● Cognitive bias
← First documented by psychologists George Quattrone and Edward Jones, 1980. Landmark demonstration: Bernadette Park and Myron Rothbart's 1982 sorority study.
The tendency to see people outside your own group as basically all the same, while seeing your own group as full of distinct individuals, even when the two groups are, in reality, equally varied.
Also called: outgroup homogeneity bias. Not the same as the cross-race effect, its face-recognition-specific cousin (Section 01).
The tendency to perceive members of a group you don't belong to (an outgroup) as more similar to one another, less varied in personality, opinion, and behavior, than members of a group you do belong to (your ingroup), even when the two groups actually contain the same amount of real variation.
Plain version
Think about your own friend group or school. You could probably describe five wildly different personalities without much effort. Now think about some group you're not part of: a rival school, a different generation, a country you've never visited, a political party you don't support. It's a lot easier to picture them as one type of person. That gap isn't because your group is actually more diverse. It's because you know your own group from the inside, one person at a time, and you mostly know other groups from the outside, as a category.
Aliases
Outgroup homogeneity bias is used interchangeably with outgroup homogeneity effect. Both refer to the same phenomenon.
Not the same as the cross-race effect
The cross-race effect (also called own-race bias) is a narrower, face-recognition-specific version of this pattern: people are measurably worse at telling apart and recognizing faces of races other than their own. It's best understood as one specific, well-studied application of outgroup homogeneity, not a separate phenomenon, and it comes with its own real-world stakes (Section 05). Related distinction, not a swap-in synonym.
Scope
Gets confused with
In-group bias and the cross-race effect: precise distinctions in Section 09 and above.
One of the hardest biases on this site to catch in the moment, and for almost exactly the same reason as confirmation bias. It doesn't feel like a snap judgment, it feels like an observation from experience ("I've seen enough of them to know"). There's an easy, reasonable-sounding cover story available the whole time, and believing it delivers real comfort: a simpler social world and a stronger sense of belonging to your own group. Nothing here trips a wire.
Why your brain does this
You spend most of your life inside your own group: family, friends, classmates, teammates, coworkers. You meet the same people over and over, in different moods, on different days, doing different things. That builds a huge, detailed mental library of individual exemplars for your ingroup. For most outgroups, you have far fewer direct encounters to draw on, and the ones you do have are more likely to be brief, one-off, or filtered through media and secondhand stories. Researchers Patricia Linville, Gregory Fischer, and Peter Salovey (1989) modeled this directly: with a smaller effective sample of examples, any group is going to look more uniform, purely as a matter of statistics, regardless of how varied its members actually are. Treating a small, thin sample as a reliable summary of a whole group isn't a broken instinct, it's the same fast, information-saving shortcut your brain uses everywhere it has to generalize from limited data.
Where it misfires
It breaks down specifically when the "small sample" problem gets mistaken for real knowledge. A handful of encounters, a few news stories, or a single stereotyped portrayal end up standing in for an entire group of potentially millions of people, each as individually varied as anyone in your own circle. The shortcut that's harmless when guessing about jellybean flavors in a jar becomes a real problem when the "jar" is a group of people, and the guess turns into a belief about who they are.
The live academic debate, briefly
Gerd Gigerenzer's ecological-rationality tradition would frame this as a generally sound strategy, generalizing from the examples you have, misapplied to a case (large, genuinely diverse human groups) where the available sample badly under-represents the true variety. The more traditional Tversky and Kahneman heuristics-and-biases framing treats it as a specific error rooted in the representativeness heuristic and the "law of small numbers," the same statistical blind spot behind the gambler's fallacy. This page doesn't take a side; both agree the underlying math is the same, and both agree it produces a real distortion.
How it activates, step by step
1. You encounter a small, unrepresentative sample of a group you don't belong to: a few individuals, a stereotype, a handful of news stories or social media clips.
2. Because your mental "library" for that group is thin compared to your ingroup's, that small sample carries more weight than it should.
3. Your brain generalizes the sample into a summary trait: "they're all like this."
4. New information about individual outgroup members tends to get absorbed into the existing summary rather than treated as evidence that the group is actually varied.
Few biases are as directly load-bearing for propaganda as this one. Most techniques that target a group rather than a person start by leaning on the audience's existing sense that "they" are all interchangeable.
This technique is, in a sense, outgroup homogeneity turned into a deliberate message: it hands the audience a single trait and explicitly applies it to an entire group, which lands easily because the audience's brain was already primed to expect the group to be uniform.
Stripping a group of individual identity, showing them only as an undifferentiated mass, a swarm, a faceless crowd, works precisely because a homogeneous group is easier to talk about (and to justify harming) than a collection of distinct individuals.
Blaming "them" for a problem only works cleanly if the audience already accepts that "they" act as a single, coordinated bloc rather than millions of individuals with their own range of views and behavior.
Splitting a population into opposing camps depends on each side seeing the other as one uniform thing to be defeated, rather than as individuals who might actually agree on more than they'd expect.
Outside politics
Rival fandoms and brand tribes run on the same instinct without any political content at all: "all [rival team]'s fans are like that," "typical [other brand's] customer." This is an everyday, illustrative pattern rather than a documented study, but anyone who has watched an online brand or sports rivalry will recognize it: flattening the other side into one caricature makes the ingroup feel more coherent and more right by comparison.
