What do the academics say? Influencing people to vote

During the 2005-10 Parliament I blogged several times pointing out how the trends in turnout in British elections were more positive than many of the media reports suggested. One reason for this was simply an interest in the gap between the widely accepted clichés in political and media circles about turnout and the reality. The other was that the gap is not simply of academic interest, as misplaced stories of turnout can in turn alter turnout.

Whether it is because of a herd effect (you know that other people are voting, so you copy them) or social pressure (you know that other people are voting and voting is seen as the right thing to do, so you feel you should vote yourself), knowing about rising or high turnout can cause a self-reinforcing circle, just as information – accurate or inaccurate – saying that people are giving up on voting can end up depressing it.

Jane Green, from the University of Manchester, has researched this question in respect of the 2010 general election:

Information about high turnout levels can increase voter turnout via a process of adherence to social norms. Research also shows that high turnout information has a greater effect among infrequent voters … Citizens follow social norms: they take cues from the actions of others and seek to comply with a norm of voting behaviour.

But what about how British voters behaved in 2010? Green organised an experiment:

Respondents were randomly assigned to one of eight experimental groups, each comprising around 100 respondents. Each respondent was shown a simple, short message, and asked their intention to vote directly afterwards. No persuasion information was given, such as messages enforcing the importance of voting or asking respondents to vote. The message simply primed information about turnout level or abstention.

In the experiment people were presented with one of eight different messages and then asked how likely they were to vote at the election. Comparing the results of the voting questions gives an insight into the impact of the different messages.

National Messages:

1: Low turnout YouGov surveys indicate that 56% of voters will turnout to vote in the next general election. Using a scale from 0 to 10, where 10 means very likely and 0 means very unlikely, how likely is it that you will vote in the general election?
2: High abstention YouGov surveys indicate that 44% of people will not vote in the next general election. Using a scale…
3: High turnout YouGov surveys indicate that 76% of people will vote in the next general election. Using a scale …
4: Low abstention YouGov surveys indicate that 24% of people will not vote in the next general election.

Local Community Messages:

5: Low turnout YouGov surveys indicate that 56% of people in your local community will vote in the next general election. Using a scale …
6: High abstention YouGov surveys indicate that 44% of people in your local community will not vote in the next general election. Using a scale …
7: High turnout YouGov surveys indicate that 76% of people in your local community will vote in the next general election. Using a scale …
8: Low abstention YouGov surveys indicate that 24% of people in your local community will not vote in the next general election. Using a scale …

The conclusions from the full research (which I have condensed and simplified above)?

First, high and low turnout and abstention information had different effects, depending on whether messages were presented before or during the election campaign. High turnout and low abstention information both increased vote intention more than a negative message before the campaign, but the reverse was found while the campaign was in progress. Second, turnout messages increased vote intention in greater measure if they were related to national turnout levels, but abstention messages increased vote intention if they were related to local community abstention. The framing of messages appears to be substantively different, with abstention being particularly salient to participants if it relates to their local community. Third, the effects of abstention are strongly conditioned by frequency of voting, but these effects are not, for the most part, mirrored in comparable conditional effects of messages relating to turnout.

This research leads to similar practical conclusions for electoral administrators and one for political campaigners. Providing information about how many people have voted in various forms can increase turnout, yet neither electoral administrators nor political campaigners do this except in very rare cases. For example, when was the last time you saw a polling station with a sign outside at 9pm advertising how many people had voted so far that day? Or a target letter to an occasional voter providing information on how high turnout was in their street at the last general election?

Both of these examples have their own drawbacks, but what really matter is that the lack of experimentation with these ideas isn’t the exception; it’s the norm. It isn’t that many other things are being tried to use information of the sort tested in this research to raise turnout; it’s that almost nothing is.

The increasing body of evidence regarding social norms and influencing people to vote is still largely ignored when it comes to actually getting people to vote in British elections. That is a big missed opportunity.