Descriptive norms are our beliefs about what “most people” do. They shape our behavior in powerful ways, from how we greet strangers to whether we recycle. Classic theories of social influence assume we conform to the majority because we prefer to feel a sense of belonging than rejection. Funnily enough, that logic assumes that we can accurately perceive what the majority is doing, but in real life, that’s rarely the case. 

We don’t observe each person once and tally the results. Some people are more visible. Some behaviors are repeated. Some individuals show up more often in our field of view. Even on social media, algorithms amplify certain voices. Over time, repetition creates the feeling of prevalence. 

Much of prior research on norms sidesteps this reality by presenting participants with summary statistics or by exposing them to each group member only once.^{1, 2, 3} Woiczyk and colleagues focus instead on how norms are actually learned: through repeated, uneven observation. They introduce a distinction between the “behavior of the majority” (what most individuals do) and the “common behavior” (the behavior most frequently seen). Their studies create situations in which these two diverge, allowing them to test which one shapes perception and action.

The researchers ran four main experiments (along with two supporting studies), all built around the same core setup where the most frequently observed behavior is performed by a minority, while the majority behaves differently.

In Study 1, nearly 200 U.S. participants observed the arrival times of 8 hypothetical coworkers across several days. In one condition, most coworkers arrived late, but participants saw examples of early arrivals more often. In another condition, the pattern was reversed. Participants then reported what they thought the norm was and when they would arrive.

Study 2 extended the design to consequential decisions. Almost 800 participants in the UK or Ireland observed 34 monetary allocation decisions made by 8 individuals in a dictator game. Again, the behavior seen most often was performed by fewer people. Participants then made their own real-money allocation decisions.

Study 3 examined cognitive load. Participants learned greeting norms in a fictional workplace. Observations were either randomly intermingled (harder to track who did what) or clustered by individual (reducing memory strain). 

Study 4 introduced identity relevance. Greeting behaviors were tied to gender subgroups to test whether people would prioritize learning norms from the gender group they identify with over the common behavior. 

Across studies, researchers measured participants’ perceived descriptive norm (what they believed most people did) and their own behavioral choices. 

The central finding was strikingly consistent. When the most frequently observed behavior differs from the behavior performed by most individuals, people overwhelmingly perceive the frequent behavior as the norm. For example, in study 1, when early arrivals were more frequently observed (even though most coworkers arrived late), 61% of participants believed early arrival was the norm. 72% said they would arrive early themselves. 

The researchers propose two cognitive explanations. First, identifying the true majority requires tracking who did what and mentally aggregating across individuals, which is a considerably demanding task. When memory is strained, people default to counting behaviors rather than people as an easy-way-out. Second, people may fail to discount repetition. Seeing the same behavior multiple times from the same person can inflate its perceived prevalence. 

When memory demands were stretched thin in Study 3, this common behavior effect weakened. In Study 4, when the numerical majority belonged to the gender group participants identified with, the pattern reversed: participants replicated the behavior of their gender group rather than the most frequently observed behavior. What this reveals is that identity relevance is a powerful moderator that could override the influence of frequency.

However, an important caveat arises. These studies were conducted in controlled environments which simplify the complexity of real-world norm formation–a methodological issue across all research in norm formation. Factors like social networks and reputational consequences can amplify or dampen these effects in a real-world setting. On top of that, this study only captures short-term norm learning rather than lasting cultural change, opening up an avenue of research that explores the longitudinal effect of visibility versus frequency. Nevertheless, the consistent pattern across diverse contexts strengthens the core finding that the perception of the majority is highly sensitive to how information is encountered.  

This research adds a layer to our understanding of social influence. Norms do not simply reflect headcounts but rather, the structure of exposure. When repetition and visibility are uneven, perceptions of what the majority is can drift away from reality. 

In the corporate world, culture change efforts can rely on communicating what most employees do. But if daily visibility tells a different story, people’s perceptions may align closer with exposure instead of official statistics. The same applies in the digital world; algorithmic amplification can make minority viewpoints dominant, shaping public discourse and political polarization. In that case, the spread of misinformation is not necessarily reliant on widespread belief–only repeated visibility. 

A key takeaway from the study is that what we pay attention to has an impact on norm creation. To influence behavioral change, leaders, designers, and policymakers should consider not just what is true of the majority, but what people repeatedly see. 

These findings highlight how deeply the structure of exposure shapes norm perception. In real-world environments, from social media feeds to workplaces, visibility is uneven and repetition is constant. Future research can move beyond controlled experiments to explore how algorithms, organizational hierarchies and network structures amplify or dampen the common behavior effect over time. 

The study also points to the importance of identity-relevant subgroups. When behaviors are tied to socially meaningful categories, signals from one’s ingroup can override frequency. That said, understanding how repeated exposure and subgroup relevance interact in complex, real-world networks would be valuable to explain why some norms spread while others remain fragmented. 

We often assume we follow the majority. Woiczyk and colleagues suggest otherwise: we frequently follow what we see most. When repeated behaviors are masked as consensus, visibility becomes a powerful tool. This matters because norms guide behavior, and behavior shapes institutions, cultures, and overall public life. If our perception of “what most people do” can be distorted by uneven exposure, then designing healthier social systems calls for going beyond correcting misinformation and rethinking how we observe behaviors. 

In an era defined by social feeds, headlines, and repeated encounters, what influences us more is who makes up the majority but who and what we are shown over and over again. 

This article summarizes: 

Woiczyk, T. K. A., Hosseini, R., & Le Mens, G. (2025). The common behavior effect in norm learning: When frequent observations override the behavior of the majority. Organizational Behavior and Human Decision Processes, 191, 104441. https://doi.org/10.1016/j.obhdp.2025.104441

1. Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015–1026.
2. Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal of Consumer Research, 35(3), 472–482.
3. Schultz, P. W., Nolan, J. M., Cialdini, R. B., et al. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18(5), 429–434.