The factor structure of an intragroup identification scale was examined. Subjects participating in small groups worked on a decision-making task, complctcd the identification measure and made in-group and out-group evaluations. Three factors reflecting emotional, cognitive and individual/group interdependence aspects of identification accounted for 70,5 percent of the scale's rota1 variance. Based upon these rcsulrs, the identification scale was decomposed into three subscales. Correlations between subscale scores and differentiation were positive, but small in magnitudc (range 0.23-0.28). In contrast, subscale scores were morc strongly associated with the in-group aspect of differentiation (rs from 0.37 to 0.62). Implications of the findings for social identity theory are discussed along with the value of multicomponent conceptualizations and methodologies.Social identity theory (SIT e.g. Tajfel & Turner, 1979) has been a major force in the resurgence of research on intergroup relations. Since the development of SIT, numerous empirical investigations have been reported which have examined diverse aspects of its formulations (see Brewer & Kramer, 1985;Tajfel, 1982). However, until recently, one of SIT'S most basic tenets, i. e. the positive relationship between in-group identification and intergroup differentiation (Kelly, l988), has not attracted substantial research attention. The study reported here examined the factor structure underlying a measure of in-group identification and the relationship between the components of identification and intergroup differentiation.Oaker & Brown (1986) summarized succinctly the nature of the link in SIT between identification and differentiation:It is assumed within this theory that associated with . . . identification is a need for some positive distinctiveness, a need which is thought to be often satisfied through social comparison to heighten di/ferencu between groups (p.768).Their research was part of a programmatic effort, carried out in both laboratory and field settings, to test the identification-differentiation hypothesis (see also Brown, Condor,

A recently proposed taxonomy describes four basic types of groups or group members by crossing the individualist-collectivist distinction with autonomous versus relational group orientations (Hinkle & Brown, 1990). The model posits that many of the basic relationships derived from Social Identity Theory will be most readily observed amongst collectivist groups with relational orientations. Three investigations (two laboratory experiments and one field study) were conducted to examine the orthogonality of the individualist-collectivist and autonomous-relational dimensions and test the hypothesis that significant, positive correlations between in-group identification and in-group bias are most apt to occur amongst collectivists with relational orientations. In all three studies, research participants completed an individualist-collectivist inventory and measures of autonomous versus relational group orientations, in-group identification, and in-group favouritism. As expected, correlations between individualism-collectivism and the measure of autonomous-relational orientations were quite low; the two dimensions manifested little overlap. Responses to the individualist-collectivist inventory and autonomous versus relational group orientations scale were used to divide the research participants into the taxonomy's four categories. As hypothesized, in all three studies significant, positive identificationhn-group favouritism correlations were observed for collectivists with relational orientations, whilst these same correlations amongst individualists with autonomous orientations were zero. The results from these studies support predictions derived from Hinkle & Brown's (1990) group taxonomy. The use of different paradigms, subject populations and measures of key constructs in the laboratory and field studies speaks to the model's generalizability. The general implication of these findings for Social Identity Theory and intergroup relations research is that our theoretical understanding and predictive precision can be enhanced by recognizing that the relevant psychological processes can vary across different types of groups and group members.*Requests for reprints.