Intergroup Biases: A Focus on Stereotype Content

Abstract

Impressions of others, including societal groups, systematically array along two dimensions, warmth (trustworthiness/friendliness) and competence. Social structures of competition and status respectively predict these usually orthogonal dimensions. Prejudiced emotions (pride, pity, contempt, and envy) target each quadrant, and distinct discriminatory behavioral tendencies result. The Stereotype Content Model (SCM) patterns generalize across time (2oth century), culture (every populated continent), level of analysis (targets from individuals to subtypes to groups to nations), and measures (from neural to self-report to societal indicators). Future directions include individual differences in endorsement of these cultural stereotypes and how perceivers view combinations across the SCM space.

The earliest social psychology of stereotypes documented their content (1, and then replicated and extended by 2–4). With few exceptions, the rest of the 20^th century focused on processes of stereotyping (e.g., social categorization, 5–6). At the outset of the 21^st century, the Stereotype Content Model identified two systematic dimensions of stereotyping (7; see Figure 1): warmth and competence.

Figure 1.

Stereotype Content Model, typical outgroup locations.

Precedents for these two dimensions include decades of impression formation research (see 7–8, for reviews), especially Asch’s (9) foundational research using a competent person who was either warm or cold and Abele and Wojciszke’s (e.g., 10–11) more modern identification of communality/morality (warmth) and agency/competence as two orthogonal dimensions, accounting for as much as 80% of the variance in impressions.

The distinctive SCM contribution, identifying mixed stereotypes high on one dimension but low on the other, also has precedents and parallels: ambivalent sexism (dumb-but-nice vs. competent but cold; 12), doddering-but-dear old-age stereotypes (13–14), smart-but-not-social anti-Asian stereotypes (15).

Overview

The Stereotype Content Model (SCM) is a simple framework (BIAS Map: 16; SCM: 7, 8, 17):

Social Structure→Stereotypes→Emotional Prejudices→Discriminatory Tendencies Stereotypes

This overview starts with the warmth × competence stereotype space. Early work (7, 17) hypothesized and found that (a) Perceived competence and warmth differentiate group stereotypes; and (b) Many stereotypes include mixed ascriptions of competence and warmth. Generally replications support these findings in more recent American convenience samples (2, 18) and in representative samples (16).

Warmth reflects the other’s intent, so it is primary and arguably judged faster (19). Competence reflects the others ability to enact that intent, so it is secondary and judged more slowly. The most valid traits reflecting warmth include seeming trustworthy and friendly, plus sociable and well intentioned. Competence includes seeming capable and skilled. Moreover, validity also increases because the four warmth-by-competence clusters also differ on the other hypothesized variables: perceived social structure, emotional prejudices, and discriminatory behavioral tendencies.

Social Structure

Given evidence of the warmth-by-competence space, SCM research has tested for their respective antecedents: (a) Status predicts perceived competence, while (b) interdependence (competition/cooperation) predicts stereotypic warmth. The status-competence correlations are surprisingly robust, usually over r = .80, and generalizing across cultures (average r = .90, range = .74 – .99, all p’s < .001; 20). Status is measured as economic success and prestigious job, so evidently the belief in meritocracy is widespread. The status-competence correlation persists across stable and unstable status systems (21).

The cooperation-warmth (and competition-cold) correlations have been more uneven until lately. In early data, perceived competition did correlate negatively with perceived warmth, r = −11. – .68), consistent but small effects (averaging −.32), sometimes not significant (20). Closer examination has refined these predictions (18). Warmth most appropriately includes both sociability and trustworthiness/morality, as in the earliest SCM studies, and consistently with the close relationship between trustworthiness and friendliness. Competition predicts most robustly when it includes not only economic resources but also values.

Emotional Prejudices

Whereas the preceding hypotheses—structure (interdependence, status) → stereotype (warmth, competence)—predict main effects, the stereotype → emotional prejudice hypotheses predict interactions. Each quadrant’s warmth-by-competence combination predicts distinctive emotions:

• High warmth, high competence, the combination that includes the society’s prototypic ingroups, such as the middle class, elicits pride and admiration.

• Low warmth, low competence, the quadrant that contains societal outcasts, such as homeless people, elicits contempt and disgust.

• Low warmth, but high competence, the mixed combination that includes successful outsiders, such as rich people, elicits envy and jealousy.

• High warmth, but low competence, the mixed quadrant includes benign subordinates, such as old or disabled people, elicits pity and sympathy.

The predictions derive from social theories of emotion, and a variety of SCM studies confirm them (7, 16). Moreover, individual groups located in each quadrant provide case studies of specific emotional dynamics of (e.g.) disgust or envy (see below).

