Fundamental attribution error

type: effect · origin: study · evidence: supported · themes: how we judge others

Claim assessed: We explain other people's behaviour by who they are, not by the situation they are in.

Published 5 October 2026 · how we work

In short

The fundamental attribution error is the tendency to put someone’s behaviour down to their character and to underestimate the situation they are in. The first experiment to show it, by Jones and Harris (1967), turned out differently from what the authors expected. Lee Ross (1977) gave it its name. The effect appears in many kinds of experiments, but not in everyone, it is weaker in East Asia, and a related idea about how we judge ourselves did not survive a meta-analysis.

What it says

The original experiment. Jones and Harris gave participants essays for or against Fidel Castro. Some were told the writer had chosen the position freely, others that it had been assigned by a debate coach. When the writer had chosen freely, it made sense for participants to think the essay showed what the writer believed. The surprise was that they drew the same conclusion, though more weakly, when they knew the position had been assigned. Gilbert and Malone (1995) note that the result replicated in many variants that ruled out simple explanations, for example that participants had misunderstood the instructions. We haven’t read the original and describe it from their account.

The name. Ross (1977) noticed that the same mistake explains why social psychology experiments are surprising: people underestimate how powerful a situation is. He called it the fundamental attribution error. Gilbert and Malone prefer the term correspondence bias: we infer a trait that “corresponds” to the behaviour.

The quiz game. Ross, Amabile and Steinmetz (1977) randomly split participants into those who asked questions and those who answered them. The questioners picked questions from their own knowledge, so the answerers often got them wrong. Observers still rated the questioners as more knowledgeable, although the role gave them an obvious advantage. The answerers reached the same conclusion. The error was larger for them and for the observers than for the questioners.

Why it happens. Gilbert and Malone describe four different causes that can produce the same result:

• You don’t see the situation. The pressure of an audience or an instruction given beforehand doesn’t show in the behaviour.
• You underestimate how strong it is. You imagine how you would react and assume everyone would do the same.
• The situation changes what you perceive. If you know the topic is embarrassing, you see more nervousness in the same behaviour.
• You stop correcting. The first impression of character comes quickly and without effort. Correcting for the situation takes attention. Gilbert and Malone describe experiments in which participants busy with another task no longer made the correction, even when the information about the situation was exactly what they had to memorise.

A recent replication. Nalipay (2017) repeated the first experiment with 145 students in the Philippines, using an essay on the reproductive health law, four months after it was passed. The difference came almost entirely from the position in the essay. Whether the writer had chosen it or not made no significant difference.

Differences between people. Scopelliti et al. (2018) built a questionnaire that measures how prone someone is to this error and tested it in ten studies with more than 2,700 Americans recruited online. The tendency was stable in the same person after three weeks and was only weakly related to intelligence. Those most prone to it blamed people more for accidents, believed confessions obtained under pressure more readily, and took less account of how hard the job was when evaluating employees. When an investment fund’s performance was put in a table next to the market’s, even they stopped favouring the fund that had merely ridden a good market.

Example

A colleague is late for the third meeting in a row and someone says “he’s disorganised”. Maybe he is. But maybe his child has moved to a nursery on the other side of town, something nobody in the meeting room can see. The example is an illustration, not a study cited here.

How to apply it

The steps below are a practical way of applying it that we propose, based on the studies cited.

1. Ask what you can’t see. The first cause Gilbert and Malone describe is that the situation simply isn’t visible. Before drawing a conclusion about someone, ask what pressures they might be under.
2. Don’t use yourself as the yardstick. “I would never do that” is a poor estimate of what most people would do in that situation.
3. Put off judgements when you’re busy. Correcting for the situation is the step that drops out first when your attention is elsewhere.
4. Put the situation next to the result. When you evaluate someone by results, compare them with others who worked under the same conditions. In Scopelliti et al.’s study, a table with the market’s performance next to the fund’s helped precisely those most prone to the error.
5. Don’t over-correct. If the person chose the situation themselves, the behaviour may well say something about them.

Limits and nuances

• “Ourselves by the situation, others by character” isn’t confirmed. The idea that we explain our own behaviour by the situation and other people’s by their character is related to the fundamental attribution error, but it is a different claim. Malle (2006) analysed 173 published studies and found average effects between d = −0.02 and d = 0.10, practically zero. The asymmetry appeared only under certain conditions, for example for negative events, and for positive events it reversed.
• Not everyone does it. Bauman and Skitka (2010) tested a representative sample of American adults rather than students. Only 53% showed the effect. It was more common among people who believe others behave the same way in every situation.
• Culture matters. Choi, Nisbett and Norenzayan (1999) review studies showing that East Asians make the error less often when the situation’s constraints are obvious. The difference is not that they don’t think in terms of traits, but that they give more weight to context.
• The lab probably exaggerates how often it happens. Gilbert and Malone note that experiments are built precisely around situations whose power is underestimated. They show that the error occurs, not how often it occurs in everyday life.
• Sometimes it isn’t a mistake. Gilbert and Malone give three cases: the person chose the situation, you always see them in the same situation, or the situational pressure changes nothing because they would have acted the same way anyway. In these cases, judging by character predicts well.
• The Philippine replication is small. It has 35–38 participants per group and a single sample. It found a larger error than the original, in which an assigned position counted much less than a chosen one.
• Excluded sources. Several recent studies on this error have Francesca Gino among their authors, who has had several papers retracted over problems with the data. We don’t use them.
• What we read. Gilbert and Malone (1995), from Daniel Gilbert’s Harvard site, Nalipay (2017), from the Psychological Association of the Philippines’ site, and Scopelliti et al. (2018), the accepted manuscript from the City, University of London repository, we read in full. For Ross, Amabile and Steinmetz (1977), Choi et al. (1999), Malle (2006) and Bauman and Skitka (2010) we read the abstract. We haven’t read Jones and Harris (1967) or Ross (1977); we describe them from Gilbert and Malone.

Sources

• Edward E. Jones, Victor A. Harris (1967). The attribution of attitudes
• Lee Ross (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution process
• Lee Ross, Teresa M. Amabile, Julia L. Steinmetz (1977). Social roles, social control, and biases in social-perception processes
• Daniel T. Gilbert, Patrick S. Malone (1995). The correspondence bias
• Incheol Choi, Richard E. Nisbett, Ara Norenzayan (1999). Causal attribution across cultures: Variation and universality
• Bertram F. Malle (2006). The actor-observer asymmetry in attribution: A (surprising) meta-analysis
• Christopher W. Bauman, Linda J. Skitka (2010). Making attributions for behaviors: The prevalence of correspondence bias in the general population
• Ma. Jenina N. Nalipay (2017). Correspondence bias in the attribution of political attitudes (A replication of Jones and Harris's (1967) experiment on correspondence bias)
• Irene Scopelliti, H. Lauren Min, Erin McCormick, Karim S. Kassam, Carey K. Morewedge (2018). Individual differences in correspondence bias: Measurement, consequences, and correction of biased interpersonal attributions

See also: Bystander effect, Confirmation bias, Curse of knowledge, Halo effect, Milgram experiment, Outcome bias, Stanford prison experiment