How to do democracy — Compulsory voting and Brexit. With BONUS: how changing your pool of eligible voters changes results
If voting were a legal requirement for all eligible citizens, would it make a difference to election outcomes? In this article we briefly look at some of the ideological arguments for and against compulsory voting, and then investigate whether compulsory voting would have affected the the outcome of the 2016 British Referendum on whether to remain in or leave the EU (we’re going to refer to it as ‘the Brexit Referendum’ because we like a good portmanteau, or ‘goodmanteau’). We will then diverge from the topic of compulsory voting to look at how results would have been affected if non-UK EU citizens who reside in Britain were allowed to vote in the Brexit Referendum. Spoiler alert: it does make a difference.
Voting is compulsory in relatively few nations across the world, and actually enforced in even fewer.
It’s also compulsory to vote in some other regions such as the Indian state of Gujarat and the Swiss canton of Schaffhausen.
You might think it’s strange that in a democracy people would be forced to do anything, including vote, but we both grew up in Australia, so we were compelled by the threat of a fine to vote. We don’t personally feel as though our human rights have been undermined. You can see voting as either a right, something we get to do; or as a duty, something we must do to uphold the democratic institutions of our society. Reasonable minds can differ.
Arguably, non-compulsory voting means that the people who actually care about politics, who have done their research, and who have engaged in the democratic process have more of a say. Perhaps it is a good thing, that votes aren’t too weighed down with ignorance. As a potential marker of ignorance, Australia, which has a high voter turn-out (~94%) also has a high proportion of informal votes where people haven’t filled in their ballot correctly, or at all, or who have adorned them with crude messages or drawings. 5.9% of voters ‘spoilt’ their ballot in the 2013 federal election.
But then again, sometimes the people with the strongest views and the strongest will to vote are among the least informed.
The main advantage of compulsory voting is that it means that politicians can’t afford to alienate segments of the population. Everyone is going to vote and if politicians want the highest number of votes, they need to address the needs of the biggest group of voters possible. When people don’t have to vote, it becomes easier for politicians to not bother courting certain segments of the population, thereby disenfranchising all those outside of their key demographics. Disenfranchisement is more of a risk for some racial, socioeconomic, religious, age, gender and other groups. As Rohan Wenn, spokesperson for the political advocacy group Get Up!, told the BBC in 2013:
In non-compulsory voting systems, the people who don’t vote are the poor and disenfranchised and those are exactly the people we think should be voting.
We also wonder how authoritative election results are in cases where a huge chunk of the voting population haven’t had their say. Theoretically, you can have a situation where most people don’t want to elect Party A, but Party A supporters were over-represented in voter turn-out — perhaps because they care more and are better informed, perhaps because they’re angry and motivated by resentment, perhaps because it’s easier for them to get to a polling booth, who knows? And thus, Party A wins but the minority of people actually feel good about it. This at least seems plausible, the voter turn-out for the 2012 US Presidential Election was only 57%, so it could easily be possible that the other 43% of Americans who are eligible to vote could have had a significant impact on the outcome.
These are all philosophical arguments, but what does the data say? We can’t directly compare electoral outcomes with and without compulsory voting because no election (that we know of) was repeated under these two conditions. So, in this article, we’re going to cross into speculative territory to see how much compulsory voting could have affected the Brexit Referendum. We will let you know our reasoning, assumptions, and limitations throughout the article, because (we repeat) this is just speculation. It’s good speculation, for reasons you’ll see, but it is not definitely how things would have gone if voting were compulsory.
The other important caveat is that although compulsory voting changes outcomes, this is not necessarily an argument for compulsory voting. Rather, we want you to think about voting procedures and systems because they do really matter and do really make a difference in the way nations forge ahead politically.
Ready? Alright!
Brexit Results
It was a close one:
51.9% of voters wanted to leave the European Union and 48.1% wanted to stay.
Though, for our purposes, we want to re-frame the results to look like this:
37.4% of eligible voters wanted to leave the European Union, 34.7% wanted to remain, and 27.8% of eligible voters did not vote. A tiny proportion of ballots were rejected (0.1%).
Note that this reflects a rather large voter turn-out in a jurisdiction that doesn’t make people vote. The high turn-out likely reflects what a polarising issue Brexit was (and continues to be), and possibly the fact that the pre-vote polls were predicting a close result. Some referenda seem likely to easily pass or fail based on the question posed and the media leading up to it. Brexit was not like that, the likely result was always in doubt, which motivates more people to turn up at their local polling station: their vote might actually count.
Still, 27.8% of people can make a difference, as we shall see.
Scaling by Location
Our first method of figuring out how the Brexit Referendum would have went if there were compulsory voting is simple scaling. As an example of what we mean, say an area has 100 people (areas are much bigger than this, but stay with us) and 25 people voted remain, another 25 voted leave and the remaining 50 didn’t vote. The scaling model assumes that if the other 50 people did vote, they would vote in the same proportions as their neighbours. So, we would estimate that if voting were compulsory, 50 would have voted for remain and 50 would have voted for leave.
Except, even in jurisdictions with compulsory voting, it’s rare to have 100% voter turn-out. To correct for this, our model assumes a 94% voter turn-out (based on Australian figures), so in the above hypothetical example, we would allocate 47 votes for remain and 47 for leave.
The British Electoral Commission collected the voting data in different areas around the country which shows each region’s voter turn-out and the proportion of votes that went to either remain or leave. Scaling this information leads to this result:
Or, you can look at it this way:
It would still be leave (at 51.7%), but not by as much.
