The Third-Person Effect
A reader's summary of the belief that media messages change other people's minds more than my own — covering Davison's 1983 paper and its wartime anecdote, how the effect was measured, the reversal for messages we like, its tie to support for censorship, and its place in today's arguments about feeds and persuasion.
The term at a glance
I know the ad did not work on me. I am less sure about everyone else. The third-person effect describes that asymmetry: people estimate that other audiences are more influenced by a message than they are themselves. As with its cousin the hostile media effect, the interesting part is not who is right about the message but what each person assumes about everyone else's mind.
Origin
W. Phillips Davison, a sociologist who studied public opinion and propaganda, named the effect in “The Third-Person Effect in Communication,” published in Public Opinion Quarterly in 1983. He had noticed the pattern in anecdotes before he tested it: people who talked about propaganda, whether enemy leaflets or commercial advertising, seemed always to be worried about someone else's exposure. His broad formulation was that people tend to overestimate the influence of mass communication on the attitudes and behaviour of others.
History and context
The paper landed in a field that had spent decades arguing about how strong media effects really are, from the 1930s fears of mass propaganda to the 1940s “limited effects” studies that found people mostly reinforced what they already believed. The third-person effect added a different question: whatever the real effect is, what do people believe it to be, and what do they do about that belief? Over the following decades researchers tested it on advertising, news, pornography, violent games and, more recently, social media and fake news. It sits alongside other findings about distorted social perception, such as the false consensus effect, and shares some ground with the fundamental attribution error, in which I explain my own behaviour by the situation and other people's by their character.
Main ideas
Others are influenced; I am not
The third-person effect is the belief that persuasive or harmful media content has a larger effect on other people (the “third persons”) than on oneself or on people like oneself. Ask someone how much an advertisement, a news story or a viral post affects them, then how much it affects the public, and the second number is usually higher. It is a perception about influence, not a measurement of it.
The origin: a leaflet drop
W. Phillips Davison, a Columbia sociologist, introduced the idea in a 1983 paper in Public Opinion Quarterly. He opened with an episode from the Pacific war: Japanese propaganda leaflets dropped on a unit of Black American soldiers on Iwo Jima, which appeared to have little effect on the troops. Yet the day after the drop, the unit was withdrawn, apparently because white officers feared the leaflets would sway the soldiers. Davison read the episode as an officer reacting to the effect he expected on others, not to any effect that had occurred.
Two parts: the perception and the behaviour
Davison's hypothesis had a perceptual half (I think others are more affected than I am) and a behavioural half (because of that, I act, for example by supporting restrictions on the content). Later researchers have tested the two separately. The perceptual gap is the better-established half.
A robust finding, with limits
Reviews and meta-analyses, including Paul, Salwen and Dupagne's in 2000 and a later one by Sun, Pan and Shen in 2008, found the perceptual gap reliable across samples and media. It tends to be larger when the content is seen as negative or harmful, when the comparison group is distant (“the public”) rather than close (“my friends”), and when the person thinks the source is biased. It is smaller or absent for content that is neutral.
It reverses for messages we approve of
For content judged beneficial, such as public-service campaigns promoting seatbelts or discouraging smoking, studies often find the opposite pattern: people say they are more affected than others. This is sometimes called the first-person effect. The asymmetry suggests the bias is partly about self-image, since being swayed by a harmful message feels foolish and being swayed by a good one does not.
From perception to censorship
The behavioural half has support too. Studies of pornography, violent entertainment and rap lyrics in the 1990s, such as Albert Gunther's work in the Journal of Communication, found that the larger the gap a person perceived between others and themselves, the more they favoured restricting the content. The reasoning, whether or not it is conscious, runs: this is harmless to me, but others cannot handle it, so someone should limit it. The effect has also been linked to willingness to “do something” about media coverage of elections and polls.
Why it happens: the explanations
Proposed causes include self-enhancement, since it is flattering to see oneself as rational; the assumption that others are more gullible, which echoes the fundamental attribution error; limited information, since I know my own reasoning but only see other people's reactions from outside; and the pull of what the message seems built for, a poster addressed to an imagined average viewer. These can operate together, and studies have not settled their relative weight.
Online, everyone else is the target audience
Modern forms of the effect are easy to spot: “the algorithm is manipulating people,” “those ads work on others,” “people fall for anything on TikTok,” “they believe whatever the bots tell them.” Each sentence locates the susceptibility elsewhere. That does not make the concern wrong, since persuasion and recommendation systems are real and measurable, but it shapes who is blamed. It shows up in arguments over disinformation and in the moderation debates behind moral panics about new media.
Critique
- Self-reports are a weak instrument. Most studies rely on people rating their own and others' susceptibility on a scale. Some of the gap may come from the comparison itself, since an abstract “public” is easier to imagine as uniform than a person one knows well.
- Other people may in fact differ. A person who is informed about a topic may really be less swayed by a claim about it than a stranger who is not. The perceived gap is a bias only to the extent that it exceeds the real one, which is rarely measured directly.
- The link to censorship is correlational. People who think others are vulnerable also tend to be more worried about the content on other grounds, such as moral objection. Studies find the perception predicts support for restrictions even after controlling for some of these, but the causal direction is hard to establish.
- It can be used as a trump card. Saying that an opponent is just a victim of the algorithm explains away their view without engaging it, a move close to the ad hominem. The effect is a reason to check how much a message really changes behaviour, not a license to dismiss the people who hold a view.
Impact
A useful habit is to apply the question to oneself first: if I believe this content is dangerous to a general audience, what evidence do I have beyond my sense that they are less careful than I am? The answer may well be good evidence, from studies of real behaviour rather than from impression. The effect also explains why policy debates about media tend to feature protecting “others”, such as children, voters or the less educated, from messages the speaker considers manageable for themselves. Naming that structure does not settle whether a particular rule is justified. It does make it easier to ask who is being protected, from what, and on what evidence, which is the same discipline as asking for the actual numbers behind a fashionable fear.