If you were ever asked as a kid whether you’d jump off a bridge if all your friends were doing it, you may have sheepishly replied that you wouldn’t. Perhaps not, but what people around us are doing can have a surprisingly powerful effect on our behavior. Other people’s behavior tells us about what they think, and about the correct way to behave, and often this is incredibly useful information. We’re generally better served going to a restaurant with a five-star review than one with a two-star review. If we arrive at our boss’s house for a holiday party, and everyone else has taken off their shoes, then it’s probably a pretty good idea to remove our shoes too.

Shifting people’s sense of norms––their perceptions of what their peers do or favor––is also a powerful way to address behavior associated with a social problem. Psychologists call the sharing of peer information a norm-based intervention. It may be as brief as sharing a statistic on a website, or as involved as leading group discussions about community practices.

Research shows that showing individuals that a particular behavior is common or supported among their peers can help people to consume less energy, turn out to vote, donate to charity, and cooperate with one another. For example, high energy consumers reduced their energy usage after learning that their neighbors used less energy than they did.

And yet a growing literature has revealed that not all types of peer information are equally effective, and some can even backfire. Crafting a norm-based intervention should involve some combination of qualitative work, observation, and piloting, so it's tailored to the target behavior and context.

What should intervention designers keep in mind during this stage? Here, I share a few considerations from a forthcoming review[i] with Betsy Levy Paluck.

Strikingly, whether a norm-based intervention has positive effects or unintended negative effects, it provides critical insights into how norms work. Unintended effects offer lessons for future interventions and point to the importance of continuing to measure these programs’ causal effects.

1.Who should the peer information be about?

If you learn that strangers in a different state are voters, you may not think twice about it. But if you learn that the coworkers you have lunch with every day are voters, you may find yourself feeling awkward or sheepish if you didn’t join in at the polls.

Peer information is powerful when the peers feel relevant to a person. Thus, an appropriate peer group could be one that someone identifies or interacts with, whether that’s at home, at work, or socially. Peers may feel relevant in different ways, however. An appropriate group may be one that is highly specific to the behavior or situation in question. For example, hotel guests in one study were especially likely to reuse towels in their rooms to help the environment when they read a sign saying that most guests staying in that same room had done so.

Some research has suggested that peer information should be about a group with a similar capacity to engage in the behavior. Otherwise an intervention may backfire. So, if a group of employees learns that higher-income coworkers are successfully saving for retirement at a high rate, such information may actually be demotivating since it belongs to a group with a greater capacity to save.

2.How believable is the peer information?

It may be tempting to tell people that everyone in their community is recycling, or to create a college brochure that inflates the level of racial diversity on a historically white campus. Well-intentioned programs may do so in an effort to exert a stronger pull on recycling or diversity. But in addition to questions about the ethics of fabricating peer information, presenting an extreme pattern that doesn’t match up with people’s observations about others around them may be less effective and risk losing people’s trust.

When using actual peer information, we must take care in selecting content that aligns with what people witness in their surroundings. Telling college students that many of their peers actually do not support binge drinking, for example, is an assertion that may be easy for them to ignore based on the evidence they have. Another option is facilitating their own discovery of their peers’ lack of support for binge drinking, perhaps through discussion groups. However, discussion-based interventions may not always be a good fit. They require substantial resources and the direction the discussions may take is inherently uncertain.

3.Do personal opinions align with the peer information or run against it?

Imagine two communities in which it is uncommon to report domestic violence to a health provider or authorities. Suppose that in one of these communities, residents personally support the reporting of domestic violence.  But they each have the impression that others do not, and therefore they don’t want to stand out by reporting incidents or sharing their opinion. In this case, a norm-based intervention that seeks to increase reporting has a lot of leverage, because residents are ready to hear the message.

Now imagine that in the other community, residents personally don’t support the reporting of domestic violence, perhaps seeing it as an action that ultimately pulls families apart. A norm-based intervention seeking to increase reporting has more work to do in this case, because it’s tasked with motivating people against their inclinations. And in some cases, this task involves a risk. Making people feel that their opinions don’t match the group’s norm could in fact lead them to distance themselves from the group or intervention. Qualitative work may suggest creative ways of communicating strong but palatable peer information in these instances.

4.Do people know who else is hearing the information?

It’s one thing to hear from friends individually that they each voted in an election, but it’s another to see on social media that friends have publicly posted “I voted” badges. In the latter case, people know that a wider audience is also seeing the voter badge.

People pick up on how many others may be hearing or observing the same peer information, based on the medium used to convey it. The medium provides additional cues about the existing norm more broadly: Would friends have posted a voter badge on social media if they suspected they would lose status among their peers for doing so? The medium also signals the potential for a norm to keep spreading: Might even more mutual friends now go vote or post that they voted, creating an even stronger norm?

Similarly, as a recent study found, broadcasting an audio-based message over a community loudspeaker or at group meetings––each a shared, social medium––may lead to a greater shift in people’s sense of norms than when using an individual-based medium like a personal audio player.

5.Does the information make a problem seem normal?

Organizations often share alarming statistics, such as that two out of three women in a community have been abused or that most households in a neighborhood are consuming extremely high amounts of energy. The intuition that increasing awareness of a problem is the first step to addressing it is sensible. And increasing awareness may indeed be one important aspect of deep social change on some issues. We must take care, however, not to unintentionally normalize a problem. For example, sharing that two out of three women in a community have been abused may paradoxically make abuse of women seem normal and okay.

When conveying peer tendencies that are unfavorable in the status quo, one available strategy is giving positive feedback to people who are doing “better” than peers (for example, those who are using less energy than their neighbors), and giving negative feedback to people who are doing “worse.” Another potential strategy is highlighting the norm’s favorable extreme (for example, some households are using a very low level of energy) or a direction of change (households are starting to reduce their energy usage).

While this list is far from comprehensive, these five questions may be helpful to consider when crafting a norm-based intervention. The growing body of research on such interventions is increasingly able to shed light on what works and what doesn’t. Read more in our forthcoming review, where we also examine different sources of peer information and consider when using norms may be a good fit for tackling a social problem.


[i] Tankard, M., & Paluck, E. L. (in press). Norm perception as a vehicle for social change. Social Issues and Policy Review.

Margaret Tankard is a PhD candidate in psychology at Princeton University. Her research has focused on the design and evaluation of behavioral interventions, perceptions of social norms, and efforts to reduce gender-based violence.