EU referendum: breaking indifference

How age affected voting

The findings suggest that, while young people voted slightly less than average, they were probably close to the national average, at only eight points below according to our survey.


A graphic image of students holding up protest placards | LSE Connect Breaking Indifference


An early tweet after the UK’s EU referendum result led many to claim that young people were simply too apathetic to vote. The data, however, reveals a different picture, as Michael Bruter and Sarah Harrison, who set out to understand how age affected voting, explain.

Youth turnout

Much has been said of the generational gap between young and old voters in elections in general and in the UK’s referendum on EU membership in particular. But while the contrast in terms of preference – with young people massively supporting a vote to remain in the EU – was unanimously accepted, the question of youth turnout was, surprisingly, far more controversial. 

Differences in turnout across age groups are rarely contentious, both because most aspects of electoral behaviour do not vary that much by age and because it is already well known that young people tend to vote very significantly less than older ones. In this case, however, the question of youth turnout was raised immediately on election night because generational differences in preferences made commentators wonder if a greater youth mobilisation would have led to a different result.

The first figures that hit the news were released by Sky Data in a tweet and were understood to mean that only 36 per cent of young people aged 18-24 voted in the referendum. Comparing that figure to the 83 per cent of those over 65 who were quoted as having voted, the verdict was returned that the young were foolishly apathetic. 

However, while most articles published shortly after the referendum on the subject referred to that alarming statistic, the figures that Sky Data released had effectively nothing to do with turnout in the referendum. As they transparently explained in a further tweet, the figure came from a survey conducted for the 2015 General Election, more than a year earlier, which looked at the proportion within each generation who say that they always vote.

In that sense, the Sky Data figure is arguably a measure of what generational turnout should have been expected to be if young people had behaved the way they usually do on the day of the vote. In truth, however, they did not behave as usual. Rather than failing to engage, they turned out in much higher numbers than past voting behaviour would lead us to expect. Alongside a general lack of certainty about young voters’ turnout there is another factor which further complicated matters.

Electoral participation

Commercial surveys typically ask whether respondents voted or not and use that figure to estimate turnout. The elephant in the room is that this is not, however, how turnout is calculated in real life. Legally, electoral participation is the proportion of people who vote among eligible (registered) voters, and not among the population as a whole. The distinction is critical because we know that young people are far less likely to be registered electorally than any other age group. In July 2014, the Electoral Commission confirmed that “younger people (under 35) are considerably less likely to be registered”, with only 70.2 per cent of 20-24 year-olds on electoral registers, against 95.5 per cent of those over 65.

While the Electoral Commission has made tremendous efforts to reduce this gap, there are well-known structural reasons why younger people remain less likely to be correctly registered electorally in any country. Therefore, if we do not control for registration in surveys, we significantly underestimate youth turnout compared to other categories. The question of whether young people voted or not is politically important for two critical reasons.

Research and insights 

  • Firstly, a year on, there is still a significant proportion of younger voters who are deeply unhappy with the result of the referendum and want to be heard, yet who are met with the answer that “they should have bothered to vote if they cared that much”.
  • Secondly, the Government chose not to allow 16-17 year-olds to vote in the referendum, and some ask whether allowing them to vote could have changed the result. 

It was in this context that the ECREP electoral psychology initiative at LSE and Opinium, a strategic insight agency, collaborated in order to get a clearer picture of youth turnout for the EU referendum. We did this through two surveys where we controlled for respondents’ electoral registration.

The first was the third wave of a panel study (meaning that the exact same people were interviewed three times in late April, late May and late June) that we ran with 3,008 respondents from April to June 2016. The second was an ad hoc study that we fielded with a representative sample of 2,008 respondents at the end of June 2016.

The two studies are very different in conception. Panel studies are in many ways the gold standard for election studies because of their invaluable insights into election effects and change over time. However, they are not ideal for inferring such elements as electoral choice and turnout because they disproportionately lose abstentionists across waves and also because the very fact that people were asked three times about the referendum could have made them more interested in it than average voters. In other words, those surveys are ideal for understanding processes and effects, but not for taking a snapshot of the population. 

