Social Discrimination

Abstract

Social discrimination is a form of social behavior: the differential, usually disadvantageous, treatment of people because of their membership in a social category rather than their individual qualities. This article develops discrimination as the behavioral component of intergroup bias, held apart from the stereotype that supplies its beliefs and the prejudice that supplies its feeling. It traces the classic finding that attitudes and discriminatory acts can come apart, sets out the cognitive route from a group categorization to a biased decision, and reviews the field-experimental methods — audit and correspondence studies — that measure discrimination directly in hiring, housing, and markets. It then surveys what reduces discrimination, why contact remains the best-supported lever, and why decades of field experiments show so little change over time. A worked callback-rate example and three interactive demonstrations make the ideas concrete.

Keywords: social discrimination, prejudice, stereotyping, audit study, intergroup contact

Social discrimination is differential treatment — in hiring, housing, lending, policing, schooling, or everyday interaction — directed at a person because of their membership in a social category such as race, sex, age, disability, or national origin. In the classic tripartite scheme it is the behavioral component of intergroup bias, distinct from the stereotype that is its cognitive component and the prejudice that is its affective one (Dovidio & Gaertner, 2000). The three are correlated but dissociable, and the gap between what people feel and what they do is one of the oldest and most instructive findings in the field (LaPiere, 1934).

The modern treatment is cognitive and structural at once. A discriminatory act can issue from an automatically activated association that controlled processing fails to override, but discrimination also operates at the level of institutions and policies that generate disparate outcomes with no prejudiced individual in the loop (Jones, 2000). The sections below move from the individual architecture of a discriminatory decision to the field methods that quantify discrimination in the world and the interventions that do, and do not, reduce it.

What Social Discrimination Is

Social discrimination is operationally the differential treatment of a person on the basis of the social group to which they are assigned, independent of their individual merit (Dovidio & Gaertner, 2000). It is worth holding three allied terms apart. A stereotype is the cognitive component, a belief about the attributes of a group; prejudice is the affective component, a negative evaluation of the group; and discrimination is the behavioral component, the act of treating the group's members differently. The components are correlated but dissociable, and much of what is distinctive about contemporary discrimination lies in their coming apart: explicit prejudice has grown less socially acceptable while discriminatory outcomes have proved far more durable.

That attitudes and actions need not align was shown strikingly before the modern apparatus existed. LaPiere travelled the United States with a young Chinese couple and was refused service at only one of some 250 establishments, yet when he later wrote to those same businesses asking whether they would serve Chinese guests, the overwhelming majority said they would not — a flat contradiction between expressed attitude and actual behavior that has framed attitude-behavior research ever since (LaPiere, 1934). The lesson is not that attitudes are irrelevant but that the link from an internal disposition to an overt act is mediated by the situation, a theme that recurs throughout the study of discrimination.

Discrimination is also not only an individual act. Jones distinguished three levels — institutionalized discrimination embedded in the structures and norms of a society, personally mediated discrimination between individuals, and internalized acceptance by members of stigmatized groups of negative messages about their own worth — and argued that the institutional level does the most damage while attracting the least attention (Jones, 2000). The cognitive account of the individual biased decision and the structural account of disparate outcomes are complementary levels of description of one phenomenon, not rivals.

Types of Social Discrimination

The Medical Subject Headings vocabulary files Social Discrimination (D063508) as a narrower term under social behavior and lists six narrower descriptors directly beneath it, shown in Table 1. These indexed subtypes do not exhaust the kinds of discrimination the research literature distinguishes — discrimination by social class, religion, or body weight are all well studied — because the table is an indexing classification, a controlled vocabulary built so that the published literature can be retrieved, not a theory that carves discrimination at its natural joints. It records how the literature is indexed, not what the true types are, and the subtypes below are neither mutually exclusive nor jointly exhaustive: a single person can face several at once, and each is itself the subject of a dedicated article.

The Cognitive Architecture of Discrimination

The root of discrimination lies in social categorization itself. The minimal group experiments showed that assigning people to groups on a basis they knew to be trivial — even an ostensibly random coin toss — was enough to make them allocate more reward to anonymous ingroup members than to outgroup members, in the absence of any conflict of interest, prior hostility, or personal gain (Tajfel et al., 1971). Social identity theory explains the result: because people draw part of their self-concept from the groups they belong to, they are motivated to see those groups as favorably distinct, and the simplest route to positive distinctiveness is to treat the ingroup better than the outgroup. Discrimination, on this account, does not require animus or competition; mere categorization supplies a standing pressure toward differential treatment that the mechanisms below then shape and release.

