NOTE-1268: Social Influence and Group Identity

Contribution

A review that puts group identity at the centre of social influence. It sets the self-categorization account against three rivals: normative influence (complying for approval), interdependence and network approaches, and persuasion research that treats the group as a source of bias. It rereads a run of classic findings in identity terms. It adds a section relating group norms to injunctive norms, morality and intrinsic motivation, which the earlier statements of the account lacked. It ends with the SIDE model’s online findings.

Key insight

If the group is part of the self, being influenced by it is not giving way to an outside pressure. “The fundamental insight of the self-categorization approach to group influence is that the group is not only an external entity but (for high identifiers, at least) becomes internalized as a level of self. Thus, social influence is partly self-influence, saved from circularity by the fact that other in-group members form its source” (p. 373). The same idea explains the SIDE paradox. Because the group is “a dimension of the self”, isolation and anonymity need not weaken it (p. 382).

Assumptions

• Self-categorization theory’s levels of self. Personal and group identities are both selves, up to the human level (p. 369). Salience depends on accessibility and fit, comparative and normative (p. 373).
• Identification as the test of process. Group identification is “the litmus test of whether group identity is the driving process” (p. 380). An effect it does not moderate is read as normative influence instead.
• Categorization over similarity. Categorization is taken to underlie inferences of similarity, after Murphy & Medin 1985 (p. 372).

Key results

The review reports results and presents none. Each is cited to the literature, much of it the author’s own.

• Three realms of influence (Figure 1, p. 370). Two axes: level of self (personal or group) and attentional focus (self or other).
  • The personal realm: self-interest, personal goals, the need to fit in. Normative influence, interdependence and networks belong here.
  • The group realm: group identity, in-group norms, group interests.
  • The other realm: focus on the task or the information, with “few concerns about creating or protecting the self” (p. 370). This is informational influence.
• Normative influence is instrumental (p. 371). Deutsch and Gerard’s normative influence is “going along to get along”, closer to power than influence. “A theoretical shift was needed to see the group as part of the self” before group influence could count as private acceptance.
• Referent informational influence (RII) (pp. 372–373). Turner’s name for group influence that is informational: members trust in-group members to know the answer. Physical and social reality testing are blurred. Asch’s participants can be genuinely perplexed, enough to affect private acceptance.
• Fit decides which identity carries influence (p. 373). Lea et al. (2007): four-person groups, two British and two Dutch, two men and two women, one of each sex in each nationality. On topics pre-tested to divide the nationalities (“British food is bad”) the nationalities bipolarized toward their norms. On topics dividing the sexes (“there should be more sport on TV”) the sexes did. That is comparative and normative fit at once.
• Classic findings reread (pp. 374–376).
  • Shared identity strengthens convergence in Sherif’s autokinetic paradigm (Abrams et al. 1990) and conformity in Asch’s.
  • Milgram’s participants identified with the scientific project, on Haslam and Reicher’s reappraisal.
  • In emergencies people “often conform to prosocial norms even if it involves self-sacrificing behavior” (Drury and colleagues; p. 375).
  • Bystanders help others in peril “primarily when they share a social identity with them” (Levine & Crowther 2008; p. 375).
  • Emotional contagion and mimicry stop or reverse at out-group boundaries (van der Schalk et al. 2011; Spears et al. 2004).
  • Moderate feminists were influenced by radical feminists cast as part of “the feminist sisterhood”, not when cast as an out-group (David & Turner 1999; p. 376).
• Crowds (p. 375). Deindividuation theory, after Le Bon, held that crowd members lose individual rationality and normal regulation. Reicher’s analysis says that “rather than losing one’s self or identity in the mass, crowd members actually switch to a group identity”. The meta-analysis of Postmes & Spears (1998) “showed that the classic conditions associated with deindividuation (i.e., group immersion and anonymity in the group) actually enhanced depersonalization and thus conformity to these local group norms”. Local group norms must be kept apart from the general societal norms that make crowd behaviour look antinormative.
• Norms and morality (pp. 376–378).
  • Cialdini et al.’s (1990) descriptive/injunctive distinction mentions “group” twice and identity not at all.
  • Group norms predicted recycling among high identifiers; social injunctive norms did not (White et al. 2009).
  • “Group identities can become so internalized (for high identifiers) that they become self-perceived as personal” (p. 377).
  • Raz’s social self-authorship and Amiot and Thomas’s work are cited for group-motivated behaviour being as intrinsically motivated as individual behaviour (p. 377).
  • In-group bias and fairness are injunctive norms at two levels of self-categorization. Right-wingers showed more in-group bias and endorsed the superordinate fairness norm less (Iacoviello & Spears 2018; p. 378).
• Moderators (pp. 379–381).
  • The prototype is the position that maximizes similarity within and difference without (meta-contrast). Group polarization is explained by an implicit out-group at the other pole: “a group effectively becomes salient and operative only in contrast to a (relevant) out-group” (p. 379).
  • A meta-analysis of 20 studies found more prototypical leaders more influential (Barreto & Hogg 2017; p. 379).
  • Peripheral high identifiers may overconform to prove their worth.
  • High identifiers may reject norms they see as harming the group (Packer’s normative conflict model), especially when secure in the group (“loyal deviance”; Masson & Fritsche 2019). “Interest flows from a focus on group identity as well as on the individual, especially for high identifiers” (p. 380).
  • Threat to personal control raised conformity to aggressive or nonviolent norms, whichever the group held, but identification did not moderate it, so it is read as normative influence (Leander et al. 2020; p. 380).
  • Group-level threat raises self-stereotyping, perceived homogeneity and loyalty (pp. 380–381).
  • Reactance follows the dominant realm of self. People with independent self-construals resisted in-group sources more; people with interdependent ones resisted out-group sources more (Graupmann et al. 2012; p. 381).
• Online influence and SIDE (pp. 382–384).
  • Cognitive SIDE effects concern the salience of identity; strategic ones concern accountability to in-group and out-group audiences.
  • Depersonalization is “not to be confused with the loss of self implied by deindividuation theories” (p. 382).
  • Spears et al. (1990): anonymous groups polarized toward the group norm when group identity was salient and away from it when personal identity was (p. 382).
  • When identifiable, people using text chat were swayed by argument strength. When anonymous, the group norm mattered more than argument quality (Lee 2008; p. 382).
  • Group anonymity raises salience, identification, cohesion, trust, knowledge sharing and “the tendency to perceive the collaborative group or dyad as an entity” (p. 383).
  • Visually cued categories reverse the pattern. Gender-based influence rose under webcam visibility while nationality-based influence rose under anonymity (Lea et al. 2007; p. 383).
• Conclusion (p. 384). Group identity is “not some product of its individual building blocks, but the primary foundation for much of our being and behavior”. “It is when we strip the individuals away that we get to commune with our group nature, perhaps in its purest form.”

