It is quite a tricky problem to find out if people are telling the truth when they respond to surveys. One of the few opportunities to investigate this arises at election times, using the official records kept by local authorities to make sure that if an election is challenged they have a record of who actually voted. These records do not show how anyone voted, just whether or not they did, but they allow us to compare the records with people’s claims about what they did on polling day.
For example, the 2010 British Election Study (BES) carried out checks on 3,515 respondents who took part in their survey. The table below compares claims of voting with actual voting for those survey respondents and shows that significant numbers of people gave misleading answers. The left-hand column shows the respondent’s validated vote obtained from the polling records after the election. The next two columns report the respondent’s self-identified turnout in the post-election wave of the survey. Thus 55.6 per cent of all respondents claimed they had voted in person when questioned in the survey, and the validation exercise confirmed they were telling the truth. Another 10.7 per cent of respondents correctly reported voting by post or by proxy. Combining these two categories, we find just over 66 per cent of people correctly reported voting, which is approximately the same as the recorded turnout in the election.
REPORTED VOTE AND ACTUAL VOTING IN THE 2010 GENERAL ELECTION
| VALIDATED VOTE | SELF-REPORTED VOTE | ||
| VOTED (%) | DID NOT VOTE (%) | TOTAL (%) | |
| Voted in person | 55.6 | 1.7 | 57.3 |
| Postal/proxy vote | 10.7 | 0.2 | 11.0 |
| Not eligible to vote | 0.3 | 0.4 | 0.7 |
| Eligible but did not vote | 6.3 | 13.6 | 19.9 |
| Not on the register | 5.0 | 6.2 | 11.2 |
| TOTAL | 77.8 | 22.2 | 100.0 |
Source: British Election Study 2010.
The interesting section of the table relates to those who either did not remember what they did on election day or provided incorrect answers. Firstly, some respondents said they voted but either could not, because they weren’t on the electoral register, or did not, because the register confirmed they never cast a ballot. Some 5.3 per cent of all respondents claimed to have voted when they were actually ineligible (0.3 per cent) or not on the register (5.0 per cent), and a further 6.3 per cent claimed to have voted, and were on the register, but never marked a ballot. Combining these categories, 11.6 per cent of survey respondents incorrectly claimed to have participated in the election. At the time of the 2010 general election this equated to about 5.3 million people, so it is a significant group.
People also make mistakes in the opposite direction, claiming not to have voted when in fact they did: 1.7 per cent of respondents in the survey forgot that they had voted in person and a further 0.2 per cent forgot that they had voted by post or by proxy. But people are more likely to report false positives – claiming to vote when they did not – than false negatives – forgetting they cast a ballot. And this is an enduring problem: self-reported turnout has been consistently higher than actual turnout, and by a fairly stable difference of about 10 percentage points, going back to the first British Election Study in 1964.
The standard explanation for this is that voting is seen as a desirable thing to do by most people, and so non-voters are tempted to claim they voted in order to conform to this social norm. Yet plenty of people are happy to disregard this social norm: 13.6 per cent of interviewees declared truthfully that they did not vote, and another 6.2 per cent correctly declared they were not registered to vote. Add these groups together, and we find that one in five citizens is quite happy to ignore the social norm sanctioning abstention. Young people, who have a weaker sense that voting is something expected of a good citizen, were particularly likely to do this. So while some over-reporting is likely to reflect a desire to conform to social norms, this is unlikely to tell the whole story.
One of the other questions in the BES survey in 2010 asked respondents to estimate how likely they were to vote in the approaching general election. Responses were scored along a scale where zero meant ‘very unlikely’ and ten meant ‘very likely’. When this is analysed in comparison with records of voting, it turns out that high scores on this scale were associated with incorrectly claiming to have voted after the election. In other words, if individuals did not expect to vote to begin with, they were quite content to admit to an interviewer that they had not voted. But if they really wanted to vote and subsequently, for whatever reason, failed to do so, they were more likely to mislead the interviewers.
The psychologist Dan Ariely claims that behaviour like this is driven by two opposing motivations – what he calls ‘fudge factor theory’. One of these motivations is to see ourselves as honest and honourable people. The other is to take advantage of opportunities to cheat and free ride on the efforts of other people. His experimental work shows that most individuals will take advantage of opportunities to cheat a little bit, if they can. But they will not do this to the point of having to admit to themselves that they are dishonest and unprincipled. In other words most people cheat a little bit, but only a few cheat a lot because the latter is incompatible with a good self-image.
This may be what drives the misleading answers given to interviewers. It is the result of ‘cognitive dissonance’ between the desire to vote on the one hand and social norms about voting on the other. Most people will recognise the social desirability norm, but the effect of this on their answers to interviewers depends on whether voting is important to them personally. People who think that their own electoral participation is not that important may well recognise that voting is seen as socially desirable, but they are less embarrassed to admit failing to meet this social standard because it is not personally important to them.
But the psychology is different for those who really wanted to vote before the election and failed to do so. They face a strong internal conflict: voting is something they regard as socially desirable and personally important, yet they failed to act in accordance with these motives. One way of dealing with this dissonance is to mislead interviewers. Another is to mislead themselves, since many respondents may not be intentionally lying but rather end up convincing themselves that they voted even when they did not.
Memories are notoriously faulty, and people often edit their recollections to suit their present self-image. However, only those who really value something go to such trouble, so when we try to assess whether respondents are giving ‘socially desirable’ responses in surveys, we have to know how important the particular topic is to respondents if we are to make sense of their answers. People only mislead when it matters.
The social desirability bias for voting is discussed in David E. Campbell’s Why We Vote (Princeton University Press, 2006). Fudge factor theory is explained in Dan Ariely’s The (Honest) Truth About Dishonesty (HarperCollins, 2012). The low levels of civic duty among young people are explored in Affluence, Austerity and Electoral Change in Britain by Paul Whiteley et al. (Cambridge University Press, 2013). Some of the interesting ways in which memory can deceive and people end up ‘rewriting’ history are discussed in Daniel Kahneman’s Thinking, Fast and Slow (Allen Lane, 2011).