Defensive decision-making

Florian Artinger, Sabrina Artinger and Gerd Gigerenzer surveyed 950 managers across all six hierarchy levels of a large German public administration. The question they asked was unusually blunt. They defined a defensive decision on the page, as one where the decision maker does not choose the option that is best for the organisation but instead chooses the option that protects them from potentially negative future consequences, and then asked: how many of the ten most important decisions you made in the last twelve months had a defensive component?

The average answer was 2.5 out of 10. Eighty percent of managers reported at least one. Seventeen percent said that at least half of their most important decisions had been defensive. It was widespread at every level of the hierarchy, not concentrated at the bottom or the top. For comparison, in a 2005 survey of American physicians in high-risk specialties, 93 percent reported practising defensive medicine, ordering scans, tests, antibiotics and surgery they judged clinically unnecessary.

The paper opens with the version everybody in business already knows. One of the services consultancies quietly provide is backing up a decision that has already been made, so that if it fails there is somebody else to point at. The authors name it in the title: C.Y.A.

The examples participants gave are the most useful part of the study. Thirty-five percent were about avoiding conflict. One manager described keeping a senior person in post until the end of their term while being confident it would damage the team, and reported that services had stalled for sixteen months and people were leaving. Nineteen percent were about not resisting pressure from a superior, including a manager who believed an external candidate was clearly better and appointed the internal one anyway.

Then the finding that makes this worth knowing rather than just recognising. Sixty-four percent of the examples people gave were psychological: conflict avoidance, pressure from above, low motivation. Only about twenty percent were about a lack of resources or information. But when the same people were asked what would fix it, forty-nine percent said more resources or more information. They described a social problem and prescribed a budget.

Something similar happened with blame. Fourteen percent of the stated causes pointed at political interference, but the regression found no significant effect from it. The authors' reading is that politics was serving as a scapegoat, which is itself a defensive move.

What did predict fewer defensive decisions was a team's approach to failure, meaning whether the first response to an error is to find who is responsible or to find out what happened, and employee voice, meaning whether people can raise concerns upward. Job satisfaction helped too. Older managers made more defensive decisions; managers with more years of experience in their specific area made fewer.

The honest limits, most of which the authors state themselves. This is self-report, and both the outcome and the explanations come from the same questionnaire. It is one organisation, a German public administration, and the sample was gathered by a cascade of invitations down the hierarchy, so managers who feared what their subordinates would report could quietly not forward it. The authors argue this makes the numbers a conservative estimate rather than an inflated one. And the 2019 study is correlational, which they acknowledge; a 2025 follow-up with 315 managers ran the scenario experimentally and found that low psychological safety combined with low authentic leadership causes defensive decisions to rise.

The reason this behaviour is stable is that it is correct. In an organisation where a failure triggers a search for who is responsible, choosing the defensible option is an accurate reading of the environment, and telling people to be braver is asking them to misread it on purpose.

You can predict the outcome of a meeting with one question. If this goes wrong, who gets blamed? If the answer is the person deciding, and the better option would expose them personally, the decision is already made.

The recognisable shapes are all around this. The larger vendor at several times the price, because if the large one fails it was a reasonable choice and if the small one fails it was your judgement. The approval chain that exists so no single name sits alone on a document. The consultancy hired to confirm a decision already taken. None of these appear anywhere as a cost. They are invisible waste, which is precisely why they survive, since visible failure gets punished and invisible waste does not.

The paper's own prescription is the sharpest thing in it, and it is one line: incentivise process quality rather than outcome quality, because the outcome is not something the decision maker controls. That is outcome bias turned into a management instruction. An organisation that cannot say out loud that a good decision lost will keep manufacturing defensive ones, no matter what it says about courage in its values.

And there is a diagnostic warning in the mismatch between what people described and what they prescribed. If you ask your team why decisions are slow or cautious, expect to be told the problem is resources, headcount or information. The examples in this study say otherwise: most of it was people avoiding a conflict or not resisting a superior. Budget will not touch that. What touches it is what happens in the room the next time something goes wrong.

Source: Artinger, Artinger and Gigerenzer, C.Y.A.: frequency and causes of defensive decisions in public administration, Business Research, volume 12, 2019, pages 9 to 25, open access. The physicians' figure: Studdert and colleagues, Defensive Medicine Among High-Risk Specialist Physicians in a Volatile Malpractice Environment, JAMA, volume 293, 2005. The experimental follow-up: Artinger, Marx-Fleck, Junker, Gigerenzer, Artinger and van Dick, Coping with uncertainty: the interaction of psychological safety and authentic leadership in their effects on defensive decision making, Journal of Business Research, volume 190, 2025.

The Fearless Organization

Amy C. Edmondson, 2018

Psychological safety is Edmondson's own construct, and the study behind this entry built its scale partly out of her survey items, so this is the mechanism at book length rather than an adjacent read. Her argument is that a team's willingness to speak up and admit error is not a matter of niceness but of performance, and that leaders build it deliberately or lose it by accident.