NOTE-1253: Social Identity Theory and Self-categorization Theory: A Historical Review

Contribution

A review, not new work. It separates the two theories that are cited under one name and tells how each changed: which claims were central, which were dropped, and which critiques still stand. For this record its use is as a map. It says which of the fandom papers’ borrowed claims belong to social identity theory (SIT) and which to self-categorization theory (SCT), and which were revised.

Key insight

The two theories answer different questions with one shared mechanism, categorization. SIT, from the minimal group paradigm, explains why groups favour themselves: people want a positive self-concept, the group’s standing is part of it, and the group’s standing is judged by comparison with other groups. SCT drops SIT’s continuum from interpersonal to intergroup behaviour and asks when a person acts as a group member at all. Its answer is levels of self-categorization, salience by fit, and depersonalization. Several popular readings of SIT, such as “high identifiers are more biased” and “bias repairs low self-esteem”, are simplifications that the theory’s architects disowned.

Assumptions

A review; its assumptions are the approach’s.

• Groups are psychologically real only by comparison: “Groups are not islands” (p. 207).
• Motive: SIT assumes a desire for a positive and secure self-concept (p. 207). SCT “offered no explicit motivational analysis” (p. 215).
• Context: categories and their content are relative to a perceiver and a frame of reference (p. 208).

Key results

• SIT (pp. 205–207). It began from the minimal group paradigm (Tajfel, Billig, Bundy & Flament 1971). Participants assigned to meaningless groups favoured their in-group in allocating points they could not themselves receive, sometimes at the cost of absolute gain, to maximize the difference. Tajfel and Turner put interactions on a spectrum from purely interpersonal (rare) to purely intergroup. Categorization accentuates similarities within and differences between groups, and shifts the self-concept from personal identity towards social identity. Social identity is “those aspects of an individual’s self-image that derive from the social categories to which he/she belongs, as well as the emotional and evaluative consequences of this group membership” (p. 206). Members seek positive distinctiveness. Members of low-status groups use exit, flattering comparisons, chosen dimensions, revaluation or social change, depending on the perceived permeability, stability and legitimacy of the hierarchy. “For Tajfel, social identity theory was at its heart a theory of social change” (p. 207).
• SCT (pp. 207–209). After Tajfel’s death in 1982, Turner and colleagues elaborated the cognitive element and turned to intragroup processes (Turner et al. 1987). Three levels of self-categorization: human, social and personal, with finer gradations allowed at the middle. A functional antagonism between levels: as one becomes more salient, the others less. Salience is set by accessibility and fit; comparative fit is the meta-contrast ratio, and normative fit is consistency with stereotypical expectation. Depersonalization: under salience, self and others are seen as interchangeable exemplars of a context-dependent prototype that both describes and prescribes.
• Reinterpretations (pp. 209–211). Group polarization comes from the perceptual shift of categorization (Turner, Wetherell & Hogg 1989). Cohesion is depersonalized liking based on prototypicality, not interpersonal attraction (Hogg & Hardie 1991). Stereotype content shifts with the comparison set (Haslam et al. 1992), and stereotypes have social functions rather than being the shortcuts of a cognitive miser. Deindividuation is a shift from personal to social identity, so anonymity makes people more responsive to the local norm, prosocial or antisocial (Reicher, Spears & Postmes 1995), with meta-analytic support (Postmes & Spears 1998). Crowds come with shared norms, and their violence is often aimed at symbolic intergroup targets (Reicher 1987). The normative/informational distinction in influence is dissolved (Turner 1991). Leaders influence to the extent they are prototypical (Hogg 2001). Turner (2005) reverses the usual order of power: prototypicality → influence → power → control of resources (p. 211).
• Evidence and its limits (pp. 211–213).
  • Minimal-group bias survived demand and similarity explanations.
  • Rabbie’s interdependence critique was the strongest challenge. Participants favour the out-group when they believe it controls their outcomes (Rabbie, Schot & Visser 1989). Bias disappears when outcomes are explicitly independent (Karp et al. 1993). Many now read the behavioural interaction model as a finer analysis within SIT, not a rival.
  • In the 1990s the research moved from minimal to real groups and, by Hornsey’s account, consolidated the socio-structural hypotheses.
  • SCT tests manipulated salience or measured identification. High salience or identification increased self-stereotyping and perceived in-group homogeneity, and prototypicality affected influence only then.
• Dead ends (pp. 213–215).
  • The weak correlation between identification and bias was read as against SIT (Hinkle & Brown 1990). Turner (1999) denied the theory ever predicted it, and “the field appears to have accepted” that it is not diagnostic.
  • Hogg and Abrams’s (1990) two self-esteem hypotheses got more meta-analytic support for “bias raises self-esteem” than for “low self-esteem causes bias” (Rubin & Hewstone 1998). The approach was criticized for measuring personal self-esteem to predict a collective outcome. Turner distanced SIT from it.
• Later directions (pp. 215–217).
  • Distinctiveness and uncertainty reduction (Hogg 2000) replaced self-esteem as the most studied motive.
  • A “subtle withdrawal” from functional antagonism, and attention to individuals shaping norms (Hornsey 2006; Postmes & Jetten 2006).
  • Group-based emotions (Smith 1993), collective guilt and intergroup schadenfreude.
  • The human level of identity first studied (Wohl & Branscombe 2005).
  • Organizational psychology.
• Critiques, not answered (p. 217).
  • Functional antagonism and salience look rigid beside subgroup, cross-cutting and relational identities (Abrams & Hogg 2004).
  • Depersonalization obscures tolerance of dissent.
  • It is too broad to be falsifiable (Hogg & Williams 2000).
  • It is itself reductionist (Farr 1996).
  • It explains in-group favouritism better than out-group hostility (Brewer 1979; Brown 1995).
• Growth. PsycInfo publications naming either theory: 11 in 1986, 47 in 1996, 568 in 2006 (endnote 2).

