Contribution
It gives identification a structure. Earlier multicomponent measures disagreed on how many components there are and mixed items across them (Table 1). The authors specify five narrow components, each with two to four items, and group them under two higher-order dimensions taken from classic sociology: Durkheim’s mechanical solidarity and Tönnies’s Gemeinschaft (likeness) for self-definition, organic solidarity and Gesellschaft (chosen commitment) for self-investment. They show by confirmatory factor analysis that the structure holds across groups and cohorts, and that the components predict different intergroup reactions.
Key insight
Defining oneself as a group member and being invested in the group are different, and under threat they pull apart. “When confronted with an in-group’s misdeeds, it is the individuals’ self-definition at the group level that seems necessary to the self-critical feeling of group-based guilt” (p. 163), while “those who most invested themselves in their in-group identity most strongly defended this identity by legitimizing the in-group’s misdeed” (p. 163). The investment dimension is not plain affect: centrality, which has no evaluative content, loads on it as highly as satisfaction does (p. 162).
Assumptions
- Identification as the individual’s relation to the group as a whole, not interpersonal ties within it (p. 148).
- Self-stereotyping as depersonalization, following self-categorization theory: seeing oneself as similar to the average member (p. 146).
- Short scales: 14 items, 7-point agreement, two to four per component; six candidate items were dropped as too general (“I see myself as [In-group]”), double-barrelled or vague (note 1, Appendix).
- Bivariate prospective correlations carry Studies 5–7; regressions are used for discriminant validity.
Key results
- The components and items (Appendix).
- Solidarity: “I feel a bond with”, “I feel solidarity with”, “I feel committed to”.
- Satisfaction: glad, proud, pleasant, good feeling.
- Centrality: “I often think about the fact that I am”, “an important part of my identity”, “an important part of how I see myself”.
- Individual self-stereotyping: “I have a lot in common with the average [In-group] person”, “I am similar to the average …”.
- In-group homogeneity: members “have a lot in common with each other”, “are very similar to each other”.
- Studies 1–2: structure (pp. 149–154). First-year students at the
University of Amsterdam, native-born Dutch only: N = 464 (Dutch,
European, university identity), then a new cohort of 363 (Dutch,
European).
- The five-component, two-dimensional model fit all five measurements (CFI .966 to .983, SRMR .034 to .051). All item loadings exceeded .60.
- The two dimensions correlated .61 to .77 (Study 1) and .69 to .71 (Study 2).
- Alternatives: five components on one factor fit marginally with AIC about twice; centrality on self-definition (Model 1f) fit satisfactorily with AIC about twice; two components only, one factor, or cognitive versus affective fit poorly (CFI .57 to .83).
- A multigroup model constraining all parameters equal across the five measurements fit (CFI .94, RMSEA .03).
- Satisfaction was the most endorsed component, centrality the least.
- Studies 3–4: concurrent validity (pp. 154–157). Partial
correlations control satisfaction.
- Self-stereotyping had the highest partial correlation with Jackson’s depersonalization items (Dutch pr = .33).
- Homogeneity alone correlated with meta-contrast items (Dutch only).
- Centrality had the highest partial correlation with identity collective self-esteem, solidarity with membership self-esteem.
- The self–in-group overlap circles (Schubert and Otten 2002). Dutch: r = .20 with self-stereotyping, .10 homogeneity, .31 satisfaction, .29 solidarity, .17 centrality. European: .22, .13, .23, .25, .29. With satisfaction controlled, self-stereotyping’s correlation was marginal (Dutch pr = .10, European .09).
- Study 5 (N = 440, weeks after Study 1; Muslims and the West). In-group homogeneity alone predicted perceived differences between Dutch and Muslims (r = .13); centrality best predicted perceived threat to the in-group (.15), solidarity less (.10). Both held with established identification scales controlled.
- Study 6 (N = 36, months after Study 1, days after the Madrid bombing
of 11 March 2004).
- Centrality predicted support for banning headscarves (.34) and, at trend, threat measures (.29, .33).
- Holding the Netherlands responsible for provoking terrorism: in-group homogeneity .35, satisfaction −.50. In regression, satisfaction b = −.507, p = .003; homogeneity b = .225, p = .07.
- Solidarity (.37) and satisfaction (.37) predicted wanting Dutch Muslims to “integrate better”.
- Self-stereotyping predicted none of these.
- Study 7 (N = 49, one month after Study 2). A fabricated Human
Rights Watch report described Europe’s mistreatment of Rwandan asylum
seekers.
- Guilt was moderate (M = 2.92 on 0–5).
- Self-stereotyping as European was the only predictor of guilt (.33). It held with the four collective self-esteem subscales controlled (b = .244) and with Jackson’s subscales controlled (b = .271, p = .041). No component predicted shame, sympathy or being appalled.
