote shares:
Using that rough yardstick, the pollsters over-predicted the Labour and arguably under-predicted the Tory vote, although in the latter case it was on the boundary of statistical significance. The other parties were within the margin of error. To be fair, different polling companies varied in their accuracy, so we need to look a little more closely at the results.
The list below shows how accurate 27 polling agencies were in forecasting the vote shares in the election. Accuracy can be measured in different ways, but the method........