Discriminatory Behavioral Tendencies (the BIAS Map)

The Behavior from Intergroup Affect and Stereotypes (BIAS) Map extends the SCM to distinctive discriminatory tendencies (16). Predictions from stereotype dimensions are main effects. Because the warmth dimension is primary, it predicts active reactions, both positive (high warmth predicts helping and protecting) and negative (low warmth predicts attacking and fighting). Because the competence dimension is secondary, it predicts more passive reactions, both positive (associating) and negative (neglecting).

The behavioral combinations, as reported by participants, are informative about varieties of discrimination. The high-high pride groups of course elicit both helping and associating. The low-low groups elicit both active harm and passive neglect, behavior characteristically directed toward homeless people.

The mixture of passive association and active harm describes reactions toward outsider entrepreneurs, whose businesses the majority may patronize in peace and stability, but the envied are also the targets of mass attacks under social breakdown. The mixture of active help but passive neglect describes institutionalizing pitied outgroups.

Between intergroup stereotypes and affect, the emotional prejudices more strongly and immediately predict behavior (16; see also 22 for a meta-analysis regarding racial biases).

Validity

Convergent and Divergent Validity: Overlap and Distinctiveness

Several parallel models are nonetheless distinct from the SCM. One comprehensive model of generic attitudinal dimensions, the Semantic Differential, identifies evaluation, potency, and activity as key (23). In social cognition, the last two dimensions collapse together, so one might assume that evaluation-by-potency/activity would be redundant with warmth-by-competence. However, these dimension operate at 45-degree angle to the SCM space (24). Evaluation runs from the low-low quadrant to the high-high quadrant (being high on either warmth or competence is good). Potency/activity runs from the unthreatening warm–but-incompetent quadrant to the threatening cold-but-competent quadrant.

In individual person perception, other relevant parallels include social good-bad and task good-bad (25), morality and competence (26), trustworthiness and dominance (19). In intergroup and inter-nation relations, similar dimensions emerge (respectively, 27, 28; see 29, 30, for discussions of the distinctions).

External Validity: Generalizing over Time

The SCM space is not simply a product of modern, immigrant-receiving liberal democracies leaning toward the politically correct notion that most outgroups have at least one positive side. The SCM dimensions emerge in the oldest studies of stereotypes (1), replicated across the 20^th century and re-analyzed in the 21^st (2). Americans and British stereotypes persist as high on both dimensions; Turks persist as low on both. Asians remain competent but cold. And “Negroes”/African Americans appear warm but incompetent. Another older dataset features Fascist magazine writing about social groups, articles content analyzed and recreating the SCM space (31).

Generalizing to Subtypes

Most SCM studies address stereotypes of societal groups, after asking pilot participants to identify the relevant ones. Subtypes reproduce the SCM space, differentiating by warm and competence different kinds of women and men (32), gay men (33), lesbians (34), African Americans (16, 35), immigrants (36), Muslims (37), and Native Americans (38). Subtypes of mental illness also spread out across the warmth-by-competence space (39).

Generalizing to Specific Groups

Drilling down to stereotypes of specific group, the SCM predicts the patronizing attitudes and behavior toward working mothers (40), sexualized women (41), and older people (13). The SCM also predicts the devaluing of lower-status people generally (42).

The SCM likewise describes patterns of resentment toward (enviable) Asians in the U.S. (15), Americans in other countries (43, 44), and politicians everywhere (45). Other enviable outgroup members (rich people, business people) specifically elicit envy-consistent Schadenfreude (malicious glee at their misfortunes), measured by self report, smile-muscle activation, and neural reward centers (46, 47).

Disgusting outgroup members (homeless, drug addicted) elicit not only reports of disgust, but also insula activation consistent with those reports (48). Moreover, participants report dehumanizing responses, such as being disinclined to imagine the minds and experiences of such individuals, and the neural areas (e.g., mPFC) implicated in such judgments likewise fail to activate.

Generalizing to Individuals

As noted, warmth and competence apply to individual person perception, as well as group perception. In a specific test of SCM predictions at the individual level, higher status individuals were expected to be competent, and competitive individuals were expected to be not warm (49). The status expectations cause higher status people to compensate for their apparent lack of warmth and lower-status individuals to compensate for their apparent lack of competence (50). Cross-racial interactions parallel these status effects (51, 52).

Generalizing to Other Intent-having Entities

If the primary warmth judgment reflects perceived intent for good or ill, and perceived competence reflects the capacity to enact that intent, then other entities seen as possessing intent should fit the warmth-by-competence space. Indeed, animals spread out across the space, with dog and cats appearing high on both, and vermin appearing low on both. Predators are competent but cold and edible farm animals are nice but stupid (53).

