A0705
Title: Modeling reciprocity in directed networks
Authors: Chenlei Leng - University of Warwick (United Kingdom)
Rui Feng - University of Warwick (United Kingdom) [presenting]
Abstract: Reciprocity-the tendency for two individuals to form mutual connections, a common feature of directed networks. Two models that incorporate covariates are presented to capture reciprocity. The first introduces a novel Bernoulli model with reciprocity, along with an associated inference procedure. A key contribution is the analysis of effective sample sizes corresponding to different components of the model's parametrization. The second model extends the classical $p_1$ framework that incorporates node-specific heterogeneity by adding link-specific reciprocity. A novel conditioning argument is developed for estimating the reciprocity parameters, and the minimax optimality of the resulting estimator is established. Numerical experiments are provided to support the theoretical results.
