Does Groupthink Explain Bad Decisions

Why We Act · 24 October 2026 · Reviewed by Faisal Zaidi

Janis built groupthink on four fiascos and two successes. Later work found no overall penalty for cohesion, and a better-tested route to the same bad decisions runs through information the group never shares.

Answer to the question

Is groupthink the real reason teams fail?

Not by itself. Groups do converge on bad plans, but Janis's causal account, that cohesion produces the convergence, has not been demonstrated: a 1994 meta-analysis found no significant overall effect of cohesiveness on decision quality, and a 1998 review judged it too early to pass judgment. Unshared information and assigned dissent explain more.

Key takeaways

Janis's 1972 theory rests on six case studies reconstructed from the historical record, four fiascos and two successes, among them the Bay of Pigs and the Cuban Missile Crisis.
Esser's 1998 review of the research judged it too early to pass judgment on groupthink theory.
Mullen and colleagues' 1994 meta-analysis found no significant effect of cohesiveness on decision quality, though cohesive groups decided worse under directive leadership.
Stasser's 1985 experiment showed that discussion kept members' distorted pictures in place, and Nemeth and colleagues found in 2001 that authentic dissent outperformed all three devil's advocate conditions.

Early in the Kennedy presidency a team of advisers approved a landing in Cuba that collapsed within days. Eighteen months later, with the same president, a team handled the missile crisis far more carefully. Irving Janis built his theory of groupthink on that contrast and published it a decade after the landing. By the end of the century, reviewers were asking whether the evidence had ever matched the theory.

Groupthink is Janis's name for a group's drive for unanimity overriding its realistic appraisal of the alternatives. He took the coinage from Orwell, a provenance that Colin Fisher records in The Collective Edge, and it explains the word's flavour: it arrived as a warning before it arrived as a measurement. The label has since travelled well beyond the cabinet room. It now serves as the default explanation whenever capable people agreed to something that nobody can defend afterwards.

What Janis counted as evidence

The founding book rests on six case studies: four failed decisions, among them the Bay of Pigs landing, and two successes, the Marshall Plan and the Cuban Missile Crisis. The set works as a design, with fiascos on one side and well-handled decisions on the other, and it gave him something to compare.

What started him was an account of the Bay of Pigs by one of Kennedy's own advisers. Janis asked how bright, shrewd men around Kennedy could have been taken in by the CIA's patchwork plan, which is close to the question in this article's own title. His answer was a mechanism. A cohesive group, led by a directive chief, comes to value agreement above accuracy. The symptoms he listed include an illusion of invulnerability, pressure on dissenters, self-censorship and an illusion of unanimity, so that the resulting quiet reads as consent.

His method was the case study, set out in Victims of Groupthink and later revised and expanded as Groupthink. Janis reconstructed each decision from the historical record, looked for the symptoms in the fiascos and their absence in the successes, and explained the difference by the conditions around the group. The method gives rich description and no control group, because every case had already been sorted by its outcome before he began.

Two claims, one word

Janis's founding book, revised a decade later, packed two claims into one word. The first claim is descriptive. Groups sometimes converge quickly on a plan, treat doubters as a nuisance, keep their reservations private and take the quiet that follows for agreement. Janis gave that pattern a vocabulary, and the vocabulary is useful, because a manager can recognise the pattern without any theory of where it came from.

The second claim, the one that makes groupthink a theory and not only a vocabulary, is causal. It says cohesion, together with conditions such as directive leadership, produces the pattern, and the pattern produces the fiasco. This is the claim that needs experiments. A story that begins with a disaster and works backwards will find causes whether or not they were operating, and Janis himself described one of his explanations, that members protect their self-esteem by going along, as a large inferential leap that later work would have to check.

A story that begins with a disaster and works backwards will find causes whether or not they were operating.

The research programme was taken stock of in Organizational Behavior and Human Decision Processes, where James Esser's 1998 review carried the title "Alive and well after 25 years". The review's abstract is not openly available, so what follows rests on secondary accounts of it. They report a conclusion more guarded than the title: that it was too early to pass judgment on groupthink theory, with the studies raising about as many questions and methodological issues as they answered. That is an undecided verdict, not a refutation.

