Analysing data from religiously diverse German schools, we find similar aggregate levels of friendship segregation among Muslim boys and girls. Using stochastic actor-oriented networks models, however, we document substantial gender differences in the underlying mechanisms. As shown in the Figure below, we find that religious in-group bias is stronger for Muslim girls than for Muslim boys, but non-Muslims of either gender are reluctant to befriend Muslim boys but not girls.
A simulation analysis further demonstrates that these gendered individual-level processes result in comparable aggregate patterns of friendship segregation among Muslim boys and girls. As the Figure below indicates, religious friendship segregation thus arises because Muslim girls tend to self-segregate and non-Muslim youth are less willing to befriend Muslim boys although the latter are open to interreligious friendships.
The bottom line is that similar aggregate levels of friendship segregation mask large gender differences in the underlying mechanisms.This gendered friendship-making behaviour has implications for the larger question of minority social integration, as we discuss in the paper.
Sebastian Pink and I will give a course on “Collecting and Analyzing Longitudinal Social Network Data”. The course is part of the GESIS Summer School in Survey Methodology; it will take part August 16 to August 20 via Zoom. Below is a short course description; click here to learn more about the course and/or register.
Social scientists often are interested in understanding how social networks emerge and/or how they shape individual behavior. These questions of network formation (“selection”) and network effects (“influence”) concern both human individuals and organizational units. Examples for selection are the emergence of friendship between people or cooperation between firms; examples for influence are adolescents start smoking because of their friends or firms copying other firms’ strategies. Selection and influence are inherently dynamic processes, but few social scientists have been trained in collecting, processing and analyzing longitudinal social network data. This practical course guides participants who intent to collect and/or analyze longitudinal social network data. We start by conceptualizing and planning data collection, discussing both general challenges and, if applicable, participants’ own projects. Thereafter, participants learn how to handle and manage network data in R by guided examples and exercises. The main part of the course focuses on specifying, estimating and interpreting stochastic actor-oriented models (SAOM) for network dynamics, again with a mix of guided examples and practical exercises using the R package RSiena. We consider selection and influence as well as how SAOM can help to empirically disentangle these competing processes.
We used the data in many of our publications. They include several waves of longitudinal social networks of 11- to 17-year-olds in German schools. Networks are rather large, covering multiple classrooms rather than single ones. Minority youth are oversampled, thus allowing rather fine-grained analyses.
In addition, the data include measures of national, ethnic & religious identities, perceived discrimination, intergroup attitudes, and more.
I am hiring two PhD researchers (three years, 65%) for my new project “Religion, religiosity, and the social-emotional integration of Muslim youth”. The project is funded by the German Research Foundation (DFG) and directed at the Mannheim Centre for European Social Research (MZES).
The project examines the mechanisms that drive the social-emotional integration of Muslim youth. For this purpose, (longitudinal) secondary data analyses (e.g., CILS4EU, FIS, NEPS), choice experiments, and group discussions will be conducted. The main aim is to better understand to what extent religion and religiosity matter for friendship choices and the development of national identification of young Muslims. It will also be examined whether non-Muslim youth exclude their Muslim peers, and, if so, what consequences this has for their social-emotional integration.
Click here for the call for applications (and here for the German version). The deadline for applications is March 31, 2018. Further information about the project is found here. Please feel also free to email me.
Together with Tobias Stark, Hanno Kruse, and Sebastian Pink I am offering a session on the causes of segregation and intergroup relations in social networks at the XXXVIII Sunbelt conference. The conference will be held in Utrecht and take place June 26 to July 1, 2018.
If you’re interested in presenting, please submit your abstract until February 1. Our call follows below.
Intergroup Relations in Social Networks: Causes of Segregation
More and more researchers use social network analysis to study intergroup relations. Various methods, including ego-centered, cross-sectional (ERGM), and longitudinal (SAOM) network analyses show that friendship networks are segregated along ethnic, cultural and religious lines. However, research has only begun to exploit the potential of social network analysis for understanding the causes and processes leading to segregation. Even fewer studies exist on interventions that aim to reduce segregation and promote positive intergroup relations. This is partially due to theoretical and methodological challenges that are specific to the network approach. For instance, people may identify with multiple groups, intergroup contact preferences contact may vary between groups, and members of multiple groups in a network have to be modelled at once. This session invites theoretical, methodological, and empirical contributions that address these and other challenges in order to deepen our understanding of how segregation arises, with a focus on intergroup relations.
My article “Young immigrants’ host country identification and their friendships with natives: Does relative group size matter?” is now available at Social Science Research. Click here to read the paper.
In short, I argue that immigrants’ host country identification only affects their own friendship choices in schools with high shares of immigrants, because only in those schools they can be picky about befriending natives. Stochastic actor-oriented models support this notion, pointing to an interplay of preferences and opportunities in shaping the relation between host country identification and interethnic friendships.