Evidence locker · triangulate the cluster
Stereotyping
Stereotyping is judging an individual by traits or behaviors imputed from their social-group membership rather than from individuating information about the person.
Hamilton & Gifford illusory correlation
— · n?
Steele & Aronson stereotype threat
— · n≈38
Correll 'Police Officer's Dilemma' shooter bias
95% · n?
Bertrand & Mullainathan resume audit
9.65% · n≈5000
★ Killing of Amadou Diallo
field incident · 1999
★ Labor-market name discrimination (Ber…
field incident · 2004
The claim
Stereotyping is the cognitive process of attributing characteristics to a person on the basis of a group category (race, gender, age, etc.), a shortcut Gordon Allport (1954) traced to the mind's normal tendency to form categories.
Reading the scatter. ★ chits are real-world incidents (vivid, but n = 1, so they sit low on strength). The cluster is split: the best-powered field evidence lands upper-right, while several lab paradigms sit lower and weaker. Read the construct paradigm by paradigm.
Exhibit index
Hamilton & Gifford illusory correlation. Participants overestimated the frequency of negative behaviors in the smaller group B and rated group B more negatively, producing a stereotype-like differentiation with no actual difference in the groups' behavior profiles.
David L. Hamilton, Robert K. Gifford, 1976 · Undergraduate samples (tens of participants per study)
Steele & Aronson stereotype threat. Black participants underperformed Whites in the diagnostic/threat framing but performed comparably in the non-diagnostic framing, interpreted as situational threat from a negative ability stereotype.
Claude M. Steele, Joshua Aronson, 1995 · Four studies, Stanford undergraduates; small Black subsamples (avg ~38)
Correll 'Police Officer's Dilemma' shooter bias. Participants were faster to shoot armed Black than armed White targets and faster to not-shoot unarmed White than unarmed Black targets; they made more errors consistent with a 'Black = threat' stereotype. Bias tracked knowledge of cultural stereotypes, not personal prejudice.
Joshua Correll, Bernadette Park, Charles M. Judd, Bernd Wittenbrink, 2002 · Multiple studies, undergraduate and community participants
Bertrand & Mullainathan resume audit. White-sounding names received 50% more callbacks than identical resumes with Black-sounding names; returns to resume quality were larger for White names. Demonstrates real labor-market consequences of name-based racial stereotyping.
Marianne Bertrand, Sendhil Mullainathan, 2004 · ~5,000 resumes to ~1,300 employment ads
Why the shortcut forms
The dominant account is cognitive-economy/categorization (Allport 1954): the mind automatically sorts people into groups to reduce information load, and group-level associations then color judgments of individuals. Hamilton & Gifford (1976) added a purely informational mechanism, distinctiveness-based illusory correlation, in which the co-occurrence of two rare events (a minority group + an infrequent negative behavior) is especially salient and over-encoded, generating a stereotype with no real group difference. Competing emphasis comes from motivational/social-identity accounts (in-group favoritism, system justification) and from a 'kernel-of-truth' view holding that some stereotypes track real statistical group differences.
Cognitive (categorization, illusory correlation, attention to distinctiveness) vs. motivational/social-identity (in-group bias, prejudice as evaluative content distinct from descriptive stereotype content) vs. ecological/accuracy ('kernel of truth') accounts that argue stereotypes can be partly statistically accurate. These are not mutually exclusive; most modern work treats stereotyping as multiply determined.
Replication
Mixed
Stereotyping as category-based judgment is a broad, well-established phenomenon, and the distinctiveness-based illusory-correlation mechanism has meta-analytic support (Mullen & Johnson 1990: moderate-to-large effects, d ~0.83-1.0 for negative distinctive behaviors). Real-world discrimination effects (e.g., Bertrand & Mullainathan 2004 callback gap) replicate well in audit literatures. However, specific high-profile sub-paradigms are contested: stereotype-threat effects on girls' math show strong publication-bias signals (Flore & Wicherts 2015: d=.22 uncorrected, falling to d=.07 and non-significant after trim-and-fill), and a large pre-registered replication (Flore, Mulder & Wicherts 2018, N=2064 Dutch students) found NO overall stereotype-threat effect. Shooter-bias reaction-time effects survive meta-analysis but are small (Mekawi & Bresin 2015, dav=-0.13). Honest summary: the umbrella construct is robust, but several celebrated effects are weaker or failed than textbooks imply.
'Stereotyping' is a broad construct, not a single experiment, so 'replication' must be assessed paradigm by paradigm; I report it as 'mixed' rather than 'robust' to honor that. The category-based-judgment and real-world discrimination evidence (audit studies, illusory-correlation meta-analysis) is strong, but two famous offshoots are weak: stereotype-threat effects on girls' math math collapse after publication-bias correction (d=.07, n.s.) and failed a large pre-registered replication (N=2064), and shooter-bias reaction-time effects are small (dav=-0.13). Steele & Aronson (1995) had small samples (avg ~38 Black participants) and no single standardized effect size; I did not fetch the full PDF, so its internal numbers are reported from secondary critique and the original abstract, not a line-by-line read. Hamilton & Gifford's original standardized effect sizes were taken from the Mullen & Johnson (1990) meta-analysis, not the 1976 paper itself. The Mekawi & Bresin (2015) figures dav=-0.13, 95% CI [-0.19, -0.06], k=32 for the armed-trial RT subanalysis are consistent with characterizations in citing literature but could not be independently confirmed from retrievable text (ScienceDirect returns 403); if drawn from the paper's Table 2 or similar subanalysis, they are retained, but note they are sourced secondhand relative to this dossier's direct retrieval. The Diallo case is a real, dated event widely cited as illustrative of shooter-bias dynamics, but a single police shooting cannot by itself prove stereotyping caused that specific outcome; it is documented as a research touchstone, not a controlled measurement.
What to log instead
Replace category-based inference with individuating information and structured, criterion-based judgment: force yourself to gather person-specific data, blind the decision to group cues where possible (e.g., anonymized resumes), and use fixed rating rubrics so a single salient group feature can't drive the call.
Bertrand & Mullainathan (2004) showed that removing or changing only the name-cued group signal on otherwise identical resumes changed callback rates by 50%, demonstrating that the group cue, not the candidate's qualifications, was driving judgment, the exact leverage point that name-blind screening targets.
Catch it in the act
You catch yourself predicting one person's traits, abilities, or behavior mainly from a group label (their race, gender, age, accent, name) rather than from what you actually know about that individual, and you'd revise the judgment if you paused to weigh individuating evidence.
Adjacent lockers
- Illusory correlationmechanistically-linkedHamilton & Gifford's proposed cognitive engine of stereotype FORMATION; perceiving a relationship between group membership and traits that isn't really there.
- PrejudicesiblingStereotype = descriptive (cognitive) content about a group; prejudice = evaluative (affective) negative attitude. Often co-occur but conceptually distinct.
- Stereotype threatchildA downstream performance effect of being aware of a negative stereotype about one's group; far more contested in replication than stereotyping itself.
- Outgroup homogeneity effectmechanistically-linkedTendency to see outgroup members as more alike ('they're all the same'), which facilitates applying a single stereotype to individuals.
- Representativeness heuristiceasily-confusedJudging probability by resemblance to a prototype; overlaps with stereotyping but is a general probability heuristic, not specifically about social groups.
- Fundamental attribution erroreasily-confusedOver-attributing others' behavior to disposition; differs in that it concerns individuals' traits generally, not group-membership-based imputation.