This paper follows the literature on separating out the structural effects of social contexts from individuals' own preferences in determining social relationship formations. While prior work has almost exclusively focused on compositional measures of social context, we posit that, net of the effect of group composition, total size has a distinct effect on the level of segregation in a population. In this paper, we first formalize a theoretical framework for the analysis of the size effect. Under the assumptions of (1) maximization of preference in choosing a friend, (2) multidimensionality of preference, and (3) racial homophily, we conducted analyses using agent-based modeling techniques and yielded three main findings. First, increasing total population size decreases the likelihood of forming intergroup relations. Second, the size effect increases with the number of preference dimensions. Third, the size effect decreases with the strength of preference. In addition, we empirically tested our theory using friendship nomination data from Add Health and found results consistent with our prediction that increasing total school size reduces the share of interracial friends in a school.
Country of focus: United States of America.