In short
The ultimate attribution error is the tendency to explain an out-group member's bad behavior by their character and their good behavior by luck or special circumstances, while reversing that pattern for one's own group. Thomas Pettigrew named it in 1979 as the group-level version of a familiar individual bias, and it is studied in intergroup conflict, hiring, policing and reporting on rival nations.
How the ultimate attribution error works
The ultimate attribution error is about group membership doing the explaining. Its better-known relative, the fundamental attribution error, says that people reach for character over circumstance when explaining anyone's behavior. Pettigrew's addition in 1979 was that the reach is selective: whether character or circumstance gets the credit depends on which side of a group line the actor stands on, and on whether the act was good or bad.
Four moves make up the pattern. A bad act by an out-group member becomes evidence of what those people are like. The same act by an in-group member becomes a bad day, a difficult road, a one-off. A good act by an in-group member shows character. A good act by an out-group member gets explained away: they were lucky, they had unusual advantages, or they are the exception that proves the rule. Pettigrew drew the idea from Gordon Allport's analysis of prejudice, where the exception is absorbed rather than allowed to change the generalization.
What keeps the pattern running is that it is self-sealing. Because out-group members' good behavior is discounted and their bad behavior is stored as character, experience with the out-group rarely updates the view of it. Conflict over real resources makes this worse: the depots in the demo above compete for the same bonuses, and competition sharpens the line that decides which explanation gets used.
The evidence is real but qualified, and overselling it is easy. Hewstone and Ward failed to find the full pattern in both groups they studied in Southeast Asia in 1985: Malay participants in Malaysia showed the ethnocentric pattern, while Chinese participants showed little of it and in some comparisons favored the out-group. Reviewing the literature in 1990, Miles Hewstone found support for intergroup attribution bias, but as a matter of degree that varies with the group's status, the kind of behavior and the setting, rather than a rule that applies everywhere. It is a tendency in groups in conflict, not a fixed feature of human judgment.
The original study
Donald Taylor and Vaishna Jaggi published "Ethnocentrism and causal attribution in a South Indian context" in the Journal of Cross-Cultural Psychology in 1974, five years before Pettigrew gave the pattern its name. Hindu participants in southern India, where Hindu and Muslim communities had a history of tension, read short accounts of a person doing something socially desirable or undesirable. In each account the actor was identified as Hindu or Muslim.
The accounts described everyday dealings between two people, some of the acts socially desirable and some undesirable. For each account participants chose among explanations that differed in whether they put the cause in the actor and their group or in the circumstances.
The choices split by the actor's group. Hindu participants picked the internal reason more often when a Hindu had behaved well and when a Muslim had behaved badly, and reached for circumstances when a Hindu behaved badly or a Muslim behaved well. That crossover, the same act getting different explanations according to the actor's group, is the finding the page is about.
Read the diagram and evidence as text
Matched plant-outcome scenes retain the same actor silhouette but change the group badge. Attribution arrows end at the actor for an in-group good act and an out-group bad act; the reverse scenes emphasize the surrounding situation. Plant scenes and badges are illustrative, not study stimuli; arrow destinations encode an ordinal tendency, not exclusive explanations.
Who gets the credit, who gets the excuse
Direction only: the paper reports the pattern and the page quotes no percentage from it.
Same kinds of everyday acts, same choice between the actor and the circumstances; only the actor's group changes
In-group (Hindu) actor does something desirable: The actor as cause: chosen more often
In-group (Hindu) actor does something undesirable: The actor as cause: chosen less often; the circumstances instead
Out-group (Muslim) actor does something desirable: The actor as cause: chosen less often; the circumstances instead
Out-group (Muslim) actor does something undesirable: The actor as cause: chosen more often
| Condition | Value and unit | Evidence quote |
|---|---|---|
| Hindu (in-group) actor behaving well | chosen more often the actor as the cause, ordinal | chose among explanations locating the cause in the actor or in the circumstances; they picked the actor more often for in-group actors behaving well and for out-group actors behaving badly |
| Muslim (out-group) actor behaving badly | chosen more often the actor as the cause, ordinal | chose among explanations locating the cause in the actor or in the circumstances; they picked the actor more often for in-group actors behaving well and for out-group actors behaving badly |
| Hindu (in-group) actor behaving badly | chosen less often the actor as the cause, ordinal | chose among explanations locating the cause in the actor or in the circumstances; they picked the actor more often for in-group actors behaving well and for out-group actors behaving badly |
| Muslim (out-group) actor behaving well | chosen less often the actor as the cause, ordinal | chose among explanations locating the cause in the actor or in the circumstances; they picked the actor more often for in-group actors behaving well and for out-group actors behaving badly |
Thomas Pettigrew's "The ultimate attribution error: Extending Allport's cognitive analysis of prejudice" appeared in Personality and Social Psychology Bulletin in 1979. It is a theoretical paper rather than an experiment: Pettigrew argued that the attribution literature of the 1970s had described the mechanism by which prejudice maintains itself, and that the group-level case deserved its own name. The empirical record since has been mixed in the way described above, which is why this page quotes the direction of Taylor and Jaggi's result rather than treating it as a settled quantity.
