Essays on the econometrics of social networks and peer effects
Chan, TszKin Julian
MetadataShow full item record
This dissertation addresses statistical issues related to endogenous peer selection in the context of social networks, social interaction models and snowball sampling methods. The first chapter studies the peer effects of friends, studymates, and seatmates on academic performance. We use data on social networks, personality traits, and cognitive ability measures collected through a unique survey conducted in three schools in Hong Kong. We estimate a social interaction model which accounts for endogenous network formation and correlation between multiple networks with a Bayesian approach. Our results show that the cognitive ability of studymates and the conscientiousness of friends positively affect a student's mathematics exam score while the conscientiousness of studymates and the cognitive ability of friends have no effect. The second chapter proposes a novel identification strategy for social interactions in a model with endogenously formed social networks. The network endogeneity arises from the correlation between the links of the network and the unobservables that determine the outcome of interest. We show that the eigenvectors of the adjacency matrix that defines the social network are control variables for network endogeneity without imposing any parametric assumption. We propose an information criterion to select the number of eigenvectors to be included as control variables. We apply the proposed method to the same empirical application as the first chapter and compare the results. The third chapter studies a snowball sampling method for social networks with endogenous peer selection. Snowball sampling is a sampling design which preserves the dependence structure of the network. It sequentially collects the information of vertices linked to the vertices collected in the previous iteration. The snowball samples suffer from a sample selection problem because of the endogenous peer selection. We propose a new estimation method that uses the relationship between samples in different iterations to correct selection. We use the snowball samples collected from Facebook to estimate the proportion of users who support the Umbrella Movement in Hong Kong.