Local Elections 2023: How the Pollsters' Predictions Measure Up
We now have a full set of results from Thursday's local elections, and a good night for the opposition parties, particularly the Liberal Democrats, saw the Conservatives losing 1,063 seats, whilst Labour gained 536, the Lib Dems 407 and the Greens 241. Meanwhile, the BBC's projected national vote share gave Labour 35% and the Conservatives 26%, as it put the Liberal Democrats on 20% and other parties' combined vote share totals at 19%. So how do these results match up to what pollsters were saying beforehand?
Electoral calculus' aggregation of national polls didn't translate particularly well into locals results – illustrating the difficulty of forecasting seats in local elections. Even their updated proposal of 338 Conservative losses, revised up from 248, was far off the 1,063 mark. Whilst their prediction of 395 Labour gains was closer, they failed to predict either the Lib Dems or the Greens making headway - they initially thought both would lose seats. This could be because their predictions were based on national polling - where the Greens and the Lib Dems see vote shares far lower than in Thursday's elections - as poll respondents may have different motivations, away from local concerns, and without consideration for tactical voting.
YouGov's MRP had mixed fortunes. Most of their local predictions were in the right ballpark, though a number were wide of the mark. As a broad source of error, they tended to slightly overstate the Labour vote share - often at the expense of smaller parties - though still managed to accurately predict a good night for the Greens.
Survation and Omnisis both ran local elections polls fairly successfully. Survation got the overall Labour lead broadly correct, predicting a 10% margin, though they overstated the vote share of independents and Greens relative to the major parties. Omnisis fared similarly well, if not better. They got the Conservative and "other" vote share spot on, with results for the Liberal Democrats and Labour within the margin for error.
Overall, a mixed set of results for pollsters, with the simpler voting intention questions coming off best. Though amongst all the predictions one stands head and shoulders above the rest. Greg Hands hit the nail on when his expectation management machine pessimistically predicted the Conservatives could lose 1,000 seats.
