Self-Serving Bias: Why Wins Feel Earned and Losses Feel Unlucky

The bias is an asymmetry in explanation: the same person, asked why things went well, reaches for causes inside themselves, and asked why things went badly, reaches for causes outside. Two kinds of account have been offered for it, and the evidence supports both to a degree.

The motivational account says that people protect and enhance their view of themselves. Taking credit for a success feels good, and owning a failure feels bad, so the explanation that preserves self-regard wins out. Campbell and Sedikides's 1999 meta-analysis found that the bias grows when the self is under threat, which is what a motivational account predicts: when a failure matters more, people work harder to place its cause elsewhere.

The cognitive account, which Miller and Ross set out in 1975, says that the asymmetry can arise without any wish to feel good. People usually expect to succeed and usually try to succeed, so success matches both expectation and effort and is naturally credited to them, while failure is unexpected and gets attributed to whatever was unusual about the situation. On this view the bias is a by-product of how people reason about causes, and it would appear even in someone with no stake in the outcome.

The 2004 meta-analysis by Mezulis and colleagues gives the clearest picture of how general the pattern is. Across 266 studies the average effect was large, and the bias appeared in almost every kind of sample. It was largest in children and in older adults, smaller in samples from Asian cultures than in US samples, and much smaller or absent in people with depression, whose attributions tend to run the other way. Those differences are the main argument that motivation is at least part of the story, since a purely cognitive mechanism would not be expected to vary so much with mood and culture.

The self-serving bias was not discovered in one experiment. By the early 1970s dozens of studies had asked people to explain their performance on a task and found that success was attributed internally more often than failure. The first landmark is a review rather than an experiment: Dale Miller and Michael Ross's 1975 paper in Psychological Bulletin, "Self-serving biases in the attribution of causality: Fact or fiction?"

Miller and Ross went through the published studies and separated two claims that had been run together. The self-enhancing claim is that people take more credit for success than they deserve; the self-protective claim is that people deny responsibility for failure. They found fairly consistent evidence for the first and weak, inconsistent evidence for the second, and they argued that the self-enhancing pattern could be explained without any motive to protect self-esteem: people expect to succeed, they see their effort covary with success, and so they credit success to themselves on reasonable informational grounds. Gifford Bradley answered in 1978 with a reexamination that argued the motivational reading still fit the data better once the conditions of the studies were taken into account.

The modern reference is the meta-analysis by Amy Mezulis, Lyn Abramson, Janet Hyde and Benjamin Hankin, published in Psychological Bulletin in 2004. They pooled 266 studies that yielded 503 independent effect sizes and found an average effect of d = 0.96, a large bias by the usual conventions. The bias was present in nearly all samples. It varied with age, being largest in children and older adults; with culture, at d = 1.05 for US samples, 0.70 for other Western samples and 0.30 for Asian samples; and with psychopathology, at d = 0.21 for samples with depression, 0.46 for anxiety and 0.55 for attention-deficit/hyperactivity disorder. The authors read the pattern as a general positivity bias in attribution that circumstances and conditions can dampen.

The pattern shows up wherever people explain outcomes they were part of. The rows below are illustrations of the pattern, not findings from the studies cited on this page.

The research on reducing the bias is thinner than the research documenting it, so the suggestions below follow from how the bias works rather than from trials that tested them.

• Decide the causes before you know the outcome. A written prediction of what will matter, made before the result is in, cannot be quietly rewritten afterward.
• Use one list of causes for both outcomes. If luck, timing and other people are allowed to explain a failure, they have to be allowed to explain a success too, and the reverse.
• Ask the mirror question. For a failure, ask what you would say if a colleague had done exactly the same thing. For a success, ask what share of it you would grant a colleague.
• Separate the account from the self-assessment. Explaining a failure by the situation is sometimes correct; the problem is only when the explanation is chosen because of how it feels. Writing the explanation first and the feeling second helps keep them apart.
• Treat the bias as a default, not a flaw. The meta-analysis found it in nearly every kind of sample, with its size varying by age, culture and clinical condition, and a sample average says nothing about how far any one person shows it. The goal is accuracy about causes, not a habit of blaming yourself.

The self-serving bias is one of several biases in how people explain behavior, and the neighbors differ in whose behavior is being explained and what flips the explanation.

• It is not the fundamental attribution error, which is the tendency to explain other people's behavior by their character and to underweight their situation. The self-serving bias is about one's own outcomes, and it changes direction with success and failure rather than always pointing at the person.
• It is not the actor-observer bias, the asymmetry between explaining one's own behavior by the situation and other people's by their traits. That contrast is between actor and observer; the self-serving contrast is between a good outcome and a bad one for the same actor. The third step of the demo touches on it.
• It is not optimism bias, the expectation that good things are more likely to happen to you than the odds warrant. Optimism bias is about the future; the self-serving bias is about explaining the past, though the expectation of success is one of the cognitive reasons Miller and Ross gave for it.