Third-person effect

The third-person effect is a self-other comparison: someone expects a message to change other people more than it changes them. It does not show that the others were actually persuaded, that the person was unaffected, or that the message was harmful.

That distinction matters whenever someone says, “This will work on them, not me.” It may reveal a gap in presumed influence. It is not a measurement of influence.

A fictional example of the third-person effect

This is an illustrative hypothetical, not a study or a report of real people. In a fictional campus transit survey, Maya sees a poster asking late-bus riders to keep earbud volume low. She says it will probably change unfamiliar riders’ habits, but not hers, because she already thinks of herself as careful.

To interpret that answer, the survey would need separate questions about Maya, her close friends, and unfamiliar riders. It would need a different measure to know whether anyone changed their listening habits. Maya’s response is a perceived influence gap, not proof that anyone did or did not change.

What the third-person effect means

Communication researcher W. Phillips Davison introduced the term in 1983 for the expectation that persuasive communication affects others more than oneself.[^davison] Researchers commonly call the self-other estimate the perceptual component. When they test whether it is associated with a later intention, such as supporting a restriction or correcting a message, they are asking a separate question about a behavioral component.

The named comparison group is part of the result. “Other people” might mean close friends, people similar to the respondent, a local community, or a distant general public. Changing that reference group can change the measured gap.

What research supports

The perceptual finding has appeared across varied message settings, including news, advertising, and entertainment. A 2000 meta-analysis of 32 studies and 121 effect sizes reported an overall self-other association of r = .50.[^paul]

A later meta-analysis included 60 papers, 106 studies, and 372 effect sizes. It found the third-person perception robust across the studies it synthesized, while identifying meaningful moderators: message desirability, perceived vulnerability of the named other group, similarity between self and others, and whether others seemed likely to encounter the message.[^sun]

Those figures summarize research measures, not the percentage of people who show the pattern. They do not mean the same gap appears on every platform or for every audience. The earlier synthesis also found larger estimates in nonrandom and college-student samples.[^paul]

Why “others, not me” is not one fixed mechanism

Self-enhancement is one proposed explanation. People may see their own judgment as more informed or resistant. It is a plausible account in some settings, not a universal diagnosis. Social distance, message desirability, reference group, and perceived exposure also matter.[^perloff]

The classic direction can reverse. For socially desirable messages, people may say the message will affect them more than others, a first-person pattern. This is a useful limit: the third-person effect is not simply a rule that people always deny influence.

A self-other comparison also says nothing conclusive about real media effects. A person can be mistaken about their own susceptibility, mistaken about others’ susceptibility, or both. Establishing actual persuasion requires evidence beyond estimates in a questionnaire.

Perception and behavior are different questions

The behavioral part of the theory asks whether perceived influence on others is associated with a relevant response, such as support for limiting a message or an intention to correct it. A meta-analysis has examined that question, and survey research has found associations under particular message conditions.[^xu][^sunbehavior]

An association is not a guaranteed sequence. A perception gap does not automatically lead to censorship, sharing, avoidance, voting, or any other action. It also cannot decide whether a restriction would be justified. Those are separate empirical and ethical questions.

Advertising provides another context boundary. An advertising-focused meta-analysis found a distinct pattern and smaller perceptual effects than communication research more generally.[^eisend] One label should not erase differences among messages, audiences, and outcome measures.