Context

On the eve of the election I looked at two likely scenarios using the slideometer app. http://r.bournemouth.ac.uk:3838/slideometer/ The second scenario I built (and posted on Facebook as evidence) turned out to be uncannily close to the actual result with Labour at 264 and the Conservatives at 305 in England and Wales. The first one (the best case scenario for the Conservative government) suggested that the result would be more or less the same as in 2015. None of my likely scenarios showed a Conservative landslide, providing I assumed some increased turnout. The scenarions were based on the assumption that most UKIP votes from the 2015 election would go to the Conservatives, but gains to Conservatives as a result would be offset by an increase in turnout of Labour voters. My best prediction was over 99% correct on a seat by seat basis and largely followed the actual pattern that arose, with a few local differences. However the actual turnout was still rather lower than that used in my prediction. This seemed iniitally baffling, as it suggested that either a block of UKIP voters must have switched to Labour, or a block of Conservatives to Labour. It is plausible to imagine some of the UKIP support in the North and Midlands going back to Labour, but it seems less likely in the solid Conservative seats in the South, particularly given the nature of the campaign against Corbyn. So, are there any clues that might explain the disparity between the turnout needed to offset the gain from UKIP and the actual turnout within the pattern of results? I tried looking at this in a number of different ways. The conclusions based on the election results themselves do seem to more or less match those of the recently published Ashcroft poll.

Change in turnout and Labour vote

If all the increase in the Labour vote was derived from a change in turnout then the data points should fall around the black line on the figure below. This is obviously a gross simplification, but it forms a useful point of reference.

In fact the points are nearly all above it (hover on the point to see the constituency name). This superficially suggests that around 2 - 3 thousand votes in each constituency were gained from other parties. Could these possibly mainly come from UKIP? It seems very unlikely, but this can be looked at by further analysis.

Fall in UKIP vote and increase in Conservative vote

What proportion of the UKIP vote went to the Conservatives? If all the increase in Conservative vote came from ex UKIP voters the points would fall along the black line belo. In fact the points fall below this. The Conservatives seem to have picked up between a half and two thirds of UKIP vote. There is not an obvious clear cut divide between North and South, when just looking at the names of some of the constituencies.

The Conservatives seem to have picked up around 50% of the ex UKIP vote, with this effect being spread accross the country fairly evenly, although some seats had a much larger UKIP vote to draw on. Many of these were solid Conservative seats anyway, so this played less of a role in the results than some Conservatives expected.

We can split by regions (slightly simplified) in order to look at this more closely.

If we divide seats roughly into those with above 50% of the population in the A B or C1 bracket and those with less than this and call these “middle class seats”" and “working class seats”" we can look at the pattern again.

The pattern in working class seats generally diverges more in those in which the UKIP vote was highest. However there are some exceptions, and Labour would be well advised to look carefully at these seats (hover on the dot to see the name) in order to understand why. In some cases the Conservative vote increased more than the lost UKIP vote, suggesting that Labour have lost some support.

Change in progressive vote

Its widely assumed that the increase in turnout may have been largely among voters turning out to vote against the Conservatives. So I added the vote for the LibDems, Greens and Plaid in Wales to form a progressive block and re ran the code.

This follows a similar pattern to Labour alone, as would be expected given that the bulk of the progressive vote is now gained by Labour. There are only a few constituencies in which the progressive total actually fell.

Conservative + UKIP vote against turnout

We wouldn’t expect much of a relationship between turnout and the change in the Conservative + UKIP vote if most of the turnout change goes to Labour.

As expected the figure does not show any clear trend. Increase in turnout does not add to the overall Conservative total. So most of the increases in the total vote for the Conservatives does indeed come from ex-UKIP voters. This might have changed the election if it hadn’t been neutralised accross the country (not just in some ex industrial areas) by an increase in Labour turnout. This may explain why the slideometer that evenly distributed changes got it right, while commentators running scenarios in their heads and basing their prognoses on unbalanced focus groups got it so wrong.

Extra votes for Labour against UKIP vote in 2015

If the extra votes for Labour are coming consistently from UKIP there should be some relationship with the increase in Labour votes after subtracting the increase in turnout and the UKIP vote in 2015.

There is no clear trend and no clear pattern in the regions. The UKIP vote in London is much lower than in other areas, and coastal towns such as Clacton have the highest UKIP vote. There is no clear generalised pattern of a large block of the working class UKIP vote going back to Labour, although individual voters may well have made this switch and there may have been a soft UKIP vote in some of the Northern constituencies that either didn’t turn out or went to Labour. Again splitting the constituencies into more working class type seats and middle class seats should help to confirm this.

Both show a similar rather random pattern rather than displaying a clear trend as would be the case if the extra votes were picked up in a consistent manner from UKIP voters. So little evidence to support the theory of large numbers of “red UKIPPers”. This is not to suggest that individual voters haven’t made the switch, simply that it does not appear to be a suitable block vote for Labour to target.

Increase in the percent of the Labour vote and age

It has been assumed that a large proportion of the increase in the Labour vote is derived from increase in turnout among young voters. I calculated the percentage of eligible voters under 30 in each constituency and plotted the percent increase in the Labour vote against this variabl.

