---
title: Group polarization
url: https://propagandaarchive.org/learn/group-polarization/
description: When people who already lean the same way discuss an issue together, they tend to come out leaning further in that direction than when they went in.
section: Learn
---

![Illustration: a Cold War intelligence collage of a non-aligned-movement conference and a map spanning Latin America, Africa and Asia.](https://propagandaarchive.org/static/generated/detail/hero-cold-war-global-south.jpg)

[Learn](https://propagandaarchive.org/learn/) · Group identity

# Group polarization

In brief

Like-minded people who discuss an issue together usually end up more extreme in the same direction, a well-replicated lab result; the popular claim that online echo chambers drive polarization is much weaker.

On this page

- [In brief](https://propagandaarchive.org/learn/group-polarization/#in-brief)
- [What the research shows](https://propagandaarchive.org/learn/group-polarization/#what-the-research-shows)
- [Has it held up?](https://propagandaarchive.org/learn/group-polarization/#has-it-held-up)
- [How it shows up in political messaging](https://propagandaarchive.org/learn/group-polarization/#how-it-shows-up-in-political-messaging)
- [How to spot it](https://propagandaarchive.org/learn/group-polarization/#how-to-spot-it)
- [Persuasion or manipulation?](https://propagandaarchive.org/learn/group-polarization/#persuasion-or-manipulation)
- [Archive techniques that lean on this](https://propagandaarchive.org/learn/group-polarization/#archive-techniques-that-lean-on-this)
- [Related topics](https://propagandaarchive.org/learn/group-polarization/#related-topics)
- [Sources](https://propagandaarchive.org/learn/group-polarization/#sources)
- [Images](https://propagandaarchive.org/learn/group-polarization/#images)

On this page

*Area: [Group identity](https://propagandaarchive.org/learn/group-identity/) · Evidence:**strong**([How we rate the evidence](https://propagandaarchive.org/learn/evidence-ratings/)). The core lab finding is one of the best replicated in social psychology; the popular claim that online "echo chambers" drive polarization is much weaker.*

## In brief

When people who already lean the same way discuss an issue together, they tend to come out leaning further in that direction than when they went in. Liberals who talk with liberals become more liberal; conservatives who talk with conservatives become more conservative. This is group polarization. It is well established for face-to-face groups. Whether social media reproduces it at the scale often claimed is a separate, and much less settled, question.

## What the research shows

The finding grew out of early-1960s studies of the "risky shift," in which groups seemed to recommend riskier choices than their members did alone. Later work showed that groups did not always become riskier; they became more extreme in whatever direction their members already leaned [1].

Serge Moscovici and Marisa Zavalloni showed this with political attitudes in France. Secondary-school students gave their private opinions on Charles de Gaulle, whom they viewed favorably on average, and on Americans, whom they viewed somewhat unfavorably. After discussing the items in small groups and agreeing on a group answer, their views of de Gaulle became more favorable and their views of Americans more negative, and the shifts largely persisted when they answered privately afterward [2].

David Myers and Helmut Lamm reviewed the growing literature in 1976 and set out the two main explanations that are still used [1]. The first is **social comparison**: people want to be seen as good members of their group, and once they learn where the group stands, they position themselves a little beyond the average to stand out in the approved direction. The second is **persuasive arguments**: in a like-minded group, most of the arguments people hear support the side they already favor, including some they had not thought of, so they become more convinced. Daniel Isenberg's 1986 meta-analysis (a study pooling many studies) found solid support for both mechanisms, with the persuasive-arguments effect somewhat larger [3].

![Diagram: a like-minded group discusses an issue; social comparison and persuasive arguments each push views further in the direction the group already leaned](https://propagandaarchive.org/learn/images/psych-group-polarization-mechanisms.svg)

*Drawn by FRAME from Myers and Lamm (1976) [1] and Isenberg (1986) [3]. CC BY 4.0 (FRAME).*

The legal scholar Cass Sunstein argued that this is close to a "law" of group behavior with direct political consequences: when like-minded people deliberate in isolated enclaves, they move toward more extreme positions [4]. A field test in Colorado in 2005, run by David Schkade, Sunstein and Reid Hastie, brought residents of liberal Boulder and conservative Colorado Springs into small same-town groups to discuss climate change, affirmative action and civil unions for same-sex couples. After discussion, Boulder groups became more liberal and Colorado Springs groups more conservative; views inside each group also became more uniform, and the gap between the two towns widened [5].