What increases it
Strong: having few or thin exemplars of the outgroup to draw on. This is the core mechanism from Section 02 (Linville, Fischer & Salovey, 1989): the fewer distinct individuals you can call to mind, the more homogeneous the group looks, independent of how varied it really is.
Moderate: group size and minority status. A 2024 study on the "minority-groups homogeneity effect" found that people tend to see members of smaller or minority groups, in general, as more similar to each other than members of a large majority group, a pattern that operates somewhat independently of simple ingroup-versus-outgroup status (PubMed, 2024).
Weaker than expected: intergroup competition and low-quality contact. Some research links perceived threat and rivalry between groups to stronger homogeneity perceptions, though this strand is less thoroughly replicated than the familiarity account above, and Gordon Allport's own 1954 contact-hypothesis work warned that casual, low-quality contact between groups can leave people falling back on stereotypes rather than revising them.
What does NOT predict it
Jones, Wood and Quattrone's 1981 study of four undergraduate clubs found the effect held "regardless of the number of in-group or out-group members known, or [...] the degree of preference for the in-group." In other words, simply knowing more outgroup members, or liking your own group more, didn't cleanly explain who showed the bias and by how much. The effect appears to run deeper than raw contact counts or in-group favoritism alone.
State-dependent factors
There isn't strong, specific evidence that stress or time pressure directly amplifies this particular bias the way they do for some others on this site. The clearer state-dependent lever is exposure: brief, superficial, or one-sided exposure to an outgroup (a single news clip, one viral post) tends to reinforce the homogeneous impression, while sustained, equal-status contact tends to work against it.
The pattern across all four: the effect shows up whether the groups in question are political parties, college social clubs, racial groups in a police lineup, or sports fans, and it shows up regardless of how much genuine variety actually exists inside the group being flattened.
First named
Psychologists George Quattrone and Edward Jones formally documented the effect in "The perception of variability within in-groups and out-groups: Implications for the law of small numbers" (Journal of Personality and Social Psychology, 1980), tying it explicitly to the same small-sample statistical error behind the gambler's fallacy.
The landmark study
Edward Jones, George Wood, and George Quattrone's 1981 follow-up study asked members of four undergraduate clubs to rate how varied their own club was, and how varied the other three clubs were, on a set of personal characteristics. Every club rated itself as more heterogeneous than it rated the other three, and the pattern held regardless of how many members of the other clubs a person actually knew, or how much they preferred their own club (Jones, Wood & Quattrone, Personality and Social Psychology Bulletin, 1981).
Since then
Park and Rothbart's 1982 sorority study (Section 05) extended the finding and connected it to social identity theory. Patricia Linville, Gregory Fischer, and Peter Salovey's 1989 work modeled the "thin sample" mechanism mathematically (Section 02). Later research identified the cross-race effect as a specific, high-stakes face-recognition version of the same pattern (Section 01, Section 05), and more recent work, including a 2024 study on the "minority-groups homogeneity effect," continues to refine which factors (group size, contact quality, threat) push the effect up or down. The core finding, that people see outgroups as less varied than ingroups, is well replicated across decades and is not a contested result; what remains actively studied is exactly which mechanism does the most work and how strongly various interventions can counter it.
Cross-tradition note
See Section 02 for Gerd Gigerenzer's ecological-rationality framing, the relevant non-English research tradition for this page's adaptive-origin section.
Score: 1 out of 5, tied with confirmation bias for one of the lowest on the whole site (see the breakdown in the card after Section 01).
The in-the-moment question
Before saying or thinking "they're all like that": how many actual individuals from that group am I basing this on, and how well do I actually know them? If the honest answer is "a handful" or "mostly what I've seen online," the confidence in the generalization is bigger than the evidence supporting it.
Why this is so hard to catch
It doesn't feel like a stereotype in the moment, it feels like a pattern you've genuinely noticed. There's an easy story available afterward no matter what happens ("I've just met enough of them"), and believing it delivers something real: a simpler, more predictable social world, and a quiet boost to your own group's sense of being the more varied, more human one. None of that produces an internal alarm.
Debiasing research note
Mixed but not hopeless. Allport himself warned that low-quality, superficial contact can backfire and reinforce stereotypes instead of reducing them, so "just put people in the same room" is not a reliable fix on its own; the specific conditions matter. The Ahler and Sood correction-information finding is promising but comes from one study and one country's two-party system, and should be treated as a strong hint rather than a settled answer.
Commonly co-occurs with
Tell me how outgroup homogeneity effect interacts with in-group bias, illusory correlation, or fundamental attribution error.
Often mistaken for
Related, not identical. In-group bias is about favoring your own group, giving it better treatment, more trust, and more benefit of the doubt. Outgroup homogeneity is about perceived variety, seeing the outgroup as less varied, regardless of whether you actually like or dislike them. The two usually travel together, but a group can be seen as homogeneous without being disliked, and disliked without being seen as uniform.
The cross-race effect is a specific, face-recognition-level manifestation of outgroup homogeneity, not a separate bias. See Section 01 and Section 05.
Compounding effect
Outgroup homogeneity feeds illusory correlation directly: because outgroup behavior already looks more uniform, a handful of vivid, negative examples get remembered as representative of the whole group far more easily than they would for a group you see as varied (the classic Hamilton and Gifford 1976 illusory-correlation studies were built around exactly this setup). It also feeds fundamental attribution error: once a group is flattened into "they're all like that," an individual outgroup member's behavior gets read as revealing their group's fixed nature rather than as a response to their specific situation.
Primary sources
Accessible reading