There is a clear general trend, but a lot more scatter around this trend than might perhaps have been expected. The Labour vote still increased in many seats with very low numbers of young voters. This could still be ascribed to an increase in turnout of the relatively small number of younger voters in these seats, but it shows a much broader pattern than some of the commentary on the election would suggest. At the same time it is worth noting that the Conservatives did not win a single seat in any constituency where over 30% of the elible voters are under 30. There is a very clear generational divide. The Conservatives have become the party of the old. The Labour party is indeed the party of the young, regardless of class.

Remain voters turning to Labour?

If Brexit were a truly emotive issue during the election we would expect a fairly strong negative relationship between the increase in Labour vote and the vote to leave the EU. This might not be because Labour made it into their issue during the campaign, but rather due to a difficulty that Leave voters may have had in voting against the Conservatives or a lower pool of potential Labour voters that could turn out in the leave prefering areas.

There is only a weak trend. The Labour vote even went up in seats such as Boston and Skegness (that have never had a Labour MP).

Conservatives gaining votes in leave seats

As the leave voting seats have more UKIP voters, and at least half of these went to the Conservatives we would expect a rise in the Conservative vote to be higher in Leave voting areas. Looking at the raw votes again, rather that percent change.

The trend is clearly there, but notice that many of the strongly leave voting constituencies actually returned a Labour MP.

Conservative + UKIP change in leave seats

Adding together the UKIP plus Conservative vote as a block and subtracting the UKIP plus Conservative vote in 2015.

Interestingly an overall trend is not really there to be seen now. There has been a generalised fall in the vote for Conservatives plus UKIP accross the spectrum of voting in the referendum.

Middle England and the distribution of the vote

Mapping the pattern of change in Labour vote and the seats gained shows an interesting and rather unexpected pattern in the sense that there is far less of a pattern than might be expected. One theory before the election was that the Labour vote would be concentrated in ex-industrial areas in the North and University towns, with the party losing all support in “Middle England” leading to a more divided country. This has not happened. It is interesting to see just how much the Labour gain is spread across the country. Most Tory constituencies, even in Middle England, saw a rise in the Labour % vote. This may explain the abrupt change towards a more consensual politics since the election. Tory MPs have seen the results in their own constituencies and got the message that they have to listen to all their voters more carefully than they assumed, especially as they can’t guarantee more than around half the UKIP voters even when they take a hardline on Brexit.

Click on any constituency to get a popup box with some key details. You can choose to only show seats that have changed or the marginals (defined here as having a margin < 5000 based on 2017 votes). Turnout and Labour % increment is shaded according to the % scale colours shown on the left. Choose one layer at a time on the left and zoom when necessary. The map can be downloaded for offline use here (in the offline version you can click on the square button on the left to go full screen. )

https://dl.dropboxusercontent.com/u/2703650/electionmap.html

This is a dynamic leaflet map which can be zoomed. Use the button on the right to switch between layers.

Where is the extra Labour vote coming from?

This quick analysis suggests something rather unusual was going on. The increase in the Labour vote apparently does not just come down simply to an increase in turnout among the young. However, this is still clearly a big factor. However, if the turnout of Conservative and UKIP voters actually fell (as it seems to have done), while the turnout of Labour voters rose, the overall turnout could stay at only slightly increased levels from 2015. The Conservative vote has been boosted to above the level in 2015 through combination with UKIP, but rather than switching to Labour many UKIP voters may just have stayed at home. This may also have occurred with some Conservative voters. So, the extra Labour votes after accounting for increased turnout, may not necessarily have been the result of much switching of loyalties. It may instead come down to the overall turnout having been drawn from a different block of voters. If this is the case then the Labour gains could potentially be reversed at the next election if the motivation behind these new voters is lost. On the other hand there may still have been more switching of loyalties among youngish professionals than was first expected, and this may prove important in the long term if the Conservatives become branded as the party of the old and angry rather than the young and hopeful.

It is also apparent that, with the notable exception of University towns, the effect of the young vote is not particularly concentrated in a few areas. Younger voters are part of a wider community and they also talk to older members of their family and may sway their vote too. There does appear to have been a change towards a more consensual, intergenerational style of politics as a result of this election. Given the viscious polemic against Corbyn coming from the Daily Mail and the Sun during the election and the tone of the Brexit debate this may have been unexpected. However Conservative MPs looking at the increase in the Labour vote in their own shire constituencies must now be influenced by this, and possibly worried by it. It looks as if a more consensual approach to politics could in the end be the result from what seemed initially to be a divisive election. The conclusion seems to be that the vote will lead to a return to a more consensual, gentler approach to issues, even in Daily Mail reading middle England. The Brexit vote seems to be a softer vote than many assumed and may not have been based on particulalry strong convictions when placed in a national context. This fits well with my analysis of the referendum results which showed that the strong anti-immigrant sentiment was only apparent in a few specific seats which got a lot of press coverage but did not represent the country as a whole.

Furthermore, if this election was at least partly fought on security issues following the terror attacks it should be noted that the people most likely to be victims of terror are young people who go out at night. That is precisely the demographic rejecting the aggresive stance on terrorism based around restricting immigration and civil liberties. This block should play an increasing role in future elections now that so many have registered to vote.