UK General Election December 2019

Electoral Calculus made the most accurate pre-poll prediction of the result of the December 2019 General Election. Our final prediction correctly forecast a Conservative victory with a substantial majority.

Regression Analysis

For the 2019 general election, we relied heavily on our own regression analysis of opinion polls. This was particularly important for this election because of the large changes in voter behaviour. With Brexit highly relevant, many traditional Labour voters either stayed at home or defected to parties which were more definitive on Brexit. The Conservatives gained pro-Brexit voters in the North of England, and lost some pro-Remain voters in the South of England. Regression analysis was particularly useful in tracking these flows in advance, and was better than the simplistic uniform national swing method.

An interesting comparison of regression methods was possible by chance on 10 December 2019. On that day, both Electoral Calculus and another pollster published regression-based predictions. We used our tested methods which used a moderate sample size of about 6,000 GB voters. The other survey used a much larger sample size of 105,000 voters. But the predictions were very similar with only a one seat difference in the prediction for the number of Conservative seats (details).

The other innovative feature we used was to take the regression analysis as the "base-line" shape for the prediction, in order to give the estimated shape of political support across all the seats. But we also allowed the absolute national level of political support to vary a little up or down, depending on the results of other public polls. This combination of geographic detail from the regression, plus a 'big picture' input from other polling, was helpful.

The Most Accurate Pre-Poll

The pre-poll prediction resulting from our analysis was the most accurate one for this election. Our final prediction forecast correctly predicted a Conservative victory with a substantial majority. We predicted the Conservatives would win 351 seats, which was closer to the actual result of 365 seats than any other final pre-poll prediction (source Wikipedia).

Working with our partners, LucidTalk, in Northern Ireland, we also made a successful prediction of the result there. Our predictions of party support were all within 2pc of their actual vote share. Our LucidTalk poll correctly predicted that the DUP and Sinn Fein would lose support compared with 2017 and that the Alliance would gain support. It also correctly predicted only small increases in support for the SDLP and UUP. In terms of seats, all but two seats were correctly predicted.

Full details of our 2019 predictions are available here, To find out more about our regression polling and election prediction services, please click here to get in touch.