Groupthink: What It Is, Famous Examples, and How to Prevent It

Groupthink is what happens when a group’s wish to agree overrides its judgment: doubts go unspoken, the favored plan goes untested, and everyone leaves believing the decision was unanimous. The psychologist Irving Janis named it in the early 1970s after studying policy disasters such as the 1961 Bay of Pigs invasion, and he listed eight warning signs and nine safeguards.¹

Decades of testing have kept the warning signs and rejected much of the recipe. In a 2005 review of three decades of research, the social psychologist Robert Baron concluded that Janis was right about the symptoms of groupthink but wrong about the conditions he said produce them, because those symptoms appear in ordinary, everyday groups, even temporary ones.²

That makes the useful part of the story practical. The counterweights this guide ends on are cheap habits, not programs: collect each person’s view before the discussion starts, let the most senior person speak last, and make naming the plan’s weak points part of everyone’s job. How one person weighs a choice is covered in how we make decisions; this page is about what changes when people decide together.

What groupthink means, in Janis’s own terms

Groupthink, in Irving Janis’s 1972 definition, is a way of thinking people fall into when they are deeply involved in a cohesive group and their striving for unanimity overrides their motivation to weigh the alternatives realistically. He chose the word to echo the newspeak of George Orwell’s 1984, and meant the unflattering ring.¹

Janis described a chain. A close-knit group, insulated from outsiders, led by someone who signals a preferred answer and working under stress, develops a strong drive for agreement. That drive shows up as symptoms, including an illusion of invulnerability, collective rationalizing of warnings, stereotyped views of rivals, pressure on dissenters, self-censorship, an illusion of unanimity and self-appointed “mindguards” who shield the group from unwelcome facts. The result is a defective process: too few options considered, a poor search for information and risks left unexamined.²

The mechanism is social, not intellectual. Disagreeing with a group you belong to carries a cost, so each member quietly shrinks their own doubts. Janis added a second step: the group then reads that silence as consent, which produces a shared illusion of unanimity.¹

Picture a launch review in which four senior people sound confident and a junior analyst has seen worrying numbers from a pilot. She waits for someone else to raise them, nobody does, and the chair closes with “no objections, then”. The group now believes it agreed. The lesson for your own meetings: a quiet room is missing data, not a vote. Listen for everyday versions of Janis’s symptoms, such as “we’re all agreed, right?” (assumed unanimity), “let’s not reopen that” (pressure on a doubter) or a colleague who keeps bad news away from the boss (a mindguard).

Famous examples of groupthink, and what they have in common

The Bay of Pigs invasion is Janis’s signature case of groupthink; the 1986 Challenger launch and the US intelligence estimate on Iraq’s weapons before the 2003 war are two much-discussed later ones. In each, relevant doubts existed inside the organization but did not shape the decision.

In April 1961 a brigade of about 1,400 Cuban exiles, backed by the United States, landed at the Bay of Pigs, and nothing went as planned. Janis asked how President Kennedy and his able advisers approved the plan. His answer drew on memoirs: advisers who had doubts kept them to themselves, partly for fear of seeming soft, and Robert Kennedy told one doubter, the historian Arthur Schlesinger, to stop pressing his objections, the role Janis called a mindguard. Janis contrasted it with the 1962 Cuban missile crisis, when the same president’s group brought in outside experts and split into subgroups that debated each other.¹

After the Challenger shuttle accident on January 28, 1986, the presidential commission that investigated found the decision to launch “flawed”: engineers at the contractor Thiokol had opposed launching in the cold, their managers reversed that position, and the officials who approved the launch did not know about the engineers’ continuing opposition. Note the difference from the classic picture: the engineers did speak up, but their objections never reached the officials who made the final decision.³ A 1989 analysis coded the commission’s report for signs of groupthink and found signs for it about twice as often as signs against, rising in the final 24 hours before launch (we read only the abstract).⁴

In 2004 the US Senate Select Committee on Intelligence used the word itself. It concluded that intelligence agencies had shared a collective presumption that Iraq had active and growing weapons programs, and that this “group think” dynamic led analysts to read ambiguous evidence as conclusive and to ignore or minimize evidence that pointed the other way.⁵

This presumption was so strong that formalized IC mechanisms established to challenge assumptions and group think were not utilized.

