Journal · Evidence
Groupthink, and the version that actually applies to you
The term is used to explain almost any bad group decision, which is a sign it explains very little. The specific mechanisms it points at are worth keeping.
Is groupthink a real phenomenon? The original theory, drawn from case studies of political fiascos, has been criticised for weak empirical support and after-the-fact reasoning. The narrower mechanisms it describes, such as self-censorship and premature convergence, are better supported and easy to observe.
Janis introduced the idea in the 1970s from retrospective case studies of policy disasters. The trouble with retrospective case studies of disasters is that you already know the outcome, and it is easy to find the causes you expect. Later experimental work supports some components and not the package as a whole.
The parts worth keeping
- Self-censorship. People with doubts do not raise them, particularly when the doubt concerns someone senior's preferred plan.
- Premature convergence. The group settles early, then spends its remaining time gathering support for what it settled on.
- Illusion of unanimity. Silence gets read as agreement, and the silent people conclude they are the only ones with reservations.
In a small company the mechanism is rarely a domineering leader. It is an enthusiastic one, which is harder to notice.
Why founder-led teams are exposed
Small teams are selected for belief, hired partly for cultural fit, and led by someone whose conviction is a large part of why the company exists. Every one of those is a strength, and together they produce a room where dissent is socially expensive and nobody has to be silenced for silence to happen.
What actually helps
Write positions before discussing them, so the first speaker does not set the anchor. Have the most senior person speak last, consistently. Assign the objection rather than waiting for it. And separate the meeting that generates options from the meeting that picks one, because doing both at once reliably means the first option survives.
Five Peers editorial note. Sources named in the text.
Structured objection, assigned rather than hoped for.
Compose my board