- Satisfaction (.39) and centrality (.29) predicted legitimizing European asylum policy; self-stereotyping did not (−.20, ns). Satisfaction held with the subscales controlled (b = .373, p = .046).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | In-group identification has five components on two correlated higher-order dimensions, group-level self-definition and self-investment | moderate to strong for this population: confirmatory analyses over three groups and two cohorts, alternatives compared; one country, students | Studies 1–2 |
| C2 | The two dimensions are distinct but closely related | moderate: correlations of .61 to .77 between them | Studies 1–2 |
| C3 | The components predict different intergroup reactions independently of established identification scales | moderate: prospective correlations, small effects, some in very small samples | Studies 3–6 |
| C4 | Facing their group’s misdeeds, self-definition predicts group-based guilt and self-investment predicts defending the group | weak to moderate: one study, N = 49, one scenario; the Study 6 pattern is in line | Studies 6–7 |
| C5 | The investment dimension is not affect or evaluation, since centrality loads on it as highly as satisfaction | moderate: follows from the loadings and the poorer fit of a cognitive-versus-affective model | p. 162 |
Concepts
- group-level self-definition: defining oneself in terms of the group, by likeness: self-stereotyping and in-group homogeneity.
- group-level self-investment: positive feeling about, bond with, and importance of the membership: satisfaction, solidarity, centrality.
- individual self-stereotyping: seeing oneself as similar to the average or prototypical member.
- in-group homogeneity: seeing the members as alike, which makes the group a coherent entity distinct from out-groups.
- centrality: the chronic salience and subjective importance of a membership.
Connections
- Ashmore et al. 2004 (LIT-1795), unread. Cited as the review showing unitary identification measures inadequate, and for satisfaction and centrality. Leach et al. do not map their components onto its elements.
- Self-categorization theory (THEORY-1114). Self-stereotyping is its depersonalization, measured as a trait-like perception rather than a salience effect. The paper puts centrality, which the theory treats as part of identification, on the investment side.
- Brewer and Gardner 1996 (LIT-1757). Their relational self is said to resemble solidarity (p. 147).
- Smith 1993 (LIT-1722) and Smith and Mackie 2016 (read in this batch, NOTE-1262). Group-based emotion presumes identification; Leach et al. say it is self-definition that guilt needs, and investment that undercuts it.
- The group entity versus a category (TERM-096). The authors suggest identification with categorical “common identity” groups rests more on self-definition, and with “common bond” groups more on self-investment (p. 163). Their homogeneity component makes a group a “coherent social entity” by likeness, which is the category side of the term.
Bearing on the record
- The triviality objection (CLAIM-492). This paper answers part
of the question the objection asks. These are inferences; the authors
do not discuss egoism.
- Identification does split into a self-extent dimension and a caring dimension. Self-stereotyping (I am like the average member) is a criterion of the self’s extent fixed by perceived likeness, not by what one values. Satisfaction, solidarity and centrality are caring.
- They come apart, though not far. The dimensions correlate .61 to .77, but they predict opposite reactions to the group’s misdeeds: self-definition predicts guilt, investment predicts defence.
- Two of the record’s candidate criteria fall on the caring side. The overlap circles, the measure fusion descends from, correlated more with satisfaction and solidarity than with self-stereotyping for the Dutch, and with centrality most for Europe; self-stereotyping’s partial link was marginal. And “an important part of how I see myself”, the plainest self-extent wording, is centrality, which loads with caring. Situational salience by fit, the objection’s other example, is not measured.
- The owner’s thesis (CLAIM-491).
- If the larger self is fixed by self-definition, its signature is sharing the group’s guilt, not defending its image. An extended self in this sense takes the group’s misdeeds as partly one’s own. That is self-concern in the thesis’s sense, but self-critical.
- If it is fixed by investment, it defends the group’s image. Then self-interest on a larger self looks like protecting a valued identity, which fits the reading the fusion note gave (a self-concept to be kept, CLAIM-491’s note on fusion).
- No study measures helping or sacrifice, so which dimension carries prosocial action is open. Solidarity predicted wanting Muslims to integrate, an in-group-directed wish.
- A dated note should be added to CLAIM-492. Proposed text: “Leach et al. (NOTE-1262) split identification into group-level self-definition (seeing oneself as like the average member) and self-investment (satisfaction, solidarity, centrality). The two correlate .61 to .77 but act differently: faced with their group’s misdeeds, self-definition predicted guilt and investment predicted defending the group. Self-stereotyping is a criterion of the self’s extent that comes apart from caring. The overlap circles and the item ‘an important part of how I see myself’ go with investment, so they are weaker candidates than the claim supposed.”
- No THEORY filed.
- No ML instruction.
Limitations
- One population: Dutch university students, mostly women, ethnic Dutch only.
- Small samples for the key tests: N = 36 in Study 6, 49 in Study 7. Studies 5–7 report bivariate correlations, many marginal.
- Self-report throughout; no behaviour, and no helping.
- Two items per self-definition component.
- Deception in Study 7: the Human Rights Watch report was fabricated, adapted from a real one.
- Not the 2004 review. Ashmore et al.’s elements and measures are not laid out.
- Typos and inconsistencies.
- “participatns” (p. 150); “illiustrate” (p. 156); “its affect on individual feelings” for effect (p. 155); “Cindy Picket” for Pickett (p. 144).
- Study 7 reports t(49) and F(1, 48) with N = 49 (p. 161), and F(1, 48) for adding four subscales; the degrees of freedom cannot be right. Its t(49) = 2.14 is given p = .051.
- Study 5 gives F(2, 438) and F(2, 439) for parallel models with N = 440 (p. 158).
Open questions
- Which dimension predicts helping other members, or sacrifice for the group? A study pairing these scales with a costly helping measure would tell which kind of extended self the owner’s thesis needs.
- Does situational salience (self-categorization by fit) behave like self-stereotyping or like centrality?