Our findings

After weighing for the over-reporting of participation, our panel study findings suggested a turnout of young people of about 70 per cent for 18-24 year-olds and 67 per cent for 25-39 year-olds – figures that had to be taken with caution, but which suggested nonetheless that the young had taken more interest in the referendum than had initially been suggested.

Our second study, which specifically asked respondents if they were registered to vote or not in order to control for it in our estimated turnout, had less explanatory power but gave a more reliable depiction of the reality. After weighting the results to match the actual overall turnout of 72 per cent, we found that turnout was 64 per cent for 18-24 year-olds and 65 per cent for 25-39 year-olds. This was almost identical for 40-54 year-olds (66 per cent), but below the 74 per cent for the 55-64 bracket, and the 90 per cent for those over 65.

The findings suggest that, while young people voted slightly less than average, they were probably close to the national average, at only eight points below according to our survey. Our figures also suggest that, although 18-24 year-olds voted less than voters above 55 in particular, the differential was likely to have been less than in General Elections, suggesting that young people made a greater effort to mobilise in the EU referendum than in recent General Elections.

In that sense, it is certainly unfair to criticise their right to have an opinion on the outcome on the grounds that they “did not bother to vote”. What this tells us about what might have happened had 16-17 year-olds been allowed to vote is far more complex. Our electoral psychology team has done a great deal of work on both first-time voters and the electoral behaviour of 16-17 year-olds across Europe. Invariably, we find that people in this age group vote less than older voters but significantly more than the 18-30 bracket. 

We had confirmation of this recently in both Austrian elections and the Scottish independence referendum, and other studies such as the ICM Scotland independence referendum survey confirmed our finding, with a 75 per cent turnout among 16-17 year-olds compared with 54 per cent for 18-24 year olds. 

Moreover, 16-17 year-olds (as well as 18-19 year-olds) are significantly more likely to register than 20-24 year-olds. There are structural reasons for those differences: many in their early twenties leave home to work or study, are new to local political stakes (an important predictor of voting), are more likely than any other group to travel or live abroad, and may not even know that this does not affect their right to vote.

Moreover, they may have registered to vote in a place where they are not physically present on the date of the vote (the 2016 referendum occurred after the examination periods in universities, for example), and may not even live at their term address to receive a postal vote.

Conversely, 16-17 year-olds often live with their parents, are more likely to be present both to register and to vote, and are likely to be connected with political debates and local networks. Allowing 16-17 year-olds to vote would have added nearly 1.6 million potential citizens to the electorate, but it is of course extraordinarily difficult to know whether it might have affected the outcome of the referendum.

On balance, the results of our surveys on the turnout of 18-24 year-olds suggest that it would not have been enough to overturn the result of the referendum, unless 16-17 year-olds were almost unanimously in favour of remaining. It would, however, have almost certainly reduced the advantage of Leave to such a point (probably fewer than 500,000 votes) that the very concept of a majority might have become controversial.

Brexit, one year on

Overall, while young people remain less likely to vote than their older counterparts – not least for technical reasons and because of a timing that disadvantaged them more than any other age group – it is clear that they considered the 2016 referendum sufficiently critical to their future and that of their country to participate far more than they had in any election in recent years. 

This is particularly clear when controlling for their lower registration levels. With a preference that was dramatically more supportive of remaining in the European Union than older voters, it is therefore not surprising that, a year after the vote, the rift not only between Remain and Leave voters but also across generations remains more dramatic than virtually ever before in post-war Britain.

Michael Bruter is Professor of Political Science at LSE and director of the ECREP Initiative in electoral psychology in the Department of Government

Sarah Harrison is Assistant Professorial Research Fellow in Electoral Psychology and co-directs the ECREP Initiative.