That case involved several agencies, not one small room, which shows how far the idea has travelled beyond Janis’s cohesive inner circles. The shared lesson of all three: when you review a failure in your own team, ask who knew, and why their knowledge did not count. Before a big decision, ask the same question the other way round: who might know something this room has not yet heard?

Further reading

• 

  The Wisdom of Crowds

  by James Surowiecki

  Why groups decide well when members judge independently first and badly when they agree too early, with the Columbia shuttle as a case.

As an Amazon Associate WiserHours earns from qualifying purchases.

How well groupthink holds up as a theory

Groupthink has far weaker research support than its fame suggests. James Esser’s 1998 review of the first 25 years of groupthink research found only a small, if growing, body of empirical work, though Esser still judged the theory useful for prompting new research (we read only the abstract).⁶ Most of the reviews Baron surveyed recommended revising, replacing or rejecting Janis’s model, because its central prediction, that cohesive groups under stress are most at risk, has found little support.²

The tests are few, and they rarely go Janis’s way. By Baron’s count, Esser’s review covered eleven laboratory studies and some 17 historical reports, and a meta-analysis of the key prediction, that cohesion lowers decision quality, found only nine studies to pool. Overall, Baron judged the results on cohesion inconsistent and often contrary to prediction.² Marlene Turner and Anthony Pratkanis, introducing the special journal issue that marked the theory’s 25th year, found that no research had supported Janis’s full chain from causes to symptoms to defective decisions, and noted the irony of a theory about the dangers of conformity being accepted with so little scrutiny.⁷

Even the founding case has shifted. Roderick Kramer revisited the Bay of Pigs and Vietnam decisions using declassified documents, oral histories and memoirs that became available after Janis wrote, and argued that Kennedy and Johnson weighed their options mainly by their political consequences. Both decisions may have been seriously flawed, he concluded, but political calculation explains them better than a small group’s striving to stay cohesive (we read only the abstract).⁸

Why, then, does the idea feel so true? Baron’s answer is that the symptoms are familiar because they are common. His “ubiquity model” proposes that three conditions are enough: members identify with the group, the group has salient norms, and members lack confidence that they can solve the problem themselves.²

1. Janis's conditions: a close-knit, insulated group with a directive leader, under stress. Tests found little consistent support for this link, especially for cohesion
2. Ordinary groups: shared identity, norms that reward agreement and low confidence in solving the problem may be enough (Baron, 2005)
3. The drive for agreement: members want unanimity more than a realistic look at the options
4. Symptoms and the decision: unspoken doubts, a false sense of unanimity and a narrow search, which historical studies have linked to poorer choices

The practical lesson: don’t judge your risk by how close your team feels. In Baron’s account the pressure comes from wanting to belong and doubting your own answer, not from friendship, so a loosely assembled project group, meeting for the first time on a hard problem, can show every symptom.

Myth

Groupthink mainly strikes tight-knit, friendly teams under pressure.

Fact

Tests found no consistent link between cohesion and poor decisions; the warning signs also appear in ordinary and temporary groups.

What goes wrong in ordinary groups: they discuss what everyone already knows

The best-measured groupthink-like effect is a narrowed discussion: groups talk far more about information all members already share than about what only one member knows. A 2012 meta-analysismeta-analysis: A study that combines the results of earlier studies on the same question into one overall estimate. Pooling makes the estimate more precise, but it cannot repair the studies it pools: a meta-analysis of surveys is still survey evidence.Full entry in the glossary of 65 studies with more than 3,000 groups found that groups mentioned shared information much more often than unique information, and that groups whose facts were split among members were much less likely to find the best answer than groups given the full picture.⁹

