Turnout: even more reason to join the Campaign for Democratic Optimism
I’ve often written in the past about the conventional wisdom of undue pessimism when it comes to turnout in British elections.
Partly because there is no Campaign for Democratic Optimism to balance it out, public commentary on turnout in elections is dominated by pessimism as so very many of those who do talk about turnout have (sometimes subconsciously) an interest in talking up gloom. That may be because they are campaigners for some sort of reform – and so gloom helps make their case – or it may be because they want to try to sound interesting and ‘it’s not quite so gloomy’ isn’t nearly as interesting, or likely to result in repeat interview requests, as ‘it’s awful’. And there’s always at least one political party in whose interest it is to promote gloom.
Yet we live in a world where anyone 34 or under has never even had a chance to vote in a Westminster general election in which turnout has fallen. We’re on a rising streak of four general elections in a row seeing rising turnout; something which is still only very rarely mentioned.
The ‘truth’ on turnout isn’t all brilliant by any means, but the default cliched tropes draped in pessimism are a long way from the truth too.
New evidence now gives even more reason for optimism, showing that turnout is much higher than official figures show. To understand why, you need to understand a simple quirk of our electoral registration system. It is legal for people to be registered in more than one place (e.g. students who go to a non-local university and so have both a ‘home’ address and a university address, or people with second homes). People, however, registered in multiple places can still only vote once at a general election.*
For example, one person might be on the register twice and vote once. The usual way of calculating turnout – comparing votes to number of electoral register entries – would score that as a 50% turnout rather than 100% turnout.
This difference between ‘register entries’ and ‘people on the electoral register’ is one of the reasons why there were many myths about the move to individual electoral registration as one of the factors overlooked in nearly all media coverage is that the switch may have reduced the number of such multi-entries.
But there’s also a much bigger implication: that ‘real’ turnout is much higher than we thought. In that example I gave about one person was able to vote and one person voted Calling that 100% turnout is a much better way of describing reality than calling it 50% turnout.
Strip out those multiple registrations, look instead at the number of people who are registered at least once and based turnout on that – and you get turnout at UK general elections being now around 9 percentage points higher than the official figures.
What’s more, as both the number of second homes and the number of students have both being going up sharply over the decades, it’s likely that this error has been growing over time and will continue to do so, even allowing for the complication that more students are now not moving away from home to go to university.
Here’s a summary of that research from one of its authors on Twitter (with some slightly odd tweet numbering), followed by the full research paper:
2. We generally want to know what proportion of people who could have voted actually ended up voting. This is known as voter eligible population (VEP) turnout which is hard to measure so no one bothers (except Michael McDonald and Daniel Stockemer)
— Jon Mellon (@jon_mellon) January 9, 2018
4. Comparative researchers normally use the https://t.co/9sfGPGWZaB turnout database to get their VAP and RV figures. This is a problem because IDEA is full of errors.
— Jon Mellon (@jon_mellon) January 9, 2018
6. This error is big enough to reverse the VAP time trend (compared to McDonald’s correct VAP figures). They converge at the end because IDEA switches sources. pic.twitter.com/OrNJOIlGhe
— Jon Mellon (@jon_mellon) January 9, 2018
8. Taken together, VEP gives a very different picture of trends and gaps between countries. UK’s turnout looks like it barely changed since 2001: it actually rose a lot. US turnout looks like it declined but actually rose. Gap between Sweden and UK/US larger than we thought. pic.twitter.com/yzoHMCemZq
— Jon Mellon (@jon_mellon) January 9, 2018
10. This implies an 18 percentage point fall in RV turnout that never happened.
— Jon Mellon (@jon_mellon) January 9, 2018
9. Register entries can exceed registered voters through inaccuracy: e.g. moving house (I was accidentally registered as Jonathan Mellon and Andrew Mellon) or dual registration: e.g. an unknown % of 7.5 million UK students and second home owners.
— Jon Mellon (@jon_mellon) January 9, 2018
11. In the US that figure is much higher. Here’s the registration rate by official figures (EAVS) compared with the current population survey (correcting for overreporting). pic.twitter.com/wCrXiTzTHH
— Jon Mellon (@jon_mellon) January 9, 2018
13. Sweden’s registers are basically without error because they have unique identifiers and very good tracking of citizens. So this is how much RV turnout is underestimated by in each country (using different correction methods) pic.twitter.com/lxhXCg9sRX
— Jon Mellon (@jon_mellon) January 9, 2018
14. These errors are really big. Here are the errors in comparison to the highest profile findings from the turnout literature: pic.twitter.com/Ha0WAFOEDg
— Jon Mellon (@jon_mellon) January 9, 2018
Opening the Can of Worms: Most Existing Studies of Aggregate Level Turnout are Meaningless, by Jonathan Mellon (University of Manchester/Nuffield College, Oxford), Geoffrey Evans (Nuffield College, Oxford), Edward Fieldhouse (University of Manchester), Jane Green (University of Manchester) and Christopher Prosser (University of Manchester/Nuffield College, Oxford)
* It is a little more complicated for local elections. You can only vote once for the same council. So if you are on the register in two different council areas you could vote in both of their elections even if both are held at the same time.
Leave a Reply