Perceived social norms: meaning and application

You want people to change their behaviour, but they estimate for themselves what others do and expect, often without checking. That estimate is called perceived social norms, and it can weigh on their choice. You will read what the concept means, where it comes from, who counts as a norm carrier and how to research it carefully.

Perceived social norms are what people think others do (the descriptive norm) and what they think others expect of them (the injunctive norm).

This is a perception, not a fact: it can differ considerably from what others actually do or think. The perception also depends on the reference group: the people someone uses as a yardstick. To demonstrate a misperception, compare the perception with the actual behaviour or views of that same group.

In 1990, Robert Cialdini, Raymond Reno and Carl Kallgren drew a sharp distinction between two types of social norms (Cialdini, Reno & Kallgren, 1990). The descriptive norm describes what most people do. The injunctive norm describes what most people approve or disapprove of. According to their focus theory of normative conduct, norms influence behaviour mainly when they are the focus of attention at that moment. In their field studies on littering, drawing attention to disapproval of littering reduced littering, even in an already littered environment.

In the theory of planned behaviour, Icek Ajzen included the subjective norm as a predictor of behavioural intention, alongside attitude and perceived behavioural control (Ajzen, 1991). He describes the subjective norm as the perceived social pressure to perform or not to perform a behaviour. This is close to the injunctive norm.

In 1993, Deborah Prentice and Dale Miller showed that such estimates can be systematically wrong (Prentice & Miller, 1993). In four studies, students believed that they were more uncomfortable with campus alcohol practices than the average student. This pattern is called pluralistic ignorance: many people privately reject something but assume that the majority supports it.

An actual descriptive norm is what a group really does; an actual injunctive norm is what group members really think. The perceived norm is an individual's estimate of these. In this article, we use perceived social norms as an umbrella term for both estimates.

A question such as "how many of your neighbours do this, in your view?" therefore measures a perception. Only data on what the neighbours actually do or think shows whether it is accurate.

Descriptive and injunctive norms can diverge

Someone may think that few people do something, but that most people approve of it. A campaign that stresses how common the undesirable behaviour is can then unintentionally communicate a descriptive norm pointing in the wrong direction.

What perceived social norms are not

Perceived social norms are not social proof: the influence principle in which other people's behaviour serves as a cue for one's own choice. The knowledge base covers that in a separate article. The influence principles themselves are covered in Influence: Cialdini principles. Perceived social norms also differ from perceived support: the support someone expects to receive when trying something.

The group someone uses as a yardstick is called the reference group. Within it, some people shape more strongly what counts as normal. In this article we call them norm carriers: people or groups who serve as the normative point of reference for others.

Deborah Terry and Michael Hogg found that the perceived norm of a relevant reference group was related to intentions to exercise regularly, but only among participants who identified strongly with that group (Terry & Hogg, 1996). A group's norm therefore matters mainly to those who count themselves as members.

Elizabeth Levy Paluck, Hana Shepherd and Peter Aronow studied norm carriers in a network experiment in 56 schools with a total of 24,191 students (Paluck, Shepherd & Aronow, 2016). They called students who, according to network data, attracted a lot of attention from their peers "social referents". A randomly selected group of students ran a campaign against conflict. In those schools, disciplinary reports of conflict fell by 30 per cent over one year compared with control schools. The effect was stronger when the group contained more of these social referents.

So in research, also ask who "the others" are. Neighbours, colleagues and friends can each produce a different norm.

Perceived social norms are usually measured with questions about a specific behaviour and a named reference group. Keep three measurements apart:

• Perceived descriptive norm: for example, an estimate of the percentage of a named group that shows the behaviour.
• Perceived injunctive norm: for example, how many people in that group approve of the behaviour, according to the respondent.
• Actual norm: the real behaviour or views of the same group, from behavioural data or group members' own answers.

Only with the third measurement can you speak of over- or underestimation. Prentice and Miller therefore compared students' own views with their estimates of the average student's views.