Why does discrimination persist in people who sincerely disavow prejudice? The justification-suppression model offers a general answer: people harbor prejudices that they are motivated to suppress because norms and self-image forbid their open expression, and a genuine prejudice is converted into an expressed discriminatory act only to the extent that the situation supplies a justification that releases the suppressed impulse (Crandall & Eshleman, 2003). Discrimination therefore appears not where prejudice is strongest but where justification is most available, which is why it so often hides behind a non-group-based rationale. The aversive-racism tradition documents exactly this signature: people who endorse egalitarian values do not discriminate when a candidate's merits are unambiguous, but do discriminate when the situation is ambiguous enough to supply an innocent explanation (Dovidio & Gaertner, 2000). The JustificationSuppressionExplorer lets the reader vary prejudice, suppression, and justification and watch expressed discrimination appear.

The content that discrimination acts on is not arbitrary. The stereotype content model holds that group stereotypes vary along two dimensions, warmth and competence, whose values follow from a group's structural position, and that the resulting quadrants predict distinct emotions — admiration, envy, pity, or contempt — and distinct discriminatory tendencies, from active harm to passive neglect (Fiske et al., 2002). A group seen as warm but incompetent (often older adults) draws paternalistic, pitying treatment; one seen as competent but cold draws envious hostility. The model explains why different targeted discriminations — the subtypes of Table 1 — take systematically different forms rather than being interchangeable expressions of a single animus. Figure 1 sets out the four quadrants and the discriminatory tendency each licenses.

Measuring Discrimination

Because attitudes and behavior diverge, and because few people will admit to discriminating, the cleanest evidence comes from field experiments that measure the act directly. In an audit study, matched testers who differ only in the group characteristic of interest apply for the same jobs or apartments; in a correspondence study, the application is on paper and the group is signalled by a name or a checkbox, holding every qualification identical. Bertrand and Mullainathan mailed nearly 5,000 fictitious resumes to Boston and Chicago employers, randomly assigning names that sounded White (Emily, Greg) or African-American (Lakisha, Jamal); the White-sounding names drew about 50% more callbacks for the identical resume, and higher resume quality raised callbacks far more for White names than for Black ones (Bertrand & Mullainathan, 2004).

The method extends to other categories and stigmas. Pager's audit study sent matched young men to apply for entry-level jobs, varying whether they reported a felony conviction; a criminal record cut callbacks sharply for all testers, but the penalty was far larger for Black applicants, and a Black applicant with no record fared worse than a White applicant who reported one (Pager, 2003). Field experiments of this kind are now the gold standard for quantifying discrimination precisely because they rule out the qualification differences that make observational wage gaps ambiguous. The ResumeAuditSimulator in the Worked Example below lets the reader run a correspondence study and read off the discrimination it reveals.

Reducing Discrimination

The oldest and best-supported individual-level intervention is intergroup contact. Allport proposed that contact reduces prejudice — and, through it, discrimination — when the groups have equal status within the situation, pursue common goals, cooperate rather than compete, and enjoy the support of authorities (Allport, 1954). Pettigrew's reformulation set out the processes through which contact works — learning about the outgroup, changed behavior, affective ties, and ingroup reappraisal — and extended the theory beyond its original conditions (Pettigrew, 1998). The empirical verdict came from a meta-analysis of more than 500 studies: contact was associated with lower prejudice in the large majority of samples, the optimal conditions strengthened but were not strictly required for the effect, and the reduction generalized beyond the specific individuals encountered (Pettigrew & Tropp, 2006). The ContactConditionsDemo lets the reader toggle Allport's conditions and watch the predicted reduction change.

Interventions that target the automatic component directly have a more sobering record. A large multi-site contest tested seventeen interventions designed to reduce implicit racial preferences and found that although several shifted the implicit measure immediately, none of the effects survived even a short delay (Lai et al., 2016). A broad review of the prejudice-reduction literature reached a similar structural conclusion: the field has many studies but few that measure behavioral discrimination, test durability, or target the policies and environments that generate disparate treatment, so the evidence base for what actually reduces discrimination in the world is thinner than the volume of research suggests (Paluck et al., 2021).

Worked Example

A correspondence study quantifies discrimination as the gap in callback rates between two groups whose applications are otherwise identical. Suppose a researcher sends 1,000 resumes signalling membership in group A and 1,000 identical resumes signalling group B, and records which receive a callback.

Group A's callback rate is 97 / 1,000 = 9.7% and group B's is 65 / 1,000 = 6.5%, so the absolute gap is 9.7 − 6.5 = 3.2 percentage points. Expressed as a ratio, group A receives 9.7 / 6.5 = 1.49 times the callbacks of group B — that is, 49% more callbacks for the identical resume. Put the other way, to win the same 97 callbacks group A obtained, group B would need to send 97 / 0.065 = 1,492 resumes, nearly 500 more than group A sent for the same result. This is the quantity the Bertrand and Mullainathan study estimated at roughly a 50% White advantage, and because the resumes differ only in the signalled group, the gap cannot be attributed to any difference in qualifications (Bertrand & Mullainathan, 2004). The ratio, not the raw difference, is the usual summary because it is comparable across settings with very different base rates. The ResumeAuditSimulator lets the reader set the callbacks for each group and watch the rate, the gap, the ratio, and the equivalent-resume figure update; the numbers here are illustrative of the method, not drawn from a specific dataset.