## Has it held up?

The laboratory effect has been reproduced across many countries, topics and decades, and the Isenberg meta-analysis puts it on a firm footing [3]. Its size depends on how like-minded the group really is, how much members care about belonging to it, and whether dissenting arguments are aired.

The extension to online "echo chambers" is where the evidence gets mixed, and it is worth saying clearly. Andrew Guess, tracking the actual web browsing of American adults, found that most people's online news diets are fairly moderate and overlapping; heavily one-sided consumption is concentrated in a small, highly partisan minority [6]. Levi Boxell, Matthew Gentzkow and Jesse Shapiro found that US polarization grew fastest among older Americans, the group least likely to use the internet and social media [7]. In a large experiment during the 2020 US election, Brendan Nyhan and colleagues reduced Facebook users' exposure to like-minded sources by about a third for three months and found no measurable effect on polarization [8].

Exposure to the other side is not a simple cure either. Christopher Bail and colleagues paid Twitter users to follow a bot that retweeted messages from the opposing party for a month. Republicans became substantially more conservative afterward; Democrats became slightly more liberal, though that change was not statistically reliable [9]. Contact with opponents' messages, at least in that form, pushed some people further apart.

## How it shows up in political messaging

- **Courts.** Sunstein and colleagues found that on US federal appeals courts, judges appointed by Republican presidents vote more conservatively when sitting with two other Republican appointees than when sitting with Democratic appointees, and Democratic appointees show the mirror image [10]. Like-minded panels move further in their shared direction.
- **Deliberative forums.** The Colorado study shows the same pattern among ordinary citizens of the left and right when each discusses only with its own side [5].
- **Radical groups.** Clark McCauley and Sophia Moskalenko list group polarization among the group-level mechanisms by which small, isolated political groups of any ideology move toward violence, as members compete to show commitment and hear only reinforcing arguments [11].
- **Movement and party messaging.** Internal communications to activists and loyal supporters, on every side, tend to reward the most committed-sounding voice, which is the social-comparison mechanism at work. Rallies, closed forums and members-only channels supply mostly one-sided arguments, which is the persuasive-arguments mechanism.

![Marchers at the 1978 March for Life carrying hand-lettered signs including "Abortion is murder"](https://propagandaarchive.org/collection/images/right-to-life-abortion-is-murder-1978.jpg)

*["Right to Life" demonstration, at White House and Capitol](https://propagandaarchive.org/collection/right-to-life-abortion-is-murder-1978/): Anti-abortion demonstrators, March for Life, Washington, D.C., 23 January 1978. Public domain. One side of a US debate at a rally of its own supporters.*

![Poster for Abortion Action Week, 1–6 May 1972, showing marchers behind a banner reading "Women unite: abortion is our right!!"](https://propagandaarchive.org/collection/images/abortion-action-week-1972.jpg)

*[Abortion Action Week, May 1-6, 1972](https://propagandaarchive.org/collection/abortion-action-week-1972/): Poster, Women's National Abortion Action Coalition, 1972. Public domain. The other side of the same debate, in its own movement material.*

## How to spot it

- A group's position has hardened over time without new evidence, only more discussion among the already convinced.
- Status within the group goes to whoever takes the strongest line; moderates are called weak or traitors.
- Arguments from the other side are absent, or appear only in their weakest form.
- The phrase "everyone here agrees" is doing the work of an argument.
- A community's statements grow more extreme with each round of internal debate.

## Persuasion or manipulation?

Organizing people who already share a goal is a normal and legitimate part of politics, and it naturally involves talking mainly with allies. Honest writing for a like-minded audience still presents the strongest opposing arguments fairly and leaves room for disagreement inside the group. It tips into manipulation when a message deliberately shuts out other views, rewards members for outbidding each other in extremity, or treats any doubt as betrayal, so that a group's movement toward the extreme is engineered rather than reasoned.

## Archive techniques that lean on this

- [Identity fusion](https://propagandaarchive.org/techniques/identity-fusion/) — Identity fusion: when questioning the group line feels like treason, the social-comparison pressure that drives polarization gets stronger.
- [Overton window shifting](https://propagandaarchive.org/techniques/overton-shift/) — Overton window shifting: positions voiced first inside like-minded enclaves can become the group's new normal and then be carried outward.