Claims

id	claim	strength	support
C1	Group influence can be private acceptance, not only compliance, when the group is a level of the self (referent informational influence)	moderate as reported: classic paradigms rerun with shared identity, cited without data	pp. 372–374
C2	Which identity carries influence depends on the comparative and normative fit of the topic	moderate as reported: one study described (Lea et al. 2007)	p. 373
C3	More prototypical members and leaders are more influential	moderate: one cited meta-analysis of 20 studies	p. 379
C4	Group immersion and anonymity produce depersonalization and conformity to local norms, not loss of self	moderate: one cited meta-analysis, by the author and a colleague, reported without numbers	p. 375
C5	Helping, mimicry and emotional contagion are bounded by shared identity	moderate as reported: several studies cited; none presented	p. 375
C6	Isolation and visual anonymity strengthen group influence, except for visually cued categories	moderate to strong as reported: many replications across labs, listed	pp. 382–383
C7	Group-based norms and motives can be intrinsic and moral, not only external pressure	weak: argued from philosophy (Raz) and two research programmes, with one thesis cited	pp. 376–378

Concepts

• referent informational influence: influence from in-group members accepted as valid information because they share one’s self-category.
• normative influence: conformity to others’ expectations to gain approval or avoid rejection; instrumental compliance.
• three realms: personal self, group self and other/outward focus, as bases of influence (Figure 1).
• descriptive and injunctive norms: what others do, and what they approve or prescribe (Cialdini et al. 1990).
• loyal deviance: a secure, highly identified member departing from a group norm for the group’s sake.
• cognitive and strategic SIDE effects: effects of anonymity on which identity is salient, and on what one can express given who can see and sanction it.
• depersonalization: seeing self and others as similar and interchangeable group members; explicitly not loss of self (p. 382).

Connections

• Turner et al. 1994 (NOTE-1284). The 1994 paper gives influence one paragraph: “private category change” (p. 461). This review is the full development of that paragraph, with the evidence named. It adds the three realms, the norms-and-morality section and the online work.
• Hornsey 2008 (NOTE-1253). Hornsey gives SIDE and Postmes & Spears 1998 one sentence each. This review gives the 1998 meta-analysis a paragraph and SIDE a section, and adds the emergency, helping and contagion findings that Hornsey does not mention.
• Skorich (NOTE-1295, THEORY-1112). Two points bear on it. The polarization account posits an implicit out-group at the other end of the scale (p. 379). And anonymity is reported to increase perceiving the group or dyad “as an entity” (p. 383).
• Spears & Postmes 2015 (read alongside). The review points there for “a more complete review” of SIDE (p. 382). The two overlap on SIDE; this one covers far more of influence.
• Identity fusion (NOTE-1256). The review does not mention fusion. Its remark that internalized group identities become “self-perceived as personal” (p. 377) describes something close to it.

Bearing on the record

• The owner’s thesis (CLAIM-491). The review bears on it in four ways. These are inferences; Spears does not discuss selflessness.

