Experimental feature

Listen to this article

00:00
00:00
Experimental feature

The FT’s poll of polls combines all voting intention surveys published by major British pollsters ahead of the 2019 general election. The trend line uses only the most recent poll from each pollster and weights them according to when they were conducted.

With the last pollsters having reported their pre-election polls, this is now the final update of the FT poll of polls:

Continuously-updating chart showing the FT's aggregation of polls of voting intention for UK political parties since the beginning of 2019.

Vote share vs number of seats

The polling average estimates the parties’ likely national vote share if the election were held today. But in the UK’s first-past-the-post voting system, it is difficult to extrapolate the number of seats likely to be won by each party from the national vote share.

Election analysts have long projected the number of seats for each party by assuming a “uniform national swing”. This is calculated by taking the difference between each party’s share in current national polls and the results at the last election. This figure is then added to the party’s results in each constituency in the previous election.

This approach worked best when there were only two large parties and little variation in voter behaviour in different parts of the country. Today, more complex methods - such as multilevel regression and poststratification (MRP) forecasting may provide more reliable estimates. On Tuesday, YouGov released its second MRP model, which suggests Boris Johnson is on course to win a parliamentary majority of 28, down from 68 in an earlier version of the model.

Nevertheless, political scientists and election forecasters generally believe that a Conservative polling lead of around 6 per cent over Labour is the dividing line between a Tory majority and a hung parliament.

The Brexit party plunge

The poll-of-polls shows plummeting support for the Brexit party since November 11, when Nigel Farage announced that the party would not contest seats won by the Conservatives in 2017. In part, this reflects how pollsters have adapted their methodologies to the announcement. ICMOpinium, and YouGov are showing respondents only the parties they are able to vote for in their constituency, while Kantar and Ipsos are reallocating respondents to their second choice if their preferred party is not standing.

Kantar’s 14-18 November poll found that the Brexit party vote was down from 9 to 5 per cent regardless of candidate selections, and then fell a further three points when voters’ second preferences were used in seats without Brexit party candidates.

All the polls currently included in the tracker
Pollsterlast fieldwork daysampleConLabLib DemGreenSNP/PCBrexit
Ipsos MoriDecember 11 20192213443312352
SurvationDecember 11 2019239545349353
DeltapollDecember 11 20191818453510343
PanelbaseDecember 11 20193174433411344
BMGDecember 11 20191660413214433
OpiniumDecember 11 20193005453312242
KantarDecember 11 20192915443213343
ComResDecember 10 20192051413612243
Number Cruncher PoliticsDecember 10 20191009433312343
ICMDecember 9 20192011423612233
QriouslyDecember 8 20192222433012423
YouGovDecember 6 20191680433313343

How it works

The FT poll-of-polls tracks voting intention polls covering England, Scotland and Wales combined, and calculates a rolling average for each party.

At each moment in time, the FT poll average is calculated using only the most recent poll from each pollster. This ensures that data from pollsters that conduct polls more frequently than others does not have a disproportionate impact.

Individual polls are also weighted to give prevalence to more recent data. The weighting diminishes over time so that a poll no longer has any effect on the trend line after 30 days.

The resulting live-updating chart displays the individual polls as dots and the trend as a line.

Get the underlying data as a CSV file

Get alerts on UK general election when a new story is published

Copyright The Financial Times Limited 2020. All rights reserved.
Reuse this content (opens in new window)

Follow the topics in this article