Cohesion did not predict worse decisions

The cleanest test of the causal claim asks whether cohesive groups make worse decisions. A meta-analysis led by Brian Mullen, published in Small Group Research in 1994, pooled the tests of the groupthink hypothesis to answer exactly that. It found no significant effect of cohesiveness on the quality of group decisions.

The detail sits in the exceptions, since the analysis combined existing tests and ran no new experiment. Cohesive groups did decide worse when another of Janis's antecedents, directive leadership, was present. The harm was larger when cohesiveness meant interpersonal attraction than when it meant commitment to the task or pride in the group, and it grew with the size of the group. The abstract does not say why larger groups fared worse, so that pattern is a finding without a mechanism, and this article does not supply one.

Taken together, those results keep part of Janis alive. A close-knit group under a leader who signals the answer he wants can decide badly. Cohesion on its own, the ingredient the theory leaned on hardest, did not predict worse decisions.

The founding case, reread

The founding case was reread in the same issue of the journal as Esser's review. Roderick Kramer asked how well the groupthink hypothesis had stood the test of time, with the Vietnam decisions as a second case, and, drawing on declassified documents, oral histories and participants' memoirs, concluded that cohesion pressures played a smaller causal role than Janis argued. Presidents Kennedy and Johnson, in his reading, mainly weighed their options by their political consequences, above all the wish to avoid political losses and damage to their reputations.

Kramer's summary was that the groupthink hypothesis overstates the influence of small group dynamics and understates the role of political considerations. Both decisions may still have been seriously flawed, and on his account the logic of the failure belongs to political psychology and not to social psychology. The two accounts agree that the decisions failed and disagree about why, and the reason matters for the remedy.

On that reading, an advisory group that watches its president weigh the political cost of cancelling an operation does not need cohesion to explain why it goes along. The remedy follows the mechanism, so the choice between the two accounts changes what a team should do about it.

A wider view came from Robert Baron, whose 2005 chapter "So Right It's Wrong" carries the subtitle "Groupthink and the Ubiquitous Nature of Polarized Group Decision Making". His ubiquity model holds that the processes Janis described are common features of group decision making, found in most groups, and not the signature of a special pathological state that strikes a few.

On that account the vocabulary survives and the diagnosis goes, because a pattern found in every outcome cannot by itself explain one of them. Groups tend to polarise toward the position most members started with, and discouraging doubt is a routine part of that tendency, which operates in decisions that turn out well as well as in those that turn out badly.

A pattern found in every outcome cannot by itself explain one of them.

Shared information wins the discussion

If agreement does not need cohesion to explain it, something else must produce it, and the best-tested candidate is information. In 1985 Garold Stasser gave four-person groups of students a political caucus to run. Each member held only part of what was known about the candidates, arranged so that pooling everything would reveal the strongest one.

The groups chose the candidate that a plurality of members had favoured before the discussion began. Recall taken before and after talking showed that the discussion had kept the distorted pictures in place rather than correcting them. The paper's title names the mechanism, biased information sampling during discussion: what everyone already knew was what got discussed.

The field calls this arrangement a hidden profile: the full picture exists in the group and in no single head. Nobody in such a group has to be pressured or silenced. Each member can speak freely and the group still settles on the wrong answer, because the agreement is built from what the members had in common. The shopping version of this problem is set out in [Group Decisions at the Checkout](https://thewhyweproject.com/articles/group-decisions-at-the-checkout), and the wider record on whether groups beat individuals is in [Is Collective Intelligence Real](https://thewhyweproject.com/articles/is-collective-intelligence-real). Members who stay quiet may also be assuming that [everyone else already wants the plan](https://thewhyweproject.com/articles/everyone-thinks-everyone-else-wants-it), which turns silence into a second source of apparent consent.

Can assigned dissent help?

