How Britain voted in 2015

For every election since 1979, Ipsos has produced estimates of how the voters voted.  Because the (very successful) exit poll we and GfK carry out for the BBC, ITV and Sky is only designed to predict seat shares, and by virtue of its design gives no demographic information, we hope that these figures provide a useful resource to politicians, commentators, academics and the public themselves to better understand voting behaviour, and the relative performance of the parties.

To forestall any questions, it should be noted that these are estimates, and are based on aggregating the data from our election polls and other surveys over the course of the campaign, which asked people how they intended to vote.  However, as we have done in the past, the voting intention figures are weighted to the final actual results and turnout at a regional level.  Whilst this means these figures are still estimates, this step should make them a more reliable guide to how different sub-groups voted (more thoughts on the performance of our polls can be found here, including how much of our overestimation of the Labour share may be due to Labour supporters being more likely to exaggerate their likeliness to vote, when we had other parties within two points of their actual share).  The larger sample size we get from aggregating (over 9,000, including over 6,000 who said they would vote, which we hope to update later with further data to over 10,000) also allows more confidence when looking at sub-groups.

So what do the findings tell us?  Here are some initial thoughts:

1. The Conservative share holds up well across most groups, as would be expected given their success in the election.   Labour, meanwhile, failed to achieve the swing it needed other than among young people and renters.  Labour only had a clear lead over the Conservatives among 18-34s, voters in social class DE, among private and social renters, and BME voters. 
2. Even worse for Labour, their vote share actually fell among those aged 65+, the highest turnout group, to just one in four.  This group is where the Conservatives were most successful, gaining a 5.5 point swing from Labour since 2010.  The Conservatives also achieved a 3 point swing from Labour among ABs, another high turnout group. 
3. While the vote share of the two main parties is broadly stable, the pattern of voting for other parties has completely changed.  The Liberal Democrats’ vote share has collapsed across the board, only getting above 10% among ABs (their smallest fall is among those aged 65+, where they had a lower share in 2010 to begin with).  They have fallen sharpest among under 34s (perhaps related to tuition fees) and private renters, who are the most likely to vote Green.  Meanwhile, UKIP take third place among nearly every group (exceptions again include social classes AB, and BME voters), and do best among older, white, working class voters. 
4. As we have seen in recent elections, the Conservative-Labour swing among the men and women vote overall was very similar.  Both vote Conservative in relatively equal proportions, while women are slightly more likely to vote Labour and less likely to vote UKIP.  There are more differences though if we do not treat men and women as homogenous groups.  Most notably, younger women had the biggest swing to Labour of any group, while older women had a small swing back to the Conservatives.  The two groups are almost exact opposites of each other: Labour has a 20 point lead among women aged 18-24, while the Conservatives have an 18 point lead among women over 55.
5. Both Conservatives and Labour increased their vote share among BME voters, but remained unchanged among white voters.  This may be related to the rise of UKIP among white voters (to 14% of the vote), which may have cancelled out some of the Liberal Democrat’s fall, while among BME voters only 2% said they would vote UKIP.
6. Patterns of turnout remain relatively unchanged, with concerning implications for the future of democratic engagement.  There appears to be no significant increase in turnout among young people, with 18-24s almost half as likely to vote as those aged 65% (43% vs 78%; in 2010 estimated turnout for 18-24s was 44%).  Similarly, turnout remains lower among the working classes, renters, and BME communities.

• For any questions, please contact Gideon Skinner, Head of Political Research

Technical Note Base: 9,149 GB adults aged 18+ (of which 6,202 were "absolutely certain to vote" or said they had already voted), interviewed 10 April - 6 May 2015.  3,196 interviews were conducted on telephone, 5,953 face-to-face (on surveys where voting intentions are asked as an analysis variable, but not comparable to our regular Political Monitor results without further weighting such as has been applied in these estimates).