  • It applies the thesis’s form to belief. Being influenced by the group is “partly self-influence” (p. 373). The thesis says that acting for the group is acting for a self. The review says that being persuaded by the group is being persuaded by a self.
  • It names a realm the thesis has to place: the “realm of the other”, with no self-focus and “few concerns about creating or protecting the self” (p. 370). It concerns attending to information, not concern for another’s good, and Spears says Turner would have denied that there is any “pure information without a social (or group) dimension” (p. 384). Still, it is a self-categorization theorist’s own category for activity with no self in play. The thesis should say whether it covers motive at all.
  • It separates a group’s interest from its norm. High identifiers can reject a norm that harms the group, and “interest flows from a focus on group identity” (p. 380). That supports reading the thesis as pursuit of a larger self’s interest rather than as conformity to it.
  • It leaves “larger or changed?” open. On one side, internalized group identities can come to feel personal (p. 377), which is close to fusion. On the other, the review calls the group self “the primary foundation” and says stripping the individuals away gets us to “our group nature” (p. 384). That is a different self, not an extended one. Either way the crowd member is not selfless, so the thesis’s denial of selflessness holds while its word “larger” is not supported.

• Batson’s objection (CLAIM-494). Levine’s bystanders help “primarily” when they share an identity with the victim (p. 375). Drury’s emergency survivors sacrifice under a shared identity. Both fit the thesis. “Primarily” also leaves helping without shared identity, which is where Batson’s case lives. Neither study is read here.

• The triviality objection (CLAIM-492). The field’s test for an identity process is moderation by identification (p. 380). Identification scales include caring about the group, so this test does not give the independent criterion the objection asks for. The salience manipulations reported in Spears & Postmes 2015 come closer.

• Self-categorization theory (THEORY-1114).

  • Lea et al. (2007) fits the promote_when’s shape. It holds the people fixed, varies the topic, and measures which category carries influence. But it comes from the author’s group and is reported without data, so it does not promote the account. It is worth filing as a LIT.
  • The implicit out-group device bears on the THEORY’s test against Skorich: “an experiment that removes the comparison group”. If the theory can always posit a contrast category at the other end of a scale, removing the visible out-group does not remove comparison. The test needs a design that rules out an implicit contrast as well.

• Group entity versus category (TERM-096). Anonymity is reported to make people see the group or dyad as an entity (p. 383). On the term’s reading, accentuating likeness yields interchangeability, not unity. If the report holds, depersonalization can yield perceived unity. This comes at second hand, and “entity” may have been measured as entitativity, which the term sets aside.

• CASE-104 and QUESTION-119. Prototypical members carry more influence, and anonymity strengthens local norms. Together they suggest that an online fanbase’s norm can be set by its most prototypical voices without any procedure. That is a derivation, not a finding about fans.

• A dated note is added to CLAIM-491. Proposed text, to be combined with the note proposed from Spears & Postmes 2015:

  Spears (2021) says the self-categorization account treats group influence as “partly self-influence” (p. 373), and treats high identifiers who defy a harmful group norm as serving the group’s interest (p. 380). Both fit the thesis read as a group self’s interest, not as conformity. But the review also names a “realm of the other” in which no self is in play (p. 370), and it treats the group self as foundational rather than as an extension of the individual (p. 384). So the thesis should say whether it is a claim about motives only, and whether its larger self is enlarged or replaced.

• No THEORY filed. The influence account would be a separate THEORY; the record has read none of its primary studies.

• No ML instruction.

Limitations

• A review by an insider. Spears is an author of the SIDE model and of several of the studies cited. The balance of evidence is his.
• No data. One count (20 studies in Barreto & Hogg’s meta-analysis) and no effect sizes. Postmes & Spears 1998 is described without its number of studies or effect sizes.
• The primary fit and frame studies are absent. Oakes, Turner & Haslam 1991 and Haslam et al. 1992 are not cited. Fit is illustrated only with Lea et al. 2007.
• No word on replication. None of the cited effects is checked against the replication debates.
• Typos.
  • “This is not to deny the absence of motivation” (p. 370) stands for its presence.
  • References: “Tavsitock” (Cartwright & Zander); “identifcaion” and “three process” (Kelman 1958); “ougroups” (Noel et al. 1995); “Jannssen” (Van Baaren et al. 2009); “Baumeister RB” for RF.
  • Mackie et al. 1990 is given as volume 57 of JPSP, which is the 1989 volume.
  • Van Zomeren et al. 2008 is given as Psychol. Bull. volume 34; the 1998 meta-analysis is in volume 123, so 2008 cannot be 34.
  • Spears & Postmes 2015 is listed in “The Handbook of Psychology of Communication”; the book is The Handbook of the Psychology of Communication Technology.

Open questions

• Is there motivated action in the “realm of the other”, with no self in play? If so, it is the case CLAIM-494 needs. If the realm covers only information processing, the thesis is untouched.
• Does referent informational influence survive preregistered tests? A replication of Abrams et al. 1990 or of the in-group Asch effect would answer it for the record.