Groupthink

What groupthink is

Groupthink is a failure of a group's procedure, not of its members' intelligence. The individuals are usually capable, informed and acting in good faith. What fails is the process that was supposed to expose the weaknesses in the plan, because raising an objection has come to feel like an attack on the group rather than a contribution to it.

The psychologist Irving Janis named the concept, in a 1971 magazine article and then at length in a 1972 book, and he built it deliberately as an unattractive word in the tradition of Orwell's coinages. Janis identified eight symptoms in three groups, and the number is worth holding on to, because groupthink in ordinary use has drifted to mean any agreement anyone dislikes. In Janis's account it is a specific syndrome with specific signs.

The one sentence version, and Janis's eight symptoms

In one sentence, groupthink is the point at which a group's desire for unanimity overrides its members' motivation to appraise alternatives realistically. Janis sorted the eight symptoms into three categories, and a group showing several of them at once is the pattern he was describing.

• Overestimating the group: an illusion of invulnerability that encourages excessive optimism, and an unquestioned belief in the group's own good intentions.
• Closed-mindedness: collective rationalizing away of warnings, and stereotyped views of outsiders and rivals as too weak or too foolish to matter.
• Pressure toward uniformity: self-censorship of private doubts, an illusion of unanimity produced by that silence, direct pressure on anyone who does speak up, and self-appointed mindguards who shield the group from inconvenient information.

The illusion of unanimity is the load bearing item. Silence is read as consent, so five people each holding the same private doubt each conclude they are the only one, and the meeting minutes record a unanimous decision that nobody in the room fully believed.

The evidence behind groupthink, and its honest limits

Janis built the concept from historical case studies rather than from a laboratory, and that is the first thing an honest account has to say. He examined government and military decisions that went badly wrong, looked for common features in how the meetings were run, and then examined cases where similar groups, having changed their procedures, made better decisions afterward. The comparison between the failures and the recoveries is what gave the model its shape.

Case study evidence has a known weakness: the researcher already knows which decisions turned out badly, and is looking backward for causes. Experimental tests of the full Janis model have produced mixed results, and some of the eight symptoms have held up better than others. The current honest position is that groupthink is a useful checklist of conditions rather than a demonstrated causal law, and its practical value lies in the remedies it suggests, which can be tested directly and which cost almost nothing to apply.

Two experimental findings from the surrounding literature are firmer. Solomon Asch's conformity studies in the 1950s showed that a person will give an answer they can see is wrong when everyone before them has given it, on a task as simple as judging which of three lines matches a standard. Research on group polarization, beginning with work on the so-called risky shift in the early 1960s, showed that discussion among people who already lean the same way tends to move the group further in that direction rather than toward the middle. Neither is groupthink, and both make it easier.

Where groupthink shows up outside the boardroom

Groupthink appears wherever a cohesive group with a directive leader makes a consequential decision under time pressure and without outside input, such as product launch committees, engineering review boards, hiring panels, school leadership teams, nonprofit boards and volunteer organizing groups.

Two worked examples make the mechanism visible.

The launch meeting

A team of eight has spent four months on a product and meets to decide whether to ship on Friday. The manager opens by saying the date matters a great deal to the company. A junior engineer knows of a defect that appears roughly once in every 200 sessions, considers raising it, notices that nobody else has raised anything, and decides it can be patched after launch. The decision is unanimous. The information needed to make the right call was in the room and never reached the table.

The timetable committee

A school committee of six agrees to restructure the school day. One member thinks the transport arrangements will not work but is the newest person present and has heard the head teacher describe the plan as settled. She says nothing. Afterward, three of the six separately tell her they had the same concern. That conversation, held after the vote, is the clearest possible evidence of an illusion of unanimity.

What groupthink is confused with: a room that agrees because one option is better

Groupthink is confused, above all, with a room that agrees because one option really is better. Groupthink is not the same as agreement, and a unanimous decision is not evidence of it. Groups agree for excellent reasons all the time, including the reason that one option is clearly better. The diagnostic question is not whether the group agreed but whether disagreement had a route to the table. Nothing here is a broken inference of the kind a deductive reasoning check would find; the premises the group used were simply never assembled.

