Linking structure and content: Friendship networks and academic achievement.

摘要

Current federal policies assume that schools and teachers can raise test scores and reduce achievement gaps. Peer relations are usually ignored in accountability reforms, but particularly among teenagers, peer relations are extremely important. The purpose of this dissertation is to examine adolescent peer networks as a context that may promote high academic performance.;Theoretically, this project conceives of the social world as consisting of macro and micro processes and dynamic social relations. It bridges several research traditions and literatures: social network analysis, the sociology of education and the sociology of the youth. The organizing argument is that to understand how friendship networks affect achievement we need to take into account both the pattern of social relationships within which students are embedded (network structure) and their action orientation, shared meanings and social emotions (network content). To address this issue, I analyze school-based friendship networks from the National Longitudinal Survey of Adolescent Health. I test whether network structural properties (ego-send density, popularity, centrality, and actor's embeddedness in school networks) act in combination with friends' school attachment (network content) to influence academic performance.;The most consistent findings are related to actor centrality and actor's embeddedness in school networks. Regarding actor centrality, I found in an OLS regression framework that the association between friends' attachment to school and achievement (math and English) is conditional on centrality levels. High-centrality students connected to high-attachment friends outperform their low-centrality counterparts linked to equally high-attached peers. However, this finding did not hold up in a fixed-effects framework, which suggests that unobserved individual heterogeneity may have been driving the statistical association found.;Regarding actor's embeddedness in school networks, I found that students embedded in highly cohesive networks and connected to high-attached students perform better than their comparable counterparts embedded in less cohesive webs of social relations. This finding shows consistency across subject-matters (math and English) and statistical models (OLS regression and fixed-effects estimation).;The findings suggest that the interplay between structure and content is a promising analytical and empirical avenue to make sense of the ways in which friendship networks may affect achievement or other social phenomena.