Actor-Observer Bias: Your Reasons Are Situations, Theirs Are Traits

Jones and Nisbett proposed the asymmetry in 1971 and gave it two sources. The first is information. You know your own history: the jobs you considered, the mood you were in, the reasons the obvious choice was closed. For a friend you have a few scenes and a general impression, and when the situation is thin the person fills the gap. The second source is perspective. When you act, your attention is on the task and the room; the situation is literally what you see. When you watch someone else act, the person is in the foreground and the situation is the backdrop, and the foreground tends to be read as the cause.

Both sources predict the 1973 pattern: a student gives more reasons about the subject for his own choice of major than for his friend's, while reasons about the chooser stay about the same. Storms showed in the same year that the perspective part can be reversed mechanically: actors shown a videotape of themselves from the observer's angle shifted their explanations toward their own dispositions, and observers who saw the scene from the actor's angle shifted toward the situation. Where the camera points is a large part of what gets blamed.

The textbook version of the bias is stronger than the evidence. Malle's 2006 meta-analysis gathered 173 published tests of the classic asymmetry and found average effect sizes between d = -0.016 and d = 0.095, close to zero. The asymmetry did appear under particular conditions: an actor portrayed as highly idiosyncratic, hypothetical events, intimates explaining each other, and free-response explanations coded as they were in 1973. Its strongest moderator was valence. For negative events the classic pattern held; for positive events it reversed, with people claiming their own successes and crediting others' to circumstances. Averaged across good and bad events, the asymmetry disappeared.

That result changes what the bias is: not a general law that actors see situations and observers see traits, but a pattern that shows up most clearly when the behavior being explained is unflattering, where it overlaps with self-serving explanation. The demo on this page uses a positive or neutral event, a career, the kind of case where the 2006 analysis found the classic pattern weak or reversed, so its live split is a test of that as much as of the 1973 finding.

Edward Jones and Richard Nisbett set out the hypothesis in a 1971 essay, "The actor and the observer: Divergent perceptions of the causes of behavior," published as a General Learning Press monograph. The first experimental tests came two years later, from Nisbett, Craig Caputo, Patricia Legant and Jeanne Marecek, in a 1973 Journal of Personality and Social Psychology paper reporting three studies.

In the first, 33 pairs of female students at Yale took part in a session in which one, the actor, was asked to volunteer time as a weekend hostess for a fundraising event while the other watched. Observers who saw the actor agree expected her to volunteer for an unrelated cause later more than the actors themselves did, treating one act of compliance as a trait.

The second study is the one the demo reproduces. Male Yale undergraduates wrote short paragraphs explaining why they liked the woman they had dated most in the past year and why they had chosen their major, then wrote the same two paragraphs about their best friend. Each reason was coded as an entity reason, a feature of the woman or the subject, or a dispositional reason, a feature of the chooser. The 23 students gave 1.52 entity reasons on average for their own choice of major and 0.43 for their friend's, while dispositional reasons were about equally common in both, 1.83 against 1.70. The difference was relative rather than a majority either way: even for their own major the students gave slightly more dispositional than entity reasons. A follow-up ruled out a quirk of writing by having fresh participants rate ready-made sentences, with the same result.

In the third study, students filled in a trait questionnaire for themselves, their best friend, their father, an acquaintance and the television newscaster Walter Cronkite, with the option of answering "depends on the situation." They chose that option most often for themselves and ascribed the fewest fixed traits to themselves and the most to Cronkite. The authors were careful to say that the evidence did not establish the hypothesis in general, a caution that Malle's 2006 meta-analysis bore out.

The rows below are illustrations, not findings from the studies cited on this page.

The bias runs on what you can see and what you know, so the counters add information and change the angle.

• Ask the situational question first. Before concluding that someone is rude or careless, ask what was happening around them, and answer as fully as you would for yourself.
• Borrow your own excuse. If you would explain the act by circumstance had you done it, try that explanation for the other person before reaching for a trait.
• Change the camera angle. Storms's 1973 videotape study showed that watching yourself from outside shifts your explanations toward disposition; imagining the scene from the other person's seat does the reverse.
• Check the valence. The asymmetry is strongest for failures and reverses for successes, so expect trait explanations to come too easily for others' bad behavior and credit too easily for your own good behavior.
• Collect the history. For people you know only slightly, the situational record is thin. Asking about it is the cheapest correction.
• Hold the trait theory loosely. The students in 1973 answered "depends on the situation" most often about themselves and least about a newscaster they knew only from television. Everyone's behavior depends on it.

The actor-observer bias compares two points of view on the same behavior. Its neighbors describe one point of view, or a different comparison.

• It is not the fundamental attribution error, which is the observer's half on its own: explaining other people's behavior by their character and underweighting their situation. The actor-observer bias is the gap between that and how the same people explain themselves.
• It is not the self-serving bias, which credits your successes to yourself and your failures to circumstances. The two overlap for failures and diverge for successes, which is why the pattern reverses for positive events.
• It is not trait ascription bias, the finding from the third 1973 study that people see their own behavior as more variable across situations than other people's. That concerns how many fixed traits are ascribed, not how a particular act is explained.
• It is not naive realism, the conviction that you see the world as it is and that others who disagree are biased or uninformed. Naive realism is about perception of facts; the actor-observer bias is about explanation of behavior.