Groupthink also gets confused with the cognitive bias family generally, and it does belong there, but it is the one that lives between people rather than inside a single head. Confirmation bias and the Dunning Kruger effect describe what one person does with evidence. Groupthink describes what a room does with a person.

One question tells a healthy consensus from groupthink, and it is not about the vote: what happened to the last person who disagreed here? If the answer is that they were heard, answered on the merits and are still in the room, the agreement is probably real. If the answer is that nobody can remember anyone disagreeing, the group has no evidence that its unanimity means anything, because it has never tested it.

What actually reduces groupthink

What reduces it is structure that makes dissent cheap, and Janis proposed most of these himself. Six moves, all of which can be applied to a meeting on Monday:

1. Collect independent written estimates from each member before any discussion begins.
2. Withhold the leader's preference until everyone else has spoken.
3. Assign one named person the job of arguing against the plan, and rotate it.
4. Split the group into two subgroups that work the problem separately, then compare.
5. Bring in an outsider with no stake in the outcome and no history with the group.
6. Hold a second meeting, after the decision, in which the only agenda item is what could still go wrong.

The first is the most powerful and the most neglected. Once one confident estimate has been said out loud, every later estimate is anchored to it, and the group has lost the independent judgments it needed. Writing first costs two minutes.

Be honest about the limit: a permanent devil's advocate stops working, because the group learns to discount the objection as a role rather than hear it as a warning. Rotation matters more than the assignment.

The test to run in your next meeting

Ask these three questions, and prefer the first:

• If I disagreed with this right now, what would it cost me, and who would notice?
• Did anyone in this room state a serious objection, and what happened to the person who did?
• What would have to be true for this plan to fail, and has anyone here been asked to describe it?

Is groupthink always bad? No. Cohesion is what allows a group to act quickly, and a group that debates every decision from first principles will not ship anything. Groupthink is the cost of cohesion when no one has built a route for dissent, and the cure is not less cohesion but a better route.

Groupthink in psychology: famous, widely taught, and weaker than its fame

In psychology, groupthink belongs to the social psychology of group decision making, and the honest summary is that it is better known than it is supported. Janis's model is taught in introductory courses, quoted in management training and used as a verb in ordinary speech, and the empirical record behind it is thinner than any of that implies.

The model makes a specific claim with three links: certain antecedents, such as a cohesive group, a directive leader and insulation from outside views, produce the eight symptoms, and the symptoms produce defective decisions. That chain is what has not held up cleanly. Laboratory studies manipulating cohesion and leadership have produced mixed and partial results, and the symptoms have not moved together as a syndrome should. What has held up is the structural end: a leader who states a preference at the start measurably shapes what a group then produces, and a group with no route for dissent does not hear the objections its members privately hold. Those are the parts worth acting on, and they are also the parts that survive without the theory.

There is a method problem worth naming, because this site is about how evidence is built. The case studies were reconstructed after the outcomes were known. A researcher who already knows which decisions ended badly, reading the record for features that explain the ending, is working under the conditions that produce hindsight bias, and the features will duly appear. That is an observation about the evidence for a theory of bad decisions, not a verdict on the theory, and it does not require naming a single decision or a single person to be a fair thing to say.

What the popular version overstates is cohesion as the villain and the label as a diagnosis. Cohesive groups make excellent decisions routinely, "groupthink" has become a name for any consensus the speaker dislikes, and applying it after a bad outcome explains nothing, since the same room running the same procedure would have been called decisive had the plan worked. Use it as Janis's list of conditions to check before the decision, not as a description to reach for afterward.

Groupthink and control group are unrelated terms

A note on a common collision, because the two phrases sit next to each other in course material and in search results. A control group is the comparison arm of an experiment: the participants who do not receive the treatment being tested, so that the outcome in the treated arm has something to be measured against. It has nothing to do with groupthink beyond the word "group". Groupthink concerns how a group makes a decision; a control group is a design feature that lets a researcher tell an effect from a coincidence, and this site treats it separately alongside blinding.

Where to go next

BIASESCognitive biasWhat it is, with examplesBIASESThe Dunning Kruger effectThe Dunning Kruger effect, and what the study actually showedBIASESConfirmation biasDefinition and examplesREASONINGDeductive reasoningDefinition and examples