Discussion

The study of discrimination has moved on two axes at once. Along the first, the unit of analysis shifted from the overtly prejudiced person to the ordinary decision-maker whose egalitarian self-image coexists with biased behavior that surfaces only where the situation supplies cover, a reframing captured by the justification-suppression and aversive-racism accounts (Crandall & Eshleman, 2003; Dovidio & Gaertner, 2000). Along the second, measurement moved from the self-report questionnaire to the field experiment, so that discrimination is now quantified as a behavioral gap in real markets rather than inferred from a stated attitude (Bertrand & Mullainathan, 2004; Pager, 2003).

The open problems lie where the axes meet. The link from any individual's measured prejudice to that individual's discriminatory act is weaker and more situation-dependent than early theory assumed, which is precisely why field experiments that bypass the attitude and measure the act have become decisive. Yet the same methods deliver an uncomfortable verdict on progress: discrimination is reproduced by institutions and routines that attitude change does not reach, so an era of steadily more egalitarian expressed attitudes has coincided with strikingly persistent behavioral gaps (Jones, 2000). The intervention evidence is correspondingly uneven — contact has strong meta-analytic support, implicit-bias retraining has weak and fleeting effects, and the structural level that may matter most is the least tractable to the experimental methods the field knows best (Pettigrew & Tropp, 2006; Lai et al., 2016).

Current Directions

The most striking recent finding concerns change over time. Quillian and colleagues meta-analyzed every available field experiment on hiring discrimination in the United States since 1989 and found no decline whatsoever in discrimination against Black applicants across 25 years, and only a modest decline against Latino applicants — a flat trend that stands in sharp contrast to the steady liberalization of survey-measured racial attitudes over the same period (Quillian et al., 2017). The result reframes discrimination as a behavioral regularity of labor markets that is decoupled from the attitudes people report, and it has pushed the field toward behavioral and structural outcomes.

A second front treats discrimination as a determinant of health. A meta-analytic review established that perceived discrimination is reliably associated with poorer mental health and, more modestly, poorer physical health, operating through chronic stress and health-damaging coping (Pascoe & Smart Richman, 2009). A later review mapped the pathways in detail and argued that reducing the resulting health disparities requires intervening at the structural level, not only the interpersonal (Williams et al., 2019). Alongside this, the intervention literature is turning self-critical: syntheses now call for studies that measure durable behavioral discrimination rather than momentary shifts on an implicit measure, after large tests showed such shifts do not last (Lai et al., 2016; Paluck et al., 2021).

Common Misconceptions

Discrimination is the same thing as prejudice.

Prejudice is an attitude, a feeling about a group; discrimination is behavior, the differential treatment of its members. The two can diverge sharply, as when a person with egalitarian attitudes discriminates where a situation supplies a justification (Crandall & Eshleman, 2003).

Discrimination requires conscious intent to harm.

Institutionalized discrimination produces disparate outcomes through ordinary policies and routines with no prejudiced individual in the loop, and individual discrimination often operates through automatic associations outside awareness (Jones, 2000).

If survey attitudes have improved, discrimination must be declining too.

Field experiments show hiring discrimination against Black applicants did not decline across 25 years, even as survey-measured attitudes grew steadily more egalitarian — behavior and stated attitude move on different tracks (Quillian et al., 2017).

Glossary

Audit study.

A field experiment in which matched testers who differ only in a group characteristic apply for the same jobs or housing, so that any difference in outcome measures discrimination directly.

Aversive racism.

A form of bias in which a person endorsing egalitarian values nonetheless holds negative associations and discriminates only where a non-group-based justification is available.

Correspondence study.

A field experiment using paper applications identical except for a signalled group membership, such as a name, measuring the callback gap it produces.

Discrimination.

The behavioral component of intergroup bias: differential treatment of people on the basis of their membership in a social category.

Field experiment.

A study that manipulates a variable in a real-world setting rather than a laboratory; audit and correspondence studies are the field experiments used to measure discrimination directly.

Institutionalized discrimination.

Differential treatment embedded in the laws, policies, and routine practices of institutions, producing disparate outcomes independently of any individual's intent.

Intergroup bias.

The umbrella term for group-based favoritism and antipathy, comprising a cognitive component (stereotype), an affective component (prejudice), and a behavioral component (discrimination).

Intergroup contact.

Direct interaction between members of different groups, which reduces prejudice and discrimination most reliably under conditions of equal status, common goals, cooperation, and institutional support.

