Turnout patterns in #EP2019 in the UK – what we know so far.

The 2019 European Election in the UK was the European Election that was not meant to be held. The inability of the Conservative Party to arrive at a consensus view on Brexit that commands a majority in parliament has resulted in the legal need to hold the EU elections — and, to many observers unsurprisingly, the UK government has made a mess of it. Thousands of British in Europe and thousands of EU citizens in the UK were denied their vote. Whether this was due to clerical or administrative errors, negligence or outright politically biased voter suppression will be a matter for the courts and a matter for social scientists.

As election results wont be published until Sunday evening, there is a growing restlessness, likely among both the Remain and the Leave side. The only data points we have so far is on turnout and this is a first attempt to provide some forensics. As I had done in the context of the 2017 General Election — turnout is a key metric. There, we noticed a significant change in turnout patterns with areas with young demographics seeing upticks in turnout; and notably, areas with a likely significant grey vote, saw lower turnout.

What statistical patterns appear revealed so far?

Following the EU Referendum, I have worked on several papers that document and characterise which areas voted to Leave highlighting that a protest motive, due to austerity, was a significant driver of the Leave vote. The question is whether attitudes among these voters since have hardened or softened. If you believe opinion polls, one may be inclined to believe the latter.

Turnout in 2019 and Leave vote in 2016

The first thing to note is that turnout has been up relative to 2014 in most local authority districts for which we have data so far. The weighted average (weighted by population in 2011 census) indicates that turnout increased by 3.6 percentage points across the 140 districts for which we have data so far.

This increase is not a trivial number, given that turnout computed using the same weights in 2014 was only 33.16% — this suggests a proportional increase by around 10%.

The increase in turnout appears concentrated in areas that were less likely to vote Leave in 2016.

Gradient indicates that turnout increased less in areas that had higher vote leave share in 2016.

2016 Referendum turnout and 2019 EP turnout?

How do 2016 EU Referendum turnout and the 2019 EP turnout line up? There appears to be a positive gradient suggesting that the polarization along the helped voter mobilization.

The question that this obviously raises is, whether turnout was skewed in any particular direction politically. UKIP won the 2014 EP elections and most opinion polls predict that essentially all UKIP voters in 2014 switch support to Nigel Farage’s limited company (which is not really a party).

So how did turnout evolve in areas where UKIP did well? And importantly, how does that compare to areas where UKIP gained votes between 2009 and 2014? This is relevant as in my research I document that austerity-induced protest voting for UKIP was a significant driver for UKIP, with 26% of 2014 EP UKIP voters indicating that they support UKIP as a protest vote “to send a message”.

Data suggests that turnout increased most in areas where UKIP did not do well in 2014 and in areas where they gained least support between 2009 and 2014. This is important to note, as UKIP has been successful in mobilizing voters back in 2014, which helped allow UKIP to win the 2014 EP elections. Below is a large poll conducted just prior to the 2014 EP election on a relatively large sample which suggested particularly high turnout intentions among UKIP supporters.

Turnout intentions among UKIP supporters in 2014 was highest.

Can we infer anything as to whether UKIP or the Brexit Party can repeat this feat?

In the run up to the EP2019 elections, for quite some time it looked like that mobilizing Leave voters may be more difficult. Yet, the most recent YouGov poll prior to the EP2019 date actually suggested that Leave voters in 2016 may have slightly higher turnout intentions (63% of Leave voters in 2016 set to vote, 61% of Remain voters). Yet, 13% of Leave voters are certain NOT to vote, while this only stands at 8% for Remain supporters.

Nigel Farages company is, according to this data, expected to attract 72% of the 2016 Leave vote, while the other unambiguous Remain parties are set to attract only 68% of the Remain vote (with a sizable chunk of 26% expected to go to Labour and the Conservatives), highlighting the division in the Remain side.

There are some reasons, why I am nevertheless optimistic that the EP2019 election results coming in on Saturday will provide some surprises.

Turnout increases appear weakly stronger in areas with younger demographic

Turnout increases are strongly driven by areas with higher share of individuals with high educational attainment (or turnout increases are lower in areas characterized by low educational attainment). My work and that of many others suggested that educational attainment variables could easily explain the bulk of the voting differences with areas.

Low turnout in urban areas has undermined the Remain side with some evidence suggesting that those that did not turn out favoring remain over leave by 2:1 or even 3:1. Similarly, a non-negligble share of Leave voters in 2016 voted for Leave not expecting Leave to win — and this group increasingly regrets or is unhappy with the vote as I document here.

EU voters, many of whom were denied their democratic right to vote on Thursday, appear to also have contributed to the increase in turnout. This is particularly visible in urban centres.

Naturally, with these types of analysis there are many caveats to bear in mind. I

Is turnout larger in areas that are Remainer Now?

In fall 2018, I wrote a paper with the title “Who is NOT voting for Brexit anymore?“. This was using data from a 20,000 member strong Survation poll and suggested that significant swing to Remain across the UK was due to gains in turnout intentions among those, who did not vote in 2016. There also appeared to be some movement indicating that some of the Leave voters in 2016 that cast a Leave vote as a protest vote may be swinging back or indicated they did not plan to turn out. This resulted in estimates suggesting that Remain would now comfortably win.

Red are districts that voted with a majority in favor to Remain in 2016.

The figure does suggest that increases in turnout were most pronounced in areas where the projected change in the Leave support was lowest — i.e. mostly, these are parts of the UK that were in support of Remain in 2016 (the red dots indicated). On the other end, turnout did not really change in aggregate in places that were projected to have the largest swing in support of Remain. This may indicate that polarization around Brexit and Leave may have further increased as the areas supporting Remain in 2016 were already then more likely to be in urban centers.

Another way to view this cut is to use the MRP estimates from 2018 to label districts as majority remain or majority leave as of 2018 October. Across the 140 districts for which turnout data is available, 76 would be majority Remain according to the MRP estimates while 64 are majority Leave.

The below plot suggests that the level increase in turnout is concentrated in areas that appear to be RemainerNow (red line versus blue line). The gradient remains intact, suggesting that increases in turnout were strongest in areas where the potential size of the swing vote is particularly small.

Red districts are ones that are classified as being majority leave according to the MRP estimates constructed from the 20,000 strong Survation poll from fall 2018.

It is important to not read too much into this, in any case. Voter mobilization can be very effective and UKIP and Farage’s new private company have been successful in getting out the vote — despite earlier polls suggesting that many more Leave voters would not plan to turn out. At the same time, many Remain voters appeared undecided for whom to cast a vote. The Remain side may have been equally effective in encouraging turnout. It is important to bear in mind that Remain and advocates of a Peoples Vote have brought millions of people on to the street ,with the UK currently being home to the EU’s largest and most active pro European mass movement.

Nevertheless, despite the fact that it was actually the Leave side that failed to deliver Brexit on tiem, the Remain side appears split across multiple parties. While the above documented patterns on turnout may suggest that explicitly pro European parties can expect significant gains on Sunday, these gains are likely split across multiple parties — in particular, Green, Change UK and the Liberal Democrats.

If Nigel Farage’s new company was less successful in mobilizing the core vote, the projected combined vote of the Brexit party may be less than many polls predicted. We know that opinion polls tend to perform poorly projecting the results if there are wild swings in turnout.

There is nothing guaranteed for Sunday, but if one thing is sure — the result is going to hold some surprises.