Pessimism Bias: When You Expect Worse Than What Arrives

Pessimism bias is measured, not diagnosed, and the measurement it shares with the optimism bias runs two ways; this page follows the pessimistic one. The comparative method asks people to rate their own chances of an event against other people's, which is what the demo above does. The outcome method asks people to predict what will happen to them and records what did, the stricter test because it has a real answer. Both methods find that most people expect better than they get; a pessimistic tilt is the minority result in both.

Comparisons with the average carry their own logic, weaker than it looks. A majority can sit on one side of a mean and be right, because a mean is pulled by the extremes: if nine people in ten face a one in two chance of a setback and the tenth faces none, the mean is 45 percent and nine of the ten are correctly above it. A group that is not a fair sample of its comparison population does the same. Scoring a comparative rating needs a reference population the rater belongs to and real outcome information, and this exercise has neither, which is why the result screen reports where your answers sat rather than whether they were right.

Low mood is the characteristic most consistently tied to the pessimistic direction. Strunk, Lopez and DeRubeis had 153 people predict the likelihood of ordinary events in their own next month and then report what happened. People with few depressive symptoms predicted a month slightly better than the one they got, people with mild symptoms predicted it accurately, and people with higher symptom scores predicted it worse than it turned out, a correlation of about minus .35 between symptoms and the direction of the error. Sharot's review of the belief-updating work adds a mechanism: most people revise expectations more readily on good news than bad, and that asymmetry is reduced or reversed in depression.

The best-known claim in this area needs careful handling. The idea that depressed people see reality more accurately, called depressive realism after Alloy and Abramson's 1979 experiments, has not held up well. Moore and Fresco pooled 75 studies with 7,305 participants in 2012 and found an overall effect of about minus .07, negligible, with both dysphoric and nondysphoric participants showing a positive bias. Dev, Moore, Johnson and Garrett ran a direct test in 2022, found no relationship between depressive symptoms and illusory control or overconfidence, and concluded that the effect is not robust to replication. Pessimistic expectations are real and measurable; accuracy is not what they are made of.

Lauren Alloy and Lyn Abramson published "Judgment of contingency in depressed and nondepressed students: Sadder but wiser?" in the Journal of Experimental Psychology: General in 1979. Students were sorted by depression questionnaire scores and then given a task in which pressing or not pressing a button was followed by a green light, and asked afterward how much control they had had over the light. In the conditions that mattered, the light was unrelated to what the student did.

Students in the nondepressed group reported substantial control over an outcome they could not influence, and they did so most strongly when the outcomes were frequent and when the light meant winning money. The students in the depressed group judged their lack of control accurately in those conditions. The title's question, whether sadder meant wiser, came from that asymmetry, and the phrase depressive realism followed from it.

The comparative method has its own founding study. Neil Weinstein asked 258 college students in 1980 to estimate how much their own chances of 42 life events differed from their classmates', and found a systematic optimistic tilt whose size varied with how desirable, how probable and how controllable each event seemed. That paper is the origin of the finding that most people place themselves on the good side of average, which is the baseline a pessimism bias departs from.

What the field has learned since is mostly about limits. The accuracy advantage reported in 1979 was found with a laboratory contingency task and a questionnaire cutoff rather than a clinical diagnosis, and it has not generalized: the 2012 meta-analysis found the effect negligible overall, and the 2022 replication did not find it at all. Strunk and colleagues' prediction study is the clearer evidence for the pessimistic direction itself, because it compared predictions with what actually happened rather than with a normative standard.

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

A forecast is a claim that can be checked, which is what makes a tilt in either direction correctable.

• Write predictions down with dates and probabilities, then score them later. A kept record is the only way to find out which direction your forecasts actually lean.
• Look up the base rate before judging your own case. Most questions of the form how likely is this have a published answer for people in general.
• Separate the probability from the consequence. Pessimistic forecasts often combine a small chance with a large cost; stating the two numbers apart makes both easier to judge.
• Ask what you know that the average person does not. A rating away from average is justified by specific facts and not by a feeling, and listing the facts shows which you have.
• Check the mood you are forecasting in. Expectations shift with mood, so a prediction made on a bad day is worth revisiting on an ordinary one.
• Treat a prediction as a test rather than a verdict: send the message, make the application, and let the outcome score the forecast.

This page is educational. Pessimism bias is a pattern in judgment, not a diagnosis, and persistent low mood is a separate matter that no demo on a website can assess.

Pessimism bias is a direction of error in expectations about your own future. Its neighbors describe different things.

• It is not the optimism bias, which is the same comparison running the other way and is much more common. The two are not simply one scale read backwards: a pessimistic lean shows up in particular domains and conditions, so the two directions are studied separately.
• It is not negativity bias, the greater weight bad information carries compared with equally strong good information. That is about how input is weighted now; pessimism bias is about what is predicted next.
• It is not declinism, the belief that things in general are getting worse. Declinism is a judgment about the world and the past; pessimism bias is a forecast about yourself.
• It is not depression. Depressive symptoms are the characteristic most reliably associated with pessimistic predictions, but a pessimistic forecast is a single judgment, and the claim that low mood brings greater accuracy has not survived replication.