Omission bias

The cleanest demonstration is a story about tennis. Spranca, Minsk and Baron gave people a scenario in which John, a club player, is about to face Ivan Lendl in a final worth real money. At dinner the night before, John remembers that Lendl is allergic to cayenne pepper, and that the house dressing contains it. In one ending, John recommends the house dressing. In another, Lendl orders it himself and John simply says nothing. Same knowledge, same hoped-for outcome, same stomach ache. Sixty-five percent of subjects judged the silence less immoral than the recommendation, and essentially nobody judged it the other way.

Then they moved from judging others to choosing. Subjects took the role of a physician, a patient, or a public health official facing an infection that carries a twenty percent chance of permanent brain damage. A treatment prevents that entirely, but the treatment itself causes the same damage in fifteen percent of cases. The right choice is the treatment, and most people chose it. But in thirteen percent of the comparisons, people rated doing nothing as the better decision, accepting the worse odds, and when the probabilities were reversed so that treating was safest, almost nobody made the equivalent mistake in the other direction. The justifications are the mechanism in plain sight. One subject wrote that the second choice had slightly less risk, but the blame would be yours if the damage occurred. Another appealed to letting nature take its course. Doing nothing has no author, so its harms feel like fate rather than like a decision.

The same team then took the finding to a harder case. Ritov and Baron asked people how deadly a hypothetical childhood disease would have to be before they would use a vaccine that itself carried a small fatal risk. The average answer, when the vaccine killed ten in a hundred thousand, was that the disease would need to kill over two and a half thousand. In a pilot study, one parent explained that if they vaccinated their child and the child died, they would feel more responsible than if they had not vaccinated and the child died, and added: sounds strange, I know. That is the bias described from the inside, by someone who can see it and holds it anyway.

Now the honest part, because this entry sits on the debated shelf for real reasons. First, the researchers themselves report that in most of the choice experiments, most people show no bias; the effect is carried by a substantial minority, which matters for policy and for markets but is not the same as a universal law. Second, the dramatic vaccination numbers were directly challenged by Connolly and Reb in 2003, who showed the response scales could only register reluctance in one direction and were unbounded above, so the striking averages were partly an artefact of measurement. Third, the effect moves with circumstances: Ritov, Hodes and Baron found that people in positions of explicit responsibility can show the opposite pattern, a bias toward action. A later preregistered meta-analysis across twenty-one samples found the asymmetry is real overall, moderated by role, perspective and outcome, which is roughly where the field rests: a genuine tendency, unevenly distributed, strongest when responsibility is informal and the harm is vivid.

The business version of this bias is an accounting asymmetry. Every organisation keeps a record of harms caused by action: the launch that flopped, the hire that failed, the acquisition that had to be written down, each with a name attached. Almost no organisation keeps a record of harms caused by inaction: the product not built, the candidate not hired, the market entered two years late, the price never raised. Both kinds of failure destroy value. Only one kind appears in anyone's file.

People respond to that asymmetry rationally. If acting and failing gets you named while not acting and failing gets recorded as prudence, the safest career strategy is to be the person who never causes anything. This is why cautious cultures do not actually reduce failure. They relocate it, from the visible column to the invisible one, where it compounds quietly because nobody is measuring it. The lawyers' version appears in the original paper: manufacturers are held liable for harms their products cause, and not at all for the products they declined to make.

There is a personal version, and it explains a pattern this library measures elsewhere. When you advise a friend, their action carries no authorship for you, so your advice is clean of the bias and tends to be bolder than your behaviour. Your own action, if it fails, would be unmistakably yours. So the gap between "go, and sooner than I went" and what you actually did is not hypocrisy. It is that their commission costs you nothing and yours costs you authorship. The related failure of judging decisions by how they landed is outcome bias. Why you reason better about their situation than your own is Solomon's paradox.

The practical countermeasure is to give inaction an author. Some firms log the deals they explicitly declined, with the reasoning and a review date, which converts a non-event into a decision that can later be scored. The moment "we chose not to" is written down with a name beside it, it competes on equal terms with "we did," and the invisible column starts to fill in.

Source: Spranca, Minsk and Baron, Omission and commission in judgment and choice, Journal of Experimental Social Psychology, volume 27, 1991, pages 76 to 105. The vaccination studies: Ritov and Baron, Reluctance to vaccinate: omission bias and ambiguity, Journal of Behavioral Decision Making, volume 3, 1990, pages 263 to 277. The methodological challenge: Connolly and Reb, Omission bias in vaccination decisions: Where's the omission? Where's the bias?, Organizational Behavior and Human Decision Processes, volume 91, 2003, pages 186 to 202. A preregistered meta-analysis by Yeung, Yay and Feldman later found an overall omission-commission asymmetry moderated by role, perspective and outcome type.

Thinking and Deciding

Jonathan Baron, 2023

The researcher's own book, and the fullest account of how judgment and decision making actually work, written by the man who ran the experiments in this entry. It is a university textbook rather than a trade book, so it is a heavier read than most on this shelf. It is also the one that will still be useful in ten years.