Group identification and perceived self-group similarity: differentiating projection from introjection

ABSTRACT

Two studies demonstrate a technique that can clarify to what degree individual differences in perceived self-ingroup similarity reflect differences in projecting the self onto the group (self-anchoring) and/or introjecting the group onto the self (self-stereotyping). In preregistered Study 1 undergraduates described their values and those of fellow students. In Study 2 (a reanalysis of Denning & Hodges, 2022) citizens described their personalities and those of compatriots with similar voting preferences. Across both studies, ingroup identification predicted perceived self-ingroup similarity. Decomposing each participant’s self-ratings and ingroup-stereotype-ratings into normative (average) and distinctive (non-normative) profiles suggested this was primarily attributable to projecting the self-concept onto the group, but in Study 2's intergroup conflict situation perhaps also attributable to ingroup enhancement and introjecting ingroup-stereotypes.

KEYWORDS:

• Perceived similarity
• assumed similarity
• social projection
• identification
• normative profiles

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15298868.2023.2244723

Notes

1. For a general mathematical proof that the sum of covariances between profile components equals the covariance between raw profiles, see the section entitled “Covariance of Linear Combinations” (especially Equation 5–7) in Nunnally and Bernstein (1994). For a specific mathematical proof that the covariance between raw profiles equals the sum of the covariances between the respective profiles’ normative and distinctive components, see Appendix A of Furr (2008).

2. To illustrate profile standardization, imagine five items are rated: 1, 1, 3, 5, 5. Since the profile has a M = 3 and SD = 2, each item’s standardized rating = (Raw Rating – M)/SD = (Raw Rating − 5)/2. Thus, the standardized profile is: −1, −1, 0, +1, +1.

3. We need not bother regressing Normative Perceived Similarity on identification because the b must be zero. Within any particular sample, normative similarity is a constant. Since normative similarity’s contribution to overall similarity is identical for every individual, it cannot contribute to individual differences in perceived similarity.

4. Reliability is more apt to be problematic when estimating normative profiles from small samples (e.g., if we were testing the association between group identification and perceived self-group similarity in small workplace teams).

5. The average projective and introjective similarity coefficients must equal zero. Recall that introjective similarity coefficients reflect the degree to which a student makes self-ratings that mirror normative stereotype-ratings more than or less than other students do; thus, to whatever degree some students do so more than average (yielding positive coefficients) others must do so less than average (yielding negative coefficients). Likewise, projective similarity coefficients reflect the degree to which a student makes stereotype-ratings that mirror normative self-ratings more than or less than other students do; thus, to whatever degree some students do so more than average (yielding positive coefficients) others must do so less (yielding negative coefficients).

6. I am very grateful to Kathryn Denning and Sara Hodges for making their data publicly available and for answering my fussy questions so quickly and kindly.

7. For a formal model of how the interaction of evaluative items with rater evaluative attitudes inflate similarity indices, see Leising et al. (2015).