What the effect is
People do not answer surveys in a vacuum. They answer as someone who is being asked, by a person or an organisation, about something that may reflect well or badly on them. Social desirability is the tendency to give responses that seem acceptable or admirable rather than ones that are completely accurate. It is a form of measurement error that comes from the respondent’s wish to be seen in a good light, sometimes consciously and sometimes without much awareness at all.
The result is that a poll can reveal what people are comfortable saying more reliably than what they privately believe or actually do.
Where it shows up most
The effect is usually discussed in connection with topics where there is a widely recognised “right” answer or a stigma attached to the “wrong” one. Researchers often point to areas such as:
- voting and civic duty, where people may over-report having voted or giving to charity
- health-related habits, where people may under-report behaviours seen as unhealthy and over-report ones seen as responsible
- attitudes about prejudice or discrimination, where openly negative views may feel unacceptable to state
- income, spending and personal finances
- illegal or embarrassing behaviour
Such patterns are typically described as tendencies. They differ by country, period and population, and they are not found in every study. What matters is that the direction of the likely distortion is often predictable: toward the answer that sounds better.
Why it happens
There are a few overlapping reasons. Some respondents want to avoid awkwardness or judgement from an interviewer. Some want to be consistent with an image of themselves. Some simply find the acceptable answer comes to mind more readily than the full truth. The effect is generally thought to be stronger when someone else, such as an interviewer, can hear the answer, and weaker when the person feels anonymous.
It also interacts with changing norms. As attitudes in a society shift, an opinion that was once openly stated may become something people hesitate to say, and vice versa. That makes social desirability relevant to any attempt to track how social norms are changing, because a change in what people are willing to say is not identical to a change in what they think.
The difference from saying one thing and doing another
People can also be inaccurate for innocent reasons, such as forgetting or estimating badly. That broader problem, the gap between what people report and what can be observed, is the subject of self-reported behaviour. Social desirability is the specific part of that gap that comes from wanting to look acceptable. The two can be hard to separate in practice.
How researchers try to reduce it
No technique removes the effect completely, but several approaches are widely used.
Choosing the survey mode
Self-completed surveys, whether online or on paper, tend to feel more private than an interview by phone or face to face. Many studies find that sensitive behaviours are reported more readily in self-completion formats, though the size of the difference varies. For especially sensitive items, some interviewer-led surveys let the respondent type answers into a device privately.
Ensuring and explaining confidentiality
Telling respondents that answers are confidential or anonymous, and meaning it, can help. Survey introductions often stress that there are no right or wrong answers and that results are reported only in aggregate.
Careful question wording
Question wording matters here as much as anywhere. Questions can be framed to normalise a behaviour, for example by acknowledging that many people find it hard to do something, before asking. Neutral phrasing avoids signalling which answer the researcher expects. Offering a range of answer options, including less flattering ones, makes it easier to choose one.
Indirect questioning
Instead of asking a person directly, a researcher might ask what they think most people in their community believe or do. These indirect approaches can sometimes reveal views that direct questions do not, though they come with their own interpretation problems, since people’s estimates about others are not the same as their own beliefs.
Randomised response and list experiments
Some methods add deliberate noise so that no single answer reveals the respondent’s position. In a list experiment, respondents are randomly given one of two lists of statements and only report how many apply to them, not which ones. One list contains an extra, sensitive item. Comparing the average counts between the groups gives an estimate of how common the sensitive item is, without any individual disclosing it. These designs protect privacy, but they are less precise and need larger samples.
Checking against other sources
Where possible, researchers compare survey answers with administrative records or other independent measures, such as official turnout figures versus reported voting. Gaps between the two hint at how much shading occurred.
What this means for reading results
Social desirability does not make surveys useless, but it suggests some habits:
- Treat results on touchy subjects as likely to understate undesirable behaviour or opinion, and probably as a lower bound.
- Notice who asked and how: an anonymous online form and a live interviewer may produce different pictures.
- Be cautious about concluding that views have changed when the topic has become more or less socially charged. The shift may be partly in willingness to say it.
- Prefer evidence that combines survey answers with other sources.
The underlying lesson is simple. A survey records what people tell a questioner under particular conditions. Understanding those conditions is part of understanding what the numbers mean.