Custodes Futurifor the advancement of humanity

Conformity and groupthink

Conformity and groupthink is a barrier to human progress within the social factors. The evidence for it, and the methods proposed to address it, follow.

Asch (1956) showed that many people agree with a group's clearly wrong answer. Janis (1972) described how close-knit groups suppress dissent. Janis's model rests mainly on historical case studies and has limited experimental support (rated C).

Evidence

  • Asch experiments. When people judged line lengths alone, they erred under 1% of the time (0.7%). With a unanimous group giving wrong answers, 36.8% of responses conformed. 94 of 123 participants (about 76%) conformed at least once, and 29 of 123 (about 24%) never did (Asch, 1956).
  • One dissenter. When one other person gave the correct answer, conformity dropped to about 5% of trials (Asch, 1956).
  • Checks. The percentages above match Table 3 of the original monograph (123 experimental and 37 control participants). The 36.8% conformity rate among 123 participants matches the figure reported by Franzen and Mader (2023). Their modern replication found 33% conformity, close to the original, though rates vary across cultures and eras.

What can be done

Actions rated on the strength-rating scale (A strong and replicated, B solid but limited, C weak or debated, D contested or failed), applied to the specific claim made.

  • Have people come up with ideas alone before pooling them (nominal groups) (B). A meta-analysis of brainstorming experiments compared face-to-face brainstorming groups with nominal groups (the same number of people working alone, with their ideas combined afterward). The meta-analysis found that face-to-face groups were significantly less productive in both quantity and quality of ideas. The loss was larger in larger groups (Mullen et al., 1991). The studies were laboratory tasks, so the size of the benefit in real organizations is less certain. This complements the independent-estimates methods in the proposed methods below.
  • Build groups with members from different backgrounds (B). A study of mock juries deciding the trial of a Black defendant found that racially diverse juries exchanged a wider range of information than all-White juries. In the same study, White jurors in diverse juries cited more case facts and made fewer errors (Sommers, 2006). A study of experimental markets in Southeast Asia and North America found that people randomly assigned to ethnically diverse markets priced assets 58% closer to their true values than people in homogeneous markets. In the homogeneous markets, traders more often copied others' speculative prices (Levine et al., 2014). Both are laboratory studies with constructed tasks.
  • Not effective: relying on open discussion to bring out information that only some members hold (B). A meta-analysis of 65 hidden-profile studies (3,189 groups) found that groups mentioned two standard deviations more of the information everyone already shared than of information held by single members. The same meta-analysis found that groups whose key facts were spread among members were eight times less likely to find the best solution than groups given full information (Lu et al., 2012). Communication medium (face to face or computer) made no difference.

Proposed and experimental methods

Methods that are proposed, under trial, approved in some places, or tried and then failed. Each shows a stage label and an evidence rating. A stage label shows how far a method has progressed, not whether it works. The stage labels are explained on the social factor page.

  • Keeping judgments independent or networks decentralized (Early trial, C). In a lab experiment with 144 people, seeing other people's estimates narrowed the spread of answers but did not make the group more accurate. It also raised confidence (Lorenz et al., 2011). A later randomized experiment found that sharing estimates made group estimates reliably more accurate in decentralized networks, where no single person dominates. The same experiment found that in centralized networks, group error became more likely to rise (Becker et al., 2017). In practice, this means collecting independent judgments before discussion and limiting the pull of central figures. No field test in real organizations was found in this review.
  • Genuine dissent rather than assigned devil's advocates (Early trial, C). This builds on "Cognitive biases" in the psychological factors (premortems and AI devil's advocates). In laboratory groups, a member who truly disagreed led to more and better solutions than a devil's advocate. This held in each of three role-play versions tested. This suggests that role-play has trouble reproducing the effect of real dissent (Nemeth et al., 2001). This argues for protecting and recruiting people who actually disagree, not only assigning the role.
  • Protected dissent channels (Approved but not scaled, C). The U.S. State Department's Dissent Channel lets staff send dissenting policy views straight to the Secretary and senior officials. Authors are "expressly protected from any penalty, reprisal, or recrimination" (Gill and Epstein, 2017). From 1972 to 2017 the channel received an average of almost nine messages a year (Berndt and Holzer, 2023). A Congressional Research Service review found that publicly known messages appear to rarely affect policy. The review cited the 1971 Blood telegram and dissents on Syria and the 2017 travel ban as cases that did not change policy (Gill and Epstein, 2017). No outcome evaluation was found.
  • Failed: fixing NASA's decision culture after Challenger (Failed or reversed, B). The board that investigated the 2003 loss of the space shuttle Columbia found organizational barriers that "stifled professional differences of opinion" (CAIB, 2003). The board concluded that "the causes of the institutional failure responsible for Challenger have not been fixed," 17 years after the 1986 Challenger accident (Congressional Research Service, 2003). The board called for an independent technical authority with full control over specifications and requirements. This authority would be separate from schedule and budget management (CAIB, 2003).

Sources cited on this page

  1. Asch, S. E. (1956). Studies of independence and conformity: I. A minority of one against a unanimous majority. Psychological Monographs, 70(9), Whole No. 416, 1-70. DOI A Strong: for the laboratory effect, B for real-world generalization
  2. Janis, I. L. (1972). Victims of Groupthink. Boston: Houghton Mifflin. C Limited
  3. Becker, J., Brackbill, D., & Centola, D. (2017). Network dynamics of social influence in the wisdom of crowds. Proceedings of the National Academy of Sciences, 114(26), E5070-E5076. link B Moderate
  4. Berndt, S., & Holzer, H. (2023, December). The State Department Dissent Channel: History and impact. The Foreign Service Journal. American Foreign Service Association. link C Limited
  5. Gill, C. R., & Epstein, S. B. (2017, April 7). The U.S. Department of State's "Dissent Channel" (CRS In Focus IF10631). Congressional Research Service. link B Moderate
  6. Lorenz, J., Rauhut, H., Schweitzer, F., & Helbing, D. (2011). How social influence can undermine the wisdom of crowd effect. Proceedings of the National Academy of Sciences, 108(22), 9020-9025. link C Limited
  7. 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. link C Limited
  8. Levine, S. S., Apfelbaum, E. P., Bernard, M., Bartelt, V. L., Zajac, E. J., & Stark, D. (2014). Ethnic diversity deflates price bubbles. Proceedings of the National Academy of Sciences, 111(52), 18524-18529. link B Moderate
  9. Lu, L., Yuan, Y. C., & McLeod, P. L. (2012). Twenty-five years of hidden profiles in group decision making: A meta-analysis. Personality and Social Psychology Review, 16(1), 54-75. link B Moderate
  10. Mullen, B., Johnson, C., & Salas, E. (1991). Productivity loss in brainstorming groups: A meta-analytic integration. Basic and Applied Social Psychology, 12(1), 3-23. link B Moderate
  11. Sommers, S. R. (2006). On racial diversity and group decision making: Identifying multiple effects of racial composition on jury deliberations. Journal of Personality and Social Psychology, 90(4), 597-612. link B Moderate

Every source for this factor is listed on the social factor page.