## Related topics

- [Conformity](https://propagandaarchive.org/learn/conformity/) — Conformity
- [Social identity](https://propagandaarchive.org/learn/social-identity/) — Social identity
- [Social proof](https://propagandaarchive.org/learn/social-proof/) — Social proof
- [Pluralistic ignorance](https://propagandaarchive.org/learn/pluralistic-ignorance/) — Pluralistic ignorance
- [Moral outrage](https://propagandaarchive.org/learn/moral-outrage/) — Moral outrage

## Sources

1. Myers, D. G., & Lamm, H. (1976). The group polarization phenomenon. *Psychological Bulletin*, 83(4), 602–627. https://doi.org/10.1037/0033-2909.83.4.602
2. Moscovici, S., & Zavalloni, M. (1969). The group as a polarizer of attitudes. *Journal of Personality and Social Psychology*, 12(2), 125–135. https://doi.org/10.1037/h0027568
3. Isenberg, D. J. (1986). Group polarization: A critical review and meta-analysis. *Journal of Personality and Social Psychology*, 50(6), 1141–1151. https://doi.org/10.1037/0022-3514.50.6.1141
4. Sunstein, C. R. (2002). The law of group polarization. *Journal of Political Philosophy*, 10(2), 175–195. https://doi.org/10.1111/1467-9760.00148
5. Schkade, D., Sunstein, C. R., & Hastie, R. (2007). What happened on deliberation day? *California Law Review*, 95(3), 915–940.
6. Guess, A. M. (2021). (Almost) everything in moderation: New evidence on Americans' online media diets. *American Journal of Political Science*, 65(4), 1007–1022. https://doi.org/10.1111/ajps.12589
7. Boxell, L., Gentzkow, M., & Shapiro, J. M. (2017). Greater Internet use is not associated with faster growth in political polarization among US demographic groups. *Proceedings of the National Academy of Sciences*, 114(40), 10612–10617. https://doi.org/10.1073/pnas.1706588114
8. Nyhan, B., Settle, J., Thorson, E., Wojcieszak, M., Barberá, P., et al. (2023). Like-minded sources on Facebook are prevalent but not polarizing. *Nature*, 620(7972), 137–144. https://doi.org/10.1038/s41586-023-06297-w
9. Bail, C. A., Argyle, L. P., Brown, T. W., Bumpus, J. P., Chen, H., Hunzaker, M. B. F., Lee, J., Mann, M., Merhout, F., & Volfovsky, A. (2018). Exposure to opposing views on social media can increase political polarization. *Proceedings of the National Academy of Sciences*, 115(37), 9216–9221. https://doi.org/10.1073/pnas.1804840115
10. Sunstein, C. R. (2009). *Going to extremes: How like minds unite and divide*. New York: Oxford University Press.
11. McCauley, C., & Moskalenko, S. (2008). Mechanisms of political radicalization: Pathways toward terrorism. *Terrorism and Political Violence*, 20(3), 415–433. https://doi.org/10.1080/09546550802073367

## Images

| Image | Source | Licence |
| --- | --- | --- |
| Diagram: two explanations for group polarization | Drawn by FRAME from Myers and Lamm (1976) and Isenberg (1986) (`make_group_identity_charts.py`) | CC BY 4.0, FRAME |
| March for Life (1978) | FRAME gallery, ["Right to Life" demonstration, at White House and Capitol](https://propagandaarchive.org/collection/right-to-life-abortion-is-murder-1978/), Library of Congress | Public domain (no known restrictions, per LoC) |
| Abortion Action Week poster (1972) | FRAME gallery, [Abortion Action Week, May 1-6, 1972](https://propagandaarchive.org/collection/abortion-action-week-1972/), Library of Congress | Public domain (no known restrictions, per LoC) |

### Image gaps

No photograph of the Colorado deliberation groups or of appellate panels was found in a verified open source, and the Schkade study's group means could not be read from an accessible copy, so no data chart is drawn; the mechanism diagram carries no numbers. The two rally images are illustrations of movement settings, not of the studies.

## Techniques

[Identity fusion](https://propagandaarchive.org/techniques/identity-fusion/) [Overton window shifting](https://propagandaarchive.org/techniques/overton-shift/)

## Related

- [Conformity](https://propagandaarchive.org/learn/conformity/)
- [Social identity](https://propagandaarchive.org/learn/social-identity/)
- [Social proof](https://propagandaarchive.org/learn/social-proof/)
- [Pluralistic ignorance](https://propagandaarchive.org/learn/pluralistic-ignorance/)
- [Moral outrage](https://propagandaarchive.org/learn/moral-outrage/)
- [Identity fusion](https://propagandaarchive.org/techniques/identity-fusion/)
- [Overton window shifting](https://propagandaarchive.org/techniques/overton-shift/)