This paper provides the semiparametric analysis of binary choice with social interactions and investigates the heterogeneity of the social interactions. We introduce the quantile regression to the binary response model with social interactions in a large network. We study the heterogeneity of the covariate effects as well as peer effects which richen the understanding beyond the center. Nested pseudo smoothed maximum score (NPSMS) algorithm is proposed to estimate the model. We illustrate the finite sample performance of the model and NPSMS with Monte Carlo experiments and an empirical application to the social interactions among students over exercise decision. |