A Model-Based Analysis of Peer Effect in Educational Transition

Educational choices after high school are related to indi-vidual characteristics and aspirations. However, growing attention has been devoted to the relational contexts in which students are embed-ded, as peers provide both normative guidance and social reinforce-ment. Classes represent bounded social environments characterised by daily interactions, making them particularly suitable settings for exam-ining how choices may develop through social interactions. The inves-tigation of these relational mechanisms is methodologically challenging. Students are embedded in hierarchical institutional contexts and con-nected through social ties, generating statistical dependence. While con-ventional regression and multilevel models account for clustering, they do not explicitly incorporate the presence of peer relations. Conversely, network-based approaches require careful specification, especially in the presence of sparse and structurally bounded networks at the school level.To address these issues, the present study employs a network modelling approach for analysing a binary educational outcome, conditional on both individual covariates and observed network structures. This framework enables the assessment of local relational dependence within intra-class networks, providing a rigorous basis for disentangling peer influence from compositional effects.