Social norms marketing tries to change behaviour by communicating actual norms and thereby correcting misperceptions, for example around alcohol use among students.

Nancy Rhodes, Hillary Shulman and Nikki McClaran meta-analysed findings from 110 articles on norm messages (Rhodes, Shulman & McClaran, 2020). They found consistent support for an effect, but average effects were small and varied widely between studies. Injunctive norms turned out to be more influential than often assumed.

The boomerang effect

Wesley Schultz and colleagues gave households information about average energy use in their neighbourhood (Schultz et al., 2007). Households above the average reduced their use; households below it increased theirs: a boomerang effect. An added injunctive message of approval or disapproval prevented that increase.

Hunt Allcott analysed randomised programmes of home energy reports comparing consumption with neighbours, covering about 600,000 households (Allcott, 2011). Consumption fell by 2.0 per cent on average. Households in the highest decile of prior consumption reduced use by 6.3 per cent, those in the lowest decile by 0.3 per cent. In an additional analysis, injunctive elements played no significant role in keeping low users from using more. So an injunctive addition does not work automatically.

Dynamic norms

What if the desired behaviour is (still) a minority behaviour? Gregg Sparkman and Gregory Walton studied dynamic norms: information that more and more people are starting to do something (Sparkman & Walton, 2017). In five experiments, such messages increased interest in eating less meat, among other outcomes, even when the current norm pointed the other way. According to the authors, one factor was the expectation that the behaviour would keep shifting. A dynamic norm is only responsible if the trend really exists.

1. Define the behaviour precisely. "Saving energy" is different from "starting the washing machine after 11 pm".
2. Determine the reference group. Ask which group matters and how strongly people identify with it.
3. Measure descriptive and injunctive norms separately. Use separate questions and report them separately.
4. Compare perception with actual data. Without an actual norm, you cannot establish misperception.
5. Base the message on the facts. No majority norm that does not exist; a dynamic norm only with a demonstrable trend.
6. Watch subgroups. Check in advance who already does better than the norm, as a boomerang effect can occur there.
7. Test in your own context. Define the outcome measure and the comparison in advance.

This example is fictional. No research results are attached to it.

Situation. An energy cooperative wants members to use more electricity during off-peak hours.

Decision question. Does highlighting what other members already do help, and if so, which members?

Available information. Aggregated consumption data from members who consented to this, and a short member survey.

Suitable approach. The survey asks each member what percentage of other members they think shift consumption (descriptive), how many they think approve of it (injunctive) and whom they use as a yardstick: neighbours, other members or friends. The aggregated consumption data show how many members actually shift.

Possible interpretation. If members underestimate the share of shifters, a message about the actual share may fit. If the actual share is small but growing, a dynamic norm comes into view. Members who already shift a lot receive no message that pulls them towards the average.

Next step. The cooperative defines in advance which message it will test against which comparison group, and which outcome measure will guide its decision.

• Presenting perception as fact. "Members think you ought to shift" is the respondent's perception, not an established view of the group.
• Merging norm types. A mixed question about "what is normal" does not show whether it concerns behaviour or approval.
• Forgetting the reference group. A norm without a named group is hard to interpret and compare.
• Announcing small average effects as certain results. Rhodes and colleagues' meta-analysis shows small, variable effects; an effect in your own context is not guaranteed.
• Ignoring the boomerang effect. An average can pull people who already do better towards that average.
• Reading correlation as causation. That norm perceptions are related to behaviour does not prove that changing them changes the behaviour. That requires experiments.

Perceived social norms are the picture people have of what others do and expect. That picture can be wrong and depends on the group someone uses as a yardstick. If you use the concept, separate descriptive and injunctive norms, name the reference group and compare perception with actual data. Norm messages can work, but their effects are small on average, context-dependent and sometimes counterproductive, so test them in your own situation.