A “People’s Vote”?

By David Hearne, Researcher, Centre for Brexit Studies

Brexit is at an impasse. In 2016, a majority of registered electors in the UK voted – by 17.4 million to 16.1 million – to leave the EU. However, it is now 2019 – how has the franchise changed? Some suggest that changing demographics alone mean that the franchise is now majority Remain[1]. Is this true?

During the 3 intervening years just over 1.8 million people have died. Whilst this includes, tragically, several thousand under 18s, the rest (over 99%) are above the age of majority. Of these, fully 85% are in the 65+ age bracket. Over half of the remaining deaths are of people in their 50s. We can be confident that almost all of these were British or Commonwealth citizens (whether through birth or naturalization), and hence entitled to vote. It is not implausible to believe that the number of electors dying is almost equal to the total number of deaths. Although it is certainly possible to perform more complex calculations based on survival models using age-specific mortality rates, there is little to be gained over a “back of the envelope” calculation.

This is for the simple (and obvious) reason that statistical uncertainty over voting preferences is much larger (several orders of magnitude) than statistical uncertainty over death rates and ages. Continuing, therefore, with our rather lackadaisical approach to modelling, we believe that the electorate “lost” approximately 1.8 million voters due to attrition, of which 85% were over 65. We therefore conclude that there were 1.53 million deaths in the over 65 age group and 0.27 million in the other age categories.

Political scientists estimate fairly consistently that around 80% of over 65s turn out to vote in General Elections in the UK, although a post-referendum Opinium poll suggested that turnout in this age group was as high as 90%. A reasonable compromise is to assume 85% turnout for this group. Polling data has been quite consistent on the electoral preferences of this age group in terms of leaving the EU. They split around 60-40 in favour of Brexit according to polls carried out over the Referendum[2]. The British Election Study[3] finds virtually identical figures, leading one to conclude with a degree of confidence that this was, in fact, the case.

Thus, we conclude that mortality alone has reduced the Remain vote by half a million and reduced the Leave vote by around three-quarters of a million. There is some debate over how to allocate the remaining 0.27 million deaths. We do know that around half of them were in the 55-64 age bracket, which skewed Leave quite heavily. A turnout assumption of 80% is not unreasonable, although this makes little difference compared to simply assuming it matched the national average of 72.2%. Ashcroft[2] estimates a 57-43 split in favour of Leave, whilst the British Election Study[3] estimates that 54% of this group voted Leave, although since their overall results skew Remain by a couple of percentage points, a figure of 56% is perhaps more reasonable.

Accordingly, we adopt an estimate of 56% and assume that the small remainder of deaths (under 55s) voted exactly in line with the UK average (a turnout of 72.2% and split 52-48 in favour of Leave). The upshot is that, overall, the electorate has lost around 600,000 of those who voted Remain in 2016 and around 900,000 of those who votes Leave in that year. Naturally, these estimates are very approximate, but they serve as initial estimates.

The gain of new voters is much more difficult to determine. Estimates in the press have typically either considered the number of 18-20 year olds or looked at the number of 16 and 17 year olds in 2016 and simply extrapolated that forward. This, however, is somewhat unsatisfactory as a significant proportion of that group are unlikely to have the right to vote. In particular, whilst there are approximately 725,000 18 year olds, there were only 600,000 births in the UK 18 years ago.

We don’t have a breakdown of the extra 125,000 by nationality. Some will be British citizens and a number will be Commonwealth citizens (and thus have the right to vote in an EU referendum) but a large number will not.

Sadly, it is impossible to know with any degree of certainty how many fall into each category. We can say that the number of births in 1999-2001 is a reasonable minimum. 85% of these births were to mothers born in the UK, but many of those born outside the UK will be British, Irish or Commonwealth citizens. Of the remainder, some will have left the country but been replaced by British citizens born abroad and some will have become British citizens in the interim. The net result is that there are likely to be an absolute minimum of 1.8 million new potential voters since 2016.

Beyond this, there could be up to an additional 450,000 new potential voters. We can test the results with and without these individuals. Not all new voters will register – many, particularly those attending university, will be living away from home and either be unaware of the need to register or unwilling to complete registration. The young are much more likely to live in rented accommodation and are thus far more likely to move regularly, making keeping on top of registration challenging (particularly those who have other priorities, including unstable work arrangements, study and financial insecurity).

Study is particularly problematic: those living at university will often be registered to vote at their place of study, but during the referendum were resident at home (after the end of examinations). Worse, a number would potentially not even have been present to receive a postal vote. There are ways around all of these problems, but it is hardly surprising to find that a number of 19-20 year olds living away from home effectively disenfranchise themselves by accident.

We do know that the overwhelming majority of potential voters overall are on the electoral role but we do not have a breakdown by age. A realistic assumption would be a 70% registration rate, based on research from the Electoral Commission[4], although this could go as high as 75%+ prior to a second referendum that many felt strongly about. An alternative assumption would be simply to sum the number of “attainers” in each year (2016 through 2018 inclusive). Realistically, these should function as effective upper and lower bounds, although given changes in the overall size of the register, it would seem that the ‘true’ figure is probably closer to the lower bound than the upper one.

Of those registered, it is highly uncertain what proportion would turn out to vote. A very conservative estimate would be 40%, since in polling 36% say that they “always vote” in General Elections[5]. Studies have suggested that the actual turnout of young voters in the EU referendum was in the region of 65%[5]. Applying a degree of caution we assume that registration is at the lower bound of our estimates and turnout is 60%. Of those new voters, a variety of data sources[2, 3] suggest that these new voters would split 75-25 in favour of Remain. This seems realistic as it appears that this new group are even more strongly pro-European than their immediate forbears.


Pleasingly, our estimates of the size of the (parliamentary) register in 2019 almost exactly match the actual size of the most recent register. The claim that the population is now “majority Remain” solely due to demographics (assuming that voting intentions remain identical in each group) is incorrect. However, literature talking of “the 17.4 million” is also now almost certainly seriously misleading. The true figure is likely to be in the region of 16.7 million and falling. Demographic change alone would imply that the Leave majority is now below 700,000 (from 1.27 million at the time of the referendum).

However, any new referendum could not take place now. Indeed, it is reasonable to surmise that it would not occur prior to mid-2020. At this point, demographics alone would predict a Leave majority of just 600,000 people. Such a narrow majority is worrying. At that point, a determined voter-registration drive and efforts to improve turnout would almost certainly ‘flip’ the results. Although polls show that few have changed their minds, they also suggest that some previous non-voters would now choose to vote Remain and would turn out. It is therefore no wonder that polling data now fairly consistently shows a lead for “Remain”. Prudence would suggest that leaving the EU without some further test of public opinion would be very unwise at this point in time. A “pause” in the endeavour until a more decisive majority for action – one way or another – emerges would probably be the best course of action.

  1. Kellner, P., In January 2019 Britain will officially switch from a pro-Brexit to an anti-Brexit country, and this is how we know, in The Independent. 2018.
  2. Ashcroft, M. How the United Kingdom voted on Thursday… and why. 2016; Available from: https://lordashcroftpolls.com/2016/06/how-the-united-kingdom-voted-and-why/.
  3. Fieldhouse, E., et al., British Election Study Internet Panel Wave 9. 2017.
  4. The Electoral Commission, The quality of the 2014 electoral registers in Great Britain. 2014.
  5. Bruter, M. and S. Harrison, EU referendum: breaking iindifference – How age affected voting. 2017.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s