A Multivariate Network Test of Extended Contact Hypothesis

• In Event: How to Address Prejudice and Discrimination in Childhood and Adolescence

Abstract

For more than two decades, anti-bias research and interventions focused on extended contact as a pathway to improve intergroup relations (Zhou et al. 2018). The classical notion is that knowing that an ingroup member has a close relationship with an outgroup member may lead to perceiving outgroup members more favorably (Wright et al. 1997). Extended contact focuses exclusively on intergroup attitudes as the outcome, but may offer additional intergroup benefits, such as future direct friendships with outgroup members. Most studies employ an individual approach that suits the “traditional definition” of extended contact and count the number of ingroup friends who have outgroup friends (for a review: Zhou et al. 2018); other studies quantify the number of outgroup friends that ingroup friends have. However, by aggregating the friend data we lose information about the relational structures and larger social network configurations in which friendships between individuals are embedded. To overcome this problem, we apply a novel multivariate network approach to examine the interdependence of ingroup (same-ethnic) and outgroup (cross-ethnic) friendships.

From a network science perspective, structural balance theory (Cartwright & Harary 1956; Newcomb 1956) predicts that positive direct relationships with someone shape a person’s sentiment about someone else (Heider 1946). The idea is that people prefer balanced configurations as enmity among one’s friends leads to strain and is avoided and, thus, prefer friendships with indirect connections (“a friend of a friend is a friend”). In our study, this would suggest that students are likely to become (or stay) friends with the cross-ethnic friends of their same-ethnic friends (Fig. 1a, H1a). Students may also be involved in existing cross-ethnic friendships that lead to the formation of new cross-ethnic friendships between their friends (Fig. 1b, H1b).

To test our hypotheses, we draw on two large, diverse U.S. high schools (total N = 4454; 50% girls) with students in 9th to 11th grade followed across three waves of data within 12 months of high school (April 2017-2018). Students were asked to name their ten closest friends at school of any gender and from any grade. To account for the grouping structure in the large school friend networks, available demographics (gender and grade) and shared extra-curricular activities were included in the analysis. We estimated a pooled model that constrained estimates to be equal across schools while recognizing that friendships were not possible between students in different schools. Following Snijders et al. (2020) the networks were split into same-ethnic and cross-ethnic friendships and analyzed by the multivariate extension of the stochastic actor-oriented model in RSiena (Snijders et al. 2013).

Support was found for both hypotheses (Table 1, Fig. 1): cross-ethnic friendships are likely to be formed (and maintained) when adolescents have same-ethnic friends who have cross-ethnic friendships (H1a) or when they have cross-ethnic friends who have cross-ethnic friendships (H1b). This study shows that bivariate network analysis with balance theory can be fruitful for understanding the development of cross-ethnic friendships. New research should clarify whether the processes are similar for different ethnic-racial groups.

Authors

• Johannes Ashwin Rambaran, Universiteit Utrecht
  Presenting Author

• Deborah Rivas-Drake, University of Michigan - Ann Arbor
  Non-Presenting Author

• Adriana J. Umaña-Taylor, Harvard University
  Non-Presenting Author

• David Schaefer, University of California - Irvine
  Non-Presenting Author

• Allison M Ryan, University of Michigan
  Non-Presenting Author