Claims

id	claim	strength	support
C1	SIT and SCT are distinct theories sharing assumptions: SIT is about intergroup relations and status, SCT about when and how people self-categorize, and about intragroup processes	strong as history: the original sources are cited and characterized consistently with Turner et al. 1994	pp. 205–209
C2	Minimal-group in-group bias is robust to demand and similarity explanations, but disappears when participants are told their outcomes are independent of others’ allocations	moderate: reported from cited studies; the Karp et al. result is a single study in a Japanese journal	pp. 211–212
C3	Identification does not reliably correlate with in-group bias, and SIT, read carefully, does not predict that it should	moderate as history; the defence is Turner’s own (1999)	pp. 213–214
C4	Meta-analysis favours “showing bias raises self-esteem” over “low self-esteem causes bias”	moderate: one cited meta-analysis (Rubin & Hewstone 1998)	p. 214
C5	The approach has drifted from a strict functional antagonism between personal and social identity	weak to moderate: Hornsey’s reading of the literature, his own work among it	p. 216
C6	Anonymity raises responsiveness to the salient group norm rather than releasing antisocial behaviour as such	moderate: one cited meta-analysis (Postmes & Spears 1998)	p. 210

Concepts

• social identity: the part of self-image from category memberships, with its emotional and evaluative consequences (p. 206). In SCT, the intermediate level of self-categorization.
• positive distinctiveness: favourable difference between in-group and relevant out-groups, which members are motivated to achieve or keep (p. 207).
• socio-structural variables: perceived status, permeability, stability and legitimacy of an intergroup hierarchy (pp. 207, 212).
• functional antagonism: as one level of self-categorization becomes more salient, the others become less so (p. 208).
• meta-contrast ratio, normative fit, depersonalization, prototype: as in Turner et al. 1994 (NOTE-1284). Hornsey adds that the prototype is prescriptive as well as descriptive (p. 209).
• social identity approach: SIT and SCT together.

Connections

• Turner et al. 1994 (NOTE-1284). The two agree on SCT’s core. Hornsey reports that functional antagonism was later softened, while the 1994 paper still describes a “continual competition” between personal and group levels (Turner et al. 1994, p. 456). The two are not in conflict, because a competition can be graded.
• The record’s fandom papers. Čater et al. 2026 (LIT-1138) and Berendt & Uhrich 2016 (LIT-1147) are framed by SIT’s distinctiveness and collective self-esteem. Wegner et al. 2020 (LIT-1140) concerns identity threat. Biel 2026 (LIT-1133) concerns recategorization into a common in-group. All are unread. Per this review, “collective self-esteem” findings belong to SIT’s contested self-esteem strand (C4), not to SCT.
• Pettigrew 1998 (LIT-1408), unread: its recategorization strategy and Brown & Hewstone’s contact theory, which Hornsey cites (p. 215), are applications of the approach.
• Collective emotion. Hornsey names Smith (1993) as the origin of social-identity work on group-based emotions (p. 216). von Scheve and Ismer use the same source (NOTE-552, THEORY-126).

Bearing on the record

• Co-source, with Turner et al. 1994, of THEORY-1114. Hornsey supplies the review-level summary of evidence (C2–C6) and the critiques that make the account Proposed.
• CASE-104. SIT’s low-status strategies and legitimacy variables give a second lens on fans who stay loyal while judging the club: a controversy threatens the group’s positive distinctiveness, and members may respond by revaluing dimensions rather than leaving. That is an inference, not a claim the review makes. It is one LIT-1138 tests.
• The social identity approach has no topic of its own. social-science holds it. Whether group processes deserve a topic word is a decision for later, once more than two works are read.
• No ML instruction.

Limitations

• A review by an insider. Hornsey works within the approach and says he will “not seek to defend” it against its critics (p. 217). The balance of evidence is his summary, with few effect sizes.
• Brief. About 14 pages of text for some 35 years; each strand gets a paragraph.
• Dated 2008. It predates the replication debates in social psychology, so none of the cited effects has been checked against them here.

Open questions

• Has minimal-group bias held up in large preregistered replications, and does the Karp et al. independence result replicate? A replication study read in full would settle it for the record.
• Is the approach falsifiable as a meta-theory (p. 217)? The record could hold one of its specific predictions to a test, as THEORY-1114 tries to.