Labour have overturned two huge majorities in Mid Bedfordshire and Tamworth. The party won Tamworth on a 24 point swing, the second largest by-election swing to Labour ever. In Mid Bedfordshire, the swing to Labour was 21 points and its victory represents the second largest numerical majority overturned in by-election history. 

This was an exceptionally good night for Labour – as it was for Survation’s constituency poll of Mid Bedfordshire conducted between the 12th – 15th September. 

Below, you can see our predicted headline voting intention compared to the actual outcome.

When we take only respondents who told us they were certain to vote, the accuracy is increased further. In our constituency poll, 42% of respondents indicated they would definitely vote in the election today (10 to vote). This figure is typically a reasonable proxy for turnout, and actual turnout in Mid Bedfordshire was 44%. This adjustment puts our assessment within 1 point of the Labour vote share, 2 points of the Conservative's, 0.2 points of the Liberal Democrats', 1 point of the Independent's, 1 point of the Green's, and 0.1 point of the True and Fair Party. 

Seemingly, voting intentions in Mid Bedfordshire were stable from mid-September through to polling day. A dynamic that we correctly identified was the presence of an anti-Conservative vote among the electorate. At the time, 9% of those expressing a party preference indicated they were not voting for their first choice party, rather are voting in order to keep another party out, with 68% of that group stating the Conservatives. This made up a fifth (19%) of all voters who expressed a vote choice in our poll. 

Voters were, however, unclear which opposition party was best placed to win the seat. In 2019, Labour won just 22% of the vote and the Lib Dems 13%. While Labour’s neck and neck polling status with the Conservatives made them the sensible option to defeat the Conservatives, voters were unsure which party to support. The opposition vote was split in such a way on polling day.

Losing in Mid Bedfordshire and Tamworth mean the Conservatives have lost eight by-elections in this Parliament. Rishi Sunak has overseen four defeats himself. As with any by-election, the statistic of principal interest is the swing. Our poll picked up the significant swing away from the Conservatives to other parties and estimated the non-Conservative vote share within 2 points. That Labour won in Mid Bedfordshire despite a split opposition vote is testament to the collapse in Conservative support.

It should be remembered that these are by-elections; voting intention polling gives you the best read on how parties are faring with the electorate. Based on our model projections, with fieldwork of around 40,000 voters conducted from mid-August through the end of September for clients Greenpeace (GP) 38 Degrees (38D) and Lodestone, Labour is on track for a very significant Parliamentary majority if a General Election taking place today. The lowest seat count we have forecast for the party is 420, translating to a comfortable majority of 206 seats (38D model).

Recent by-elections have exhibited much larger swings against the incumbents than our General Election MRP projections. For Mid Beds and Tamworth, our data reveals notable increases in Labour's vote share — 12.2% and 16.4% respectively. Despite this, our MRP averages do not currently predict a Labour victory in either seat in a general election. Labour’s path to a 200+ seat majority does not, in our modelling, run through either Tamworth or Mid Beds; nor do last night’s results mean we expect Labour to hold both of these seats in a General Election. But swings of 21 and 24 points in seats we do not think Labour needs to win in order to achieve a landslide majority do suggest the Conservatives are heading towards a heavy defeat.

Get the data

Survation interviewed 559 people aged 18 plus, living in the Mid Bedfordshire constituency. Interviews were mainly conducted by telephone, with a small online boost. Further details and full data tables can be viewed here.