Action bias
Also known as: Bias for action, Do-something bias
The pull to do something rather than wait, even when waiting is better.
What it means
Action bias is the tendency to favor doing something over doing nothing, even when the action carries no expected benefit or a worse one than waiting. The mechanism is part psychological and part reputational: acting restores a felt sense of control and signals competence and effort, while standing still looks passive and invites blame if things go badly, so we intervene partly to be seen intervening. It shows up in goalkeepers diving at penalty kicks, doctors ordering tests and procedures of marginal value, investors trading through market noise, managers reorganizing teams, and legislators responding to every crisis. It is not always a bias — where situations move fast and moves are reversible, acting first is often correct — but it turns costly when the base rate favors patience, when the action is irreversible, or when its side effects exceed the original problem. It matters because most institutions reward visible action over quiet restraint, so the bias compounds instead of correcting.
Why it happens
The deeper mechanism is norm theory: regret attaches not to action as such but to whichever choice departs from what is normal in the setting. The abnormal choice is the one we replay afterwards, so it is the one that stings. This is why action bias and omission bias are not rival findings but the same machinery pointed in opposite directions. In vaccination, the jab is the conspicuous departure and parents shy from it; in goal, standing still is the departure and keepers avoid that instead. Feldman and Albarracin (2017) show that social norms drive whether action or inaction produces more regret. Change what a room treats as normal and the bias flips sign, which makes it a norm effect rather than a standing preference for movement.
What the evidence shows
Patt and Zeckhauser (2000) named the bias and tested it with surveys of environmental choices, finding people prefer fostering an improvement over preventing an equivalent deterioration even when the framing makes the choice arbitrary. The sharpest field evidence is Bar-Eli and colleagues (2007), who coded 286 penalty kicks: keepers dived left 49.3 percent of the time and right 44.4 percent, staying centre in only 6.3 percent. Yet staying centre stopped 33.3 percent of kicks, against 14.2 percent diving left and 12.6 percent diving right; pooling the two dives into a single jump action and testing that against staying centre, the difference is unlikely to be chance (chi-square(1, N=286)=5.33, p=0.02). A companion survey of 32 top professional goalkeepers confirmed the norm: 25 rated jumping the more normal response against 7 who rated staying centre so (chi-square(1, N=32)=10.13, p=0.001). Jumping is what a keeper is supposed to do. The regret half of the story is thinner. Ceiling effects made 17 of the 32 unusable, and among the 15 who remained, 11 judged conceding while rooted worse than conceding mid-dive against 4 the reverse: chi-square(1, N=15)=3.27, p=0.07, which the authors themselves describe as not significant at the 5 percent level and merely compatible with their hypothesis.
Limits and caveats
The famous example is weaker than its fame. Because keepers so rarely stay centre, the 33.3 percent figure rests on 18 kicks: the behaviour under criticism starves the comparison of data. The survey side is weaker still, and it matters more, because the regret asymmetry is the mechanism the norm-theory account here runs on. Ceiling effects gutted the sample — 15 of 32 keepers gave the maximum bad-feeling rating to every scenario — and the 11-to-4 split among the 15 usable responses lands at p=0.07, short of significance, as the authors concede. The feeling the explanation rests on is suggested, not established. The deeper issue is that a penalty is a game, not a lottery: the centre pays partly because keepers vacate it, and if they stopped diving, kickers would aim wider. Bar-Eli and colleagues concede that the mixed-strategy equilibrium lies outside their analysis, and later reframed the behaviour as socio-emotional rationality: keepers optimise a utility that includes how a choice feels and looks. The reputational story has conditions too. Across three pre-registered experiments, Fillon and colleagues (2023) found people judge inaction agents more competent and trustworthy after bad outcomes, unless negative prior outcomes had made acting the normal move.
Using it in practice
The practical move is to give waiting a name and a slot. If the options on the table are three interventions, doing nothing is not competing; if they are three interventions plus a stated watch-and-wait with a review date, it is. Fix the base rate before the meeting: estimating how often this class of problem resolves itself is honest in advance and nearly impossible once a specific case is in the room. Clinical medicine has pushed this furthest: Scott and colleagues (2017) set out debiasing strategies for minimising low-value care, the setting where the pull runs toward the extra test or procedure and the discipline is making not ordering it a documented, defensible choice rather than an absence. The organisational fix follows the same shape. Give watch-and-wait an owner, as any intervention has one: a named person who presents it, argues for it, and reports back at the review date on what the wait produced. Then make that report a deliverable that counts — written up, circulated, and weighed at appraisal the way a shipped change is — so restraint leaves the same trail as action instead of leaving none.
Examples
Facing a penalty kick, goalkeepers almost always dive left or right, though staying centred stops more shots; diving feels like doing the job, while standing still looks like failing to try.
A patient's ordinary back pain gets an MRI, then a referral, then surgery, because at each step both doctor and patient find escalating easier to justify than watching and waiting.
An investor rattled by a market dip sells and rebuys through the noise, paying fees and taxes, while the colleague who did nothing quietly ends the year ahead.
A product team ships a redesign to arrest a dipping metric that was merely seasonal. The change disrupts behaviour enough that nobody can tell whether the dip would have recovered anyway.
Mid-outage, an engineer restarts services and reverses config while the real fix is already propagating, adding changes the team must later untangle from the actual cause.
First described in Patt & Zeckhauser (2000); Bar-Eli et al. (2007).
Key references
- Fillon, A. A., Strauch, L., & Feldman, G. (2023). Evaluations of action and inaction decision-makers in risky decisions resulting in negative outcomes: Inaction agents are preferred to and perceived as more competent and normative than action agents. Collabra: Psychology, 9(1), 74817. doi.org/10.1525/collabra.74817
- Scott, I. A., Soon, J., Elshaug, A. G., & Lindner, R. (2017). Countering cognitive biases in minimising low value care. Medical Journal of Australia, 206(9), 407-411. doi.org/10.5694/mja16.00999
- Feldman, G., & Albarracin, D. (2017). Norm theory and the action-effect: The role of social norms in regret following action and inaction. Journal of Experimental Social Psychology, 69, 111-120. doi.org/10.1016/j.jesp.2016.07.009
- Bar-Eli, M., Azar, O. H., & Lurie, Y. (2009). (Ir)rationality in action: Do soccer players and goalkeepers fail to learn how to best perform during a penalty kick? Progress in Brain Research, 174, 97-108. doi.org/10.1016/S0079-6123(09)01309-0
- Bar-Eli, M., Azar, O. H., Ritov, I., Keidar-Levin, Y., & Schein, G. (2007). Action bias among elite soccer goalkeepers: The case of penalty kicks. Journal of Economic Psychology, 28(5), 606-621. doi.org/10.1016/j.joep.2006.12.001
- Patt, A., & Zeckhauser, R. (2000). Action bias and environmental decisions. Journal of Risk and Uncertainty, 21(1), 45-72. doi.org/10.1023/A:1026517309871