This new result reflects the fact that strongly remain areas, particularly Scotland, had a lower voter turn-out than strong leave areas. The case of Scotland is interesting in the sense that national politicians overwhelmingly supported the remain campaign and so Scottish voters may have been insulated from the extent of the leave sentiment from their southern counterparts. This would have had a demotivating effect for eligible voters. If you’re pretty sure that one side is going to win, it becomes more difficult to prioritise the trip to the polling station on an otherwise busy Thursday.
In terms of limitations, the areas defined by the Electoral Commission are not uniform in size — the smallest (Isles of Scilly) had 1799 eligible voters while the biggest (Northern Ireland — yes, the whole of it) had 1.2 million eligible voters. Breaking it down by regions in this way makes for very blunt, not very nuanced estimations. Additionally, is it a good assumption that people in the same, very roughly defined, geographical areas tend to vote the same way? The geographical differences in the Brexit Referendum outcome were striking, but we aren’t so sure that this was the only, or even the main factor, leading to the outcome.
Scaling by Age
One of the big observations from the Brexit Referendum was the disparity between the attitudes of younger and older voters. Take a look:
The data maps an overwhelming trend, that the younger you are the more likely you are to want to remain in the EU. And yet, areas with a younger demographic tended to turn-out in fewer numbers:
As FT Data points out, the observed trend isn’t perfect because you only know the median age of people in each region, you don’t know the precise demographic figures on who actually voted. Note though that low voter turn-out among young people is observed in various parts of the world. Nobody seems to know why it is the case exactly, perhaps apathy or even disenfranchisement, or a lack of understanding of politics.
If all eligible voters had to vote, though, would the inclusion of a greater proportion of youth make a difference to the overall result? We estimated an answer to this question by scaling the BBC bar graph by the number of eligible voters in each age group. To do this, we found out how many people were in each age group (note that this figure includes people who may not be eligible to vote). Here’s our head-to-head results:
50.7% remain and 49.3% leave. Huh.
Again, these results are highly, highly speculative because the assumption that the proportion of leave and remain would be consistent over each age group. In actuality, it’s likely that the young people who did vote were more passionate about the issue and held stronger opinions than those who didn’t. We’d expect that people who didn’t vote would have been a more euro-ambivalent than their peers, but we don’t know. And then again, maybe the older people who didn’t vote would likewise be more ambivalent, thereby correcting the estimate to some extent.
Yet another problem is that we assumed that the voter turn-out would be consistent across age groups, which seems unlikely. We also haven’t included Gibraltar in these estimates because we couldn’t find an available age break-down of the population, but they are tiny (roughly 0.05% of the population) and positively pro-EU, so this would likely have a small effect on our estimated figure.
In any case, there is a lesson here for young people: you should vote. Your collective opinions actually can make a difference. Plus, if you show politicians that you’re going to vote, they may make more policies designed to please you.
We could use scaling methods to play with other variables that were likely important in the Brexit Referendum. Things like whether you’re born overseas, whether you live among many people who were born overseas, how educated you are, and so on, are predictors of whether you’d prefer to leave or remain. We’re not going to go through each of those factors though, because we think you get the gist. It’s quite likely the Brexit referendum results would have been different if voting were compulsory, and it’s possible that the outcome may have been different also.
Bonus — Who Votes?
Forget about compulsory voting for a second, that part is over. In this section we don’t scale results, instead we speculate on what the results might have looked like if one group of people excluded from voting in the Brexit Referendum were included.
In the Brexit Referendum, people residing in Britain who were citizens of the EU but not of the UK were not allowed to vote, except those from Ireland. However, UK residents from Commonwealth countries such as Australia, Canada, New Zealand, India, South Africa, and plenty more were allowed to vote.
Is this fair? We have lived in Britain for the last 10 months and voted thanks to our Commonwealth citizenship, but we know plenty of EU citizens who have resided in the UK for years, even decades, have jobs and families here, and so on, but couldn’t have their say at the polls. Obviously, if this group of people — comprising of some 3 million people — were eligible to vote, they would probably have tended to vote remain. Note also that non-UK EU citizens who reside in the UK are allowed to vote in general elections.
Here’s another way of framing the Brexit Referendum result, noting the proportion of votes for leave and remain, the proportion of people who did not vote, and the rough proportion of non-UK EU citizens who reside in the UK and were excluded from voting:
In our calculations, we decided that they would have voted remain at a rate commensurate with the pro-EU Lambeth (around 79%). We’ve also assumed that non-UK Europeans would have had a higher than average turn-out rate, of 80%, because they care more than average about the result. Honestly, we’ve just pulled these figures from nowhere, but they seem plausible, more of a safe floor than a likely ceiling. Keep in mind, the people of Gibraltar voted for remain at an enormous rate of 95% and had a high turn-out of 84%. They actually live on mainland Europe so likely have similar levels of skin in the game as non-UK Europeans.
Making these assumptions, here’s the result we get:
Or, in a head-to-head comparison:
Bremain. Slightly.
Conclusion
The referendum was really quite close, so different results come quite naturally from changing the parameters of who actually votes, and who gets to vote. There probably would have been different outcomes also if the voting age was lowered, if all residents were eligible to vote, if there weren’t horrible floods through London on the day of the Brexit Referendum, etc.
People in places like Britain have inherited a lot of fantastic democratic instruments from their foreparents. We have to remember though that foreparents are like actual parents, many are good, some are bad, the vast majority sit in between, and all are humans as flawed as us today. It is a good thing, we think, to at least consider different alternatives because, as we’ve shown, the choices our foreparents made affect our lives. And thus, they should really be our choices.
Our raw data comes from the Electoral Commission, the BBC, and the Office of National Statistics 2015 population data.