Organizations of people, such as corporations, also appear to have intent, and the public responds according with trust only for apparently well-intentioned brands and respect only for competent ones that deliver (54, 55).

Causality among the SCM Variables

Most SCM studies are descriptive and correlational, so the structure-stereotype-prejudice-behavior sequence rests on correlations consistent with predictions (7, 16). Experiments directly manipulating interdependence and status structures for two hypothetical groups do yield the predicted patterns of warmth and competence stereotypes (56). As noted, manipulated interdependence and status between two individuals show the same effects on perceived warmth and competence traits (49). Similarly, manipulating status alone (by housing price) predicts the inhabitants’ expected competence (57).

Likewise, manipulating apparent warmth and competence in vignette studies leads to the predicted emotions (56). These emotions mediate the link between stereotypes and behavior (16).

Dynamics between the Dimensions: Compensation and Innuendo

Warmth and competence themselves often correlate negatively, contrary to halo-effect predictions (58, 59), especially in comparative contexts (60, 61), and regardless of direct or indirect measurement (62; see 63, for a review).

Lay people understand and use these tradeoffs in communicating stereotypes. They will mention the positive dimension and not mention the negative one, knowing that innuendo will imply it, a phenomenon dubbed stereotyping by omission (2), which allows stereotypes to stagnate over time. Listeners understand the innuendo (64), and impression-managers likewise use it, downplaying one dimension to emphasize the other (65).

Moderators

Individual and Group Moderators

Although not much tested, some individual difference variables moderate how much people endorse the SCM model. Status-justifying ideologies reinforce the status-competence correlation (57).

Group-level moderators include group membership. Slight ingroup favoritism emerges for students rating students, across nations, and for countries rating themselves, within the EU (29). Groups especially favor themselves on their stronger dimension, higher status groups on competence and lower-status groups on warmth (66); strength of group identification affects interpretation of outgroup behavior on SCM dimensions (67).

Individual and group-level moderators of the warmth dimension and its correlation with interdependence have been even less evident. Cooperative or competitive orientations are likely candidates. Morality might also be relevant, because warmth includes morality and trustworthiness. Suggestive support comes from preliminary work on system legitimacy (21).

Cultural and Macro Moderators

Cultural differences emerged immediately, as East Asian samples demoted societal ingroups and reference groups to the middle of SCM space, consistent with cultural modesty norms (68).

More broadly, the central feature of the SCM space is the warmth-by-competence differentiation of groups, in which the two dimensions are roughly orthogonal. To the extent the two dimensions correlate, they boil down to a single vector of evaluation. Cross-cultural samples from every populated continent indicate that the SCM space does differentiate groups, but that the warmth-competence correlation varies (20, 68). Countries with more income inequality show less warmth-competence correlation, indicating that they use the ambivalent (mixed) quadrants; these and related data suggest justifying inequality (some high-status groups are allegedly nice and some not; some low-status groups are allegedly deserving and some not). Under income equality, most groups locate in the acceptably medium to high-high space and qualify for social benefits; the extreme low-low outgroups (homeless, nomadic, migrant) do not.

Another macro dimension that apparently affects use of the warmth-competence space is conflict (43). Higher-conflict countries adopt more of an us-them cultural map, minimizing use of the ambivalent parts of the space. (See 69 for a more detailed review of cultural patterns.)

Conclusion

For the past 15 years, the stereotype content model has accumulated evidence that warmth and competence differentiate social groups in more that three dozen countries, over time, and at levels of analysis that include subtypes and individuals. Perceived social structures of competition and status predict the two dimensions, which together predict distinct emotional prejudices and discriminatory tendencies. Moderators appear at the individual, group, cultural, and macro level, but many of the patterns are consistent: Citizens and the middle class are admired as high on both dimensions; unhoused people such as migrants, homeless, and nomads disgust as low on both. Older and disabled people are pitied as well intentioned but incompetent. Rich and business people are envied as competent but cold. These patterns occur in self-reports and neural signature.

Ongoing work addresses specific groups’ profiles (e.g., older people: 14, 70). Future work could address combinations of groups across the space. Also, individual differences in endorsing the SCM space might be of interest, as would be moderators of its use.

Highlights.

• Two fundamental dimensions -- warmth and competence -- explain stereotyping, as the Stereotype Content Model shows.

• The model identifies ambivalent stereotypes of respect but not liking or of liking but not respect.

• Perceived social structure (competition, status) predicts these stereotypes, which in turn predict distinct emotional prejudices (envy, disgust, pity, pride), which then predict distinct discriminatory tendencies.

• SCM principles generalize across dozens of cultures, decades of time, and levels from neural response to social interaction to cultural ideology.