Researchers call these tasks hidden profiles, because the best option becomes visible only if members pool what each alone knows. The reasons for the bias are partly arithmetic, since a fact everyone knows has more people able to raise it, and partly social: people rate colleagues who repeat shared views as more competent and credible, Baron notes.²

Disagreement before the discussion starts turns out to help. In a 2006 experiment with 135 three-person groups choosing a job candidate, groups whose members all preferred the same wrong candidate beforehand hardly ever found the right one; groups whose members disagreed did better, even when none of them favored the right candidate at the start (we read only the abstract).¹⁰

Think of a hiring panel in which each interviewer noticed a different warning sign, while the meeting is spent on the CV everyone read. Asking each person for their verdict and their one unique concern before anyone speaks puts those warnings on the table.

How to prevent groupthink: what the evidence supports

The prevention steps with experimental support change what a group rewards: a norm that values critical debate over quick agreement, and genuine differences of opinion surfaced before discussion. A leader who holds back a preference is sensible advice with more mixed evidence.

The clearest test of the norm idea is a pair of small experiments.

The study

Limited evidence

Critical versus consensus norms (Postmes, Spears & Cihangir, 2001)

Before a hiring decision, student groups did a short warm-up task meant to build either a consensus norm (making a poster together) or a critical norm (debating a policy proposal). In the first study, 67% of decisions in critical-norm groups picked the best candidate after discussion, against 22% in consensus-norm groups. Consensus groups rated information everyone had known from the start more highly, and the two kinds of group were equally cohesive.¹¹

The striking part is how little it took: a ten-minute task, unrelated to the decision, changed how critically the groups then discussed the candidates. The caveats are the size of the studies, the student participants and the short lab setting; nobody has shown how long such a norm lasts in a real team.

1. Make critical evaluation everyone’s job

Janis’s first safeguard was for the leader to give every member the role of critical evaluator and to accept criticism of their own judgments, encouraging everyone to air objections and doubts openly.¹ In practice, that can be one sentence at the start: “Before we agree, each of us names the strongest reason this could fail.” This is also what the team research on psychological safety asks of leaders.

2. Have the leader speak last

Janis counted the lack of impartial leadership among the conditions for groupthink. The evidence is mixed: Baron counts directive leadership among the few conditions linked to poor decisions in both laboratory and historical studies, and argues it would harm decisions with or without groupthink.² Turner and Pratkanis, by contrast, found that laboratory studies of leadership style showed no harm to performance.⁷ The social cost from the first section is still a reason for caution: once the most senior person has named a favorite, disagreeing means disagreeing with them. If you chair the meeting, keep your view until everyone else has given theirs.

3. Collect views before the meeting

Ask for short, independent written views first, including each person’s main concern. The aim is to keep the disagreement that exists before a discussion from being smoothed away, and the job-candidate experiment above suggests that disagreement is what helps. The simplest version is five minutes of silent writing at the start of the meeting, with each view read out in turn before open discussion begins.

4. Bring in outsiders and hold a second-chance meeting

Janis also recommended inviting outside experts to challenge the core group, splitting into subgroups, and holding a “second chance” meeting after a tentative decision, at which every member voices remaining doubts. He rated an assigned devil’s advocate as of limited value, warning that many executives go through the motions with no apparent effect.¹ These safeguards rest mainly on case studies such as the missile crisis, so treat them as sensible habits rather than proven fixes.

Structured advocacy has thin support too: in a 2006 experiment, groups following an advocacy procedure discussed more information than groups in free discussion, but their decisions did not significantly improve as they gained experience (we read only the abstract).¹² A rival reading comes from Jerry Harvey’s Abilene paradox, in which a group does something that contradicts what its members really want: Harvey argued that many events Janis described as conformity pressure are better seen as mismanaged agreement.¹³

The bottom line

Groupthink is real as a description and weak as a theory. The famous recipe, a close-knit group under stress, has little support; the symptoms, unspoken doubts and a discussion that circles what everyone already knows, show up in ordinary meetings. Build the counterweights into your process: collect views first, speak last if you lead, and make naming the flaws part of the job.