Abstract
The role of network structure in intergroup differentiation--the bipolarization of stereotypes that are defensive (ingroup-positive/outgroup-negative) and non-defensive (outgroup-positive/ingroup-negative)--was investigated using a Dynamic Social Impact Theory (DSIT) framework. Three computer-mediated laboratory social network experiments were pooled to test the interaction of network clustering (cliquish structure) and segregation (personal network homogeneity) on intergroup differentiation. Democrats and Republicans during the five months preceding the 2008 U.S. Presidential election, deaf and hearing persons, and women and men participated. Twenty-six 16-person groups (e.g., 8 Democrats, 8 Republicans) serially discussed nine controversial stereotyped rumors in lattice (unclustered) or "family" (clustered) network structures. Support was found for an Ingroup Echo Chamber Effect : segregation led to intergroup differentiation (stronger defensive belief, weaker non-defensive belief) in clustered, but not unclustered, structures. At the individual level, network clustering amplified ingroup neighbor social influence, leading participants to think more positively of their ingroup and more negatively of their outgroup.