Why Doesn’t a Good Outcome Necessarily Mean a Wise Decision?

When a decision produces a good result, we readily call it wise. If the same decision ends badly, it can suddenly look careless. This is outcome bias: once the ending is known, the result is given too much weight in judging the quality of the earlier decision.

In experiments published in 1988, Baron and Hershey asked participants to evaluate medical decisions. The information and decision process stayed the same while the eventual success or failure of surgery changed, and ratings moved with the outcome. A preregistered 2023 study replicated the signal and direction of one of those experiments. These studies mostly use written scenarios, however. They establish an average influence on judgement, not that every real-world evaluation works this way.

Outcomes are not always irrelevant. Repeated failure may reveal a missing factor in a model, an implementation problem or a poor original probability estimate; that is new evidence. Bias arises when an outcome contains a substantial element of luck but a single ending is still treated as proof that the earlier reasoning was good or bad. If a forecast gives an 80 per cent chance of rain and the day stays dry, that one event alone does not show that the forecast was poor.

Outcome bias is related to hindsight bias but distinct from it. Hindsight makes the past seem as though it should have been predictable; outcome bias changes the score we give the decision because the ending was good or bad. A sounder review first hides the ending and asks what was known, what the probabilities were and which alternatives were reasonable. It then reveals the result and asks what genuinely new evidence it supplies. That avoids rewarding luck without treating every unlucky result as incompetence.

https://pubmed.ncbi.nlm.nih.gov/3367280/
https://pmc.ncbi.nlm.nih.gov/articles/PMC12372742/