Is the minimal group paradigm culture dependent? A cross-cultural multi-level metaanalysis Ronald Fischer & Crysta Derham (Victoria University of Wellington, New Zealand)
Two main paradigms of individualgroup relations Social identity theory (Tajfel & Turner, 1979) Individualism-collectivism (Triandis, 1995)
In-group bias = positive evaluation of the ingroup compared to the out-group (Brewer, 1979), even if membership is nonconsequential or arbitrary
Minimal group paradigm (Tajfel, 1971) a) individuals have no face-to-face contact, b) group membership is completely anonymous, c) there is no instrumental or rational link between the categorization and the experimental tasks to be completed by participants, d) the responses by participants have real implications for the groups, but not the responding individual Tajfel et al., 1971, pp. 153-154
How strong is this effect? Mullen et al. (1992): moderate effect size Buhl (1999): moderate effect size
Is in-group bias universal? Second major paradigm of individual-group relations: Individualism-collectivism (Triandis, 1995) Hinkle & Brown (1990): in-group bias stronger in collectivistic contexts Brown et al. (1992): effects stronger for relational collectivists But: Morales et al. (1998) = no effect; Capozza et al. (2000) = reverse effect Individualists may be more sensitive to group information (see review by Fischer, 2008)
Uncertainty Identity Theory (Hogg, 2000, 2007) Self-uncertainty = primary motive underlying identification with social groups Feeling uncertain = aversive psychological state; motivation to reduce uncertainty Categorization of self & others as part of social groups => Proto-types (fuzzy set of attributes: define/differentiate groups + provide prescriptive norms and guidelines for appropriate behaviour). For others, depersonalization helps to cognitively predict how people will behave as members of their group. For the self, depersonalization prescribes acceptable in-group behaviour as well as providing a sense of identification and belonging, which is a fundamental human need (Baumeister, 2005). Depersonalization into group prototypes therefore reduces uncertainty about who one is, how one should behave and how one will be treated by others (Hogg, 2009, p. 222).
Questions How universal is in-group bias? Does individualism-collectivism affect the relative strength of in-group bias? Does uncertainty avoidance as a contextual variable affect in-group bias (in line with uncertainty identity theory)?
Meta-analysis PsycInfo search ( in-group bias or minimal group ) + published meta-analyses (e.g., Aberson, Healy, & Romero, 2000; Mullen, Brown, & Smith, 1992, Robbins & Krueger, 2005) 498 articles + 20 unpublished ms & dissertations Final coded data: 269 samples from 121 articles based on a total N of 21,266 participants from 18 societies
Variables Dependent variable: Effect size for in-group bias Predictor variables: Individualism (Hofstede, 1980) Uncertainty avoidance (Hofstede, 1980)
Analytical strategy Effect sizes converted to z (r-to-z transformed correlations) 3-level mixed effects v-known meta-analysis Level 1: effect size, variance component (v-known analysis) Level 2: publication year, group status (high, equal, lower), group type (artificial, real) Level 3: individualism, uncertainty avoidance
Descriptive results Mean effect (random effects model) =.369, standard error.021, 95% confidence intervals:.327 to.410; z = 17.37, p <.0001. There was significant variation in effect sizes: Q (268) = 2224.06, p <.0001.
Level 2 effects Publication year: γ =.001, t =.45, n.s. Group type: γ =.010, t = 1.33, n.s. Group status: γ =.055, t = 1.62, p =.13; with robust standard errors: γ =.001, t = 4.06, p <.01); higher status groups show higher ingroup bias
Level 3 effects Individualism : γ =.003, t robust = 2.20, p <.05. Greater individualism was associated with more in-group bias, supporting the postmodern paradox hypothesis (but only when controlling for UA). Uncertainty avoidance : γ =.007, t robust = 4.29, p <.01. Greater uncertainty avoidance was associated with greater in-group bias. This supports the uncertainty-identity model. Interaction between group type (level 2) and countrylevel characteristics (level 3) on in-group bias (level 1): Uncertainty avoidance: γ =.012, t robust = 3.34, p <.05. Individualism: γ =.011, t robust = 1.76, p =.08.
0.941 Low Uncertainty Avoidance High Uncertainty Avoidance UAI= -17.933 UAI= 19.067 0.684 Ingroup ZR Bias 0.426 0.168-0.089 Laboratory 0.00 Real 1.00 Target REALGroup
Discussion Moderate in-group bias, but highly variable across societies Consistent effect in line with uncertainty identity theory Individualism: postmodern paradox versus relational collectivism
Implications To date, theoretical and empirical work has emphasized the effects of threat (Stephan & Stephan, 2000), little research on uncertainty. A societal concern with uncertainty avoidance increase the likelihood to favour the in-group over the outgroup. Significant real-world implications (money, jobs, access to basic resources or health, education and social services). The overall levels of uncertainty and the social climate can potentially have a large impact on intergroup relations.
Cross-cultural studies Cross-cultural research can be used to test and confirm mono-cultural research More studies of in-group bias are needed, taking into account specific societal contexts and parameters
Thank you! Ronald.Fischer@vuw.ac.nz