The $\beta$-model for Random Graphs --- Regression, Cram\'er-Rao Bounds, and Hypothesis Testing

TL;DR