Justification-suppression model.

The account on which a genuine prejudice becomes an expressed discriminatory act only to the degree that a situation supplies a justification releasing the normally suppressed impulse.

Perceived discrimination.

A target's own subjective experience of unfair group-based treatment, a documented pathway from discrimination to poorer mental and physical health.

Prejudice.

The affective component of intergroup bias: a negative attitude or evaluation of people because of their group membership.

Social behavior.

The broad MeSH category of behavior directed toward or involving others, under which social discrimination is classified.

Social discrimination.

Differential, usually disadvantageous, treatment of people because of their membership in a social category rather than their individual qualities.

Social identity theory.

The account on which people derive part of their self-concept from group membership and are therefore motivated toward positive distinctiveness, producing ingroup favoritism from mere categorization.

Stereotype content model.

The framework placing group stereotypes on dimensions of warmth and competence, predicting distinct emotions and distinct forms of discriminatory treatment.

Stereotype.

The cognitive component of intergroup bias: a belief about the attributes shared by members of a social group.

Key Researchers

Gordon W. Allport

(1897–1967). Professor at Harvard University whose 1954 book The Nature of Prejudice grounded prejudice in ordinary categorization and set out the contact hypothesis that still frames the reduction of discrimination. Wikipedia - Wikidata

John F. Dovidio

(Yale University). Carl Iver Hovland Professor of Psychology and a principal architect of aversive-racism theory, documenting how subtle discrimination surfaces when a non-group-based justification is available. ORCID - Wikidata

Susan T. Fiske

(Princeton University). Eugene Higgins Professor of Psychology and Public Affairs who developed the stereotype content model and the BIAS map linking group stereotypes to the emotions and discriminatory behaviors they license. Wikipedia - Wikidata

Elizabeth Levy Paluck

(Princeton University). Professor of Psychology and Public Affairs whose field-experimental work on prejudice and discrimination reduction re-evaluated the contact hypothesis and the evidence base for interventions that change behavior, not just attitudes. ORCID - Wikidata

Thomas F. Pettigrew

(University of California, Santa Cruz). Research Professor of Social Psychology whose reformulation of intergroup contact theory and meta-analytic test, with Linda Tropp, established contact as the best-supported route to reducing prejudice and discriminatory treatment. Wikipedia - Wikidata

David R. Williams

(Harvard University). Florence Sprague Norman and Laura Smart Norman Professor of Public Health whose work established discrimination as a determinant of population health and developed widely used measures of everyday discrimination. Wikipedia - Wikidata

Frequently Asked Questions

What is social discrimination in psychology?

Social discrimination is the behavioral component of intergroup bias: the differential, usually disadvantageous, treatment of people because of their membership in a social category such as race, sex, age, or disability, rather than their individual qualities (Dovidio & Gaertner, 2000).

How is discrimination different from prejudice and stereotyping?

A stereotype is a belief about a group (the cognitive component), prejudice is a negative feeling about it (the affective component), and discrimination is differential behavior toward its members (the behavioral component). They are correlated but dissociable, and attitudes and actions can diverge sharply (LaPiere, 1934).

Why do people who hold egalitarian attitudes still discriminate?

The justification-suppression model holds that people suppress prejudices they are motivated not to express, so a prejudice becomes a discriminatory act mainly where the situation supplies a justification that releases it, which is why discrimination hides behind non-group-based rationales (Crandall & Eshleman, 2003).

How do researchers measure discrimination?

The cleanest method is the field experiment: audit studies send matched testers, and correspondence studies send identical applications that differ only in a signalled group, so any gap in callbacks or offers measures discrimination directly, free of qualification differences (Bertrand & Mullainathan, 2004).

What are the main types of social discrimination?

MeSH classifies ageism, disability discrimination, homophobia, perceived discrimination, racism, sexism, and xenophobia as narrower kinds of social discrimination; these are indexing categories for retrieving literature, not mutually exclusive natural kinds, and a person may face several at once.

Does discrimination affect health?

Yes. A meta-analytic review found that perceived discrimination is reliably associated with poorer mental health and, more modestly, poorer physical health, acting largely through chronic stress and health-damaging coping responses (Pascoe & Smart Richman, 2009).

What reduces discrimination?

Intergroup contact has the strongest meta-analytic support, especially under equal-status, cooperative, goal-sharing, institutionally supported conditions, while interventions that retrain implicit associations produce effects that do not last (Pettigrew & Tropp, 2006; Lai et al., 2016).

Has discrimination declined over time?

Survey-measured attitudes have grown steadily more egalitarian, but a meta-analysis of hiring field experiments found no decline in discrimination against Black applicants across 25 years, showing that behavioral discrimination can persist even as stated attitudes improve (Quillian et al., 2017).

References