Everyday examples
The error needs two things: a group line that matters to the observer, and an act that could be explained either way. The rows below are illustrations, not findings from the studies cited on this page.
| Situation | What the bias does | What helps |
|---|---|---|
| A missed deadline by another department | Reads as the way that team works, while your own team's missed deadline was the client's fault | Write down the circumstances of both misses before explaining either |
| Driving, with a bumper sticker or a plate from elsewhere | One bad maneuver becomes a fact about drivers from there | Count how many drivers from that place did nothing notable in the same hour |
| Coverage of a rival country's government | Its concessions are read as tactics, its aggression as national character | Read one account of the same events written from the other side |
| A hiring panel reviewing a candidate from an unfamiliar background | A gap in the resume becomes a character question rather than a circumstance | Score written criteria before discussing the person, and ask candidates about gaps |
| Fan judgment of a foul in a match | The other side's foul was intent, your side's identical foul was momentum | Watch the replay with the shirts out of frame, or swap the teams in the description |
How to counter it
Nothing in the studies cited here tested a remedy, so the moves below follow from the mechanism rather than from measured results.
- Swap the actor. Ask what explanation you would reach for if the same act had come from the other side of the line. A difference in the answer is the bias, visible from the inside.
- Ask what the situation was before asking what the person is like. In Taylor and Jaggi's materials the situational reasons were sitting right there on the page; the group of the actor decided whether they got picked.
- Notice the exception move. When an out-group member does something good and the explanation is luck, advantage, or that they are not typical, that is the error in its less visible half.
- Use criteria you wrote earlier. Written, specific standards in hiring, grading and discipline leave less room for group membership to supply the explanation.
- Reduce the competition where you can. The pattern is strongest between groups contesting the same resources, so shared goals and mixed teams change the conditions rather than the attitudes.
- Expect it in your own group too. The bias includes excusing your own side, which is the part that feels like fairness.
What it is not
Group-level attribution has several close neighbors, and the differences are what make them testable.
- It is not the fundamental attribution error, which is the general preference for character over circumstance when explaining one person's behavior. The ultimate attribution error adds the group line that decides which explanation is applied.
- It is not the group attribution error, which is about treating one member's view as the group's or the group's decision as every member's view. That one concerns what gets generalized, not what gets credited.
- It is not self-serving bias, in which a person claims credit for their own successes and blames circumstances for their failures. Swap "a person" for "a group" and the two patterns rhyme, but self-serving bias needs no out-group at all.
- It is not stereotyping or prejudice themselves. Those are the content, the beliefs and feelings about a group; the ultimate attribution error is one of the mechanisms by which that content survives contact with evidence.
Tests and tools that use this
Related biases
Ultimate Attribution Error FAQ
What is the ultimate attribution error?
It is the tendency to explain an out-group member's bad behavior by their character and their good behavior by luck or special circumstances, while giving one's own group the opposite treatment. Thomas Pettigrew named it in 1979, extending Gordon Allport's account of how prejudice survives contact with contrary evidence. The pattern was documented by Taylor and Jaggi in 1974 with Hindu and Muslim actors in short written scenarios.
How is it different from the fundamental attribution error?
The fundamental attribution error is the general tendency to explain behavior by what someone is like rather than by their situation. The ultimate attribution error says that tendency is applied selectively: character gets the credit when an out-group member does something bad or an in-group member does something good, and circumstances get it the other way around. The group line is the addition.
How does the demo on this page work?
You were assigned one of two versions of the same incident, differing only in whether the van carried your own depot's sticker or the rival depot's, and you chose character or circumstance. Once each version has 50 responses, the page shows both splits side by side, which is the comparison the research makes. One answer from one person cannot show a bias; the difference between the two groups of visitors is where it would appear.
Does the ultimate attribution error always happen?
No, and the honest summary is that it depends. Hewstone and Ward found the pattern among Malay participants in 1985 but not among Chinese participants in the same region, and Hewstone's 1990 review concluded that intergroup attribution bias varies with the status of the groups, the kind of behavior and the setting. It is a tendency in groups in conflict rather than a universal rule.
Why does the demo use delivery depots instead of real groups?
Because a web page cannot know which real groups matter to you, and inventing a membership for you is the only way to keep the comparison fair across visitors. An assigned workplace identity is a weak version of the manipulation: the original studies used groups with real history between them. Expect any split between the two versions here to be smaller than the research describes.
Can the ultimate attribution error be reduced?
The studies cited here measured the bias rather than testing cures, so the suggestions are reasoned rather than proven. Swapping the actor's group in your own head, listing the situational facts before judging the person, and using written criteria in decisions all attack the point where group membership enters. Shared goals and mixed teams change the competitive conditions that make the pattern strongest.
References
- Taylor, D. M., & Jaggi, V. (1974). Ethnocentrism and causal attribution in a South Indian context. Journal of Cross-Cultural Psychology, 5(2), 162-171. doi.org
- Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport's cognitive analysis of prejudice. Personality and Social Psychology Bulletin, 5(4), 461-476. doi.org
- Hewstone, M., & Ward, C. (1985). Ethnocentrism and causal attribution in Southeast Asia. Journal of Personality and Social Psychology, 48(3), 614-623. doi.org
- Hewstone, M. (1990). The 'ultimate attribution error'? A review of the literature on intergroup causal attribution. European Journal of Social Psychology, 20(4), 311-335. doi.org
Cite this source
Psychology.com. (2026, October 23). Ultimate Attribution Error. Psychology.com. https://psychology.com/biases/ultimate-attribution-error