The usual repair is to appoint someone to argue the other side, though an earlier study had already suggested that devil's advocates tend to reinforce the initial viewpoint rather than prompt divergent thinking. Charlan Nemeth and two colleagues compared that with the real thing in the European Journal of Social Psychology in 2001. Their devil's advocate conditions varied whether the advocate's true position was known or unknown, and whether it matched the position assigned. Authentic dissent, from a minority that held its position sincerely, outperformed all three devil's advocate conditions, and the authors read the result as showing that role-playing struggles to reproduce genuine dissent.

Fisher reaches a matching preference in the chapter of The Collective Edge on the remedy for conformity. He doubts devil's advocacy on the ground that it is still advocacy, and favours arrangements that do not depend on anyone playing a part: outside perspectives, psychological safety, boundary-spanning and brokerage, newcomers, systematic aggregation of independent judgements, and protected dissent. Each changes who is in the room, what they know or when they speak.

Those remedies speak to both halves of the problem. A decision nobody argued against has been agreed but not tested, and agreement costs a group almost nothing. A newcomer brings information the incumbents do not share, which addresses the hidden profile, and a protected dissenter supplies dissent that does not depend on a volunteer. They work whether or not Janis's mechanism is right, since they target what the research does support, which is that members' information goes unpooled and that assigned objections change little.

Common questions

Is groupthink the real reason teams fail?

Did groupthink cause the Bay of Pigs?

Janis said so in 1972. Roderick Kramer's 1998 reappraisal found cohesion pressures played a smaller causal role than Janis argued and that the presidents mainly weighed political consequences. The decision was a fiasco on either account, and the accounts differ on what a team should change.

Why do smart teams still agree on bad decisions?

One well-tested route is information. Stasser's 1985 experiment showed that groups tend to discuss what members already share, so a decision can turn on the common picture while the unshared facts that would change it stay unspoken. Pressure from a leader can add to this, as the 1994 meta-analysis found.

How can a team avoid agreeing too fast?

The evidence favours changing the structure over asking someone to play critic. Nemeth and colleagues found in 2001 that authentic dissent beat assigned devil's advocacy, and Fisher recommends newcomers, outside perspectives, protected dissent and aggregating independent judgements.

Sources

Fisher, C. M. (2025). The Collective Edge. New York: Avery, an imprint of Penguin Random House

Janis, I. L. (1972). Victims of Groupthink: A Psychological Study of Foreign-Policy Decisions and Fiascoes. Boston: Houghton Mifflin

Janis, I. L. (1982). Groupthink: Psychological Studies of Policy Decisions and Fiascoes (2nd ed.). Boston: Houghton Mifflin

Esser, J. K. (1998). 'Alive and well after 25 years: A review of groupthink research.' Organizational Behavior and Human Decision Processes, 73(2-3), 116-141. doi:10.1006/obhd.1998.2758 (abstract not openly available; conclusion as reported in secondary accounts)

Kramer, R. M. (1998). 'Revisiting the Bay of Pigs and Vietnam decisions 25 years later: How well has the groupthink hypothesis stood the test of time?' Organizational Behavior and Human Decision Processes, 73(2-3), 236-271. doi:10.1006/obhd.1998.2762

Mullen, B., Anthony, T., Salas, E., & Driskell, J. E. (1994). 'Group cohesiveness and quality of decision making: An integration of tests of the groupthink hypothesis.' Small Group Research, 25(2), 189-204. doi:10.1177/1046496494252003

Baron, R. S. (2005). “So right it's wrong: Groupthink and the ubiquitous nature of polarized group decision making.” Advances in Experimental Social Psychology, 37, 219-253. doi:10.1016/S0065-2601(05)37004-3

Stasser, G., & Titus, W. (1985). 'Pooling of Unshared Information in Group Decision Making: Biased Information Sampling During Discussion.' Journal of Personality and Social Psychology, 48(6), 1467-1478. doi:10.1037/0022-3514.48.6.1467

Nemeth, C., Brown, K., & Rogers, J. (2001). “Devil's advocate versus authentic dissent: stimulating quantity and quality.” European Journal of Social Psychology, 31(6), 707-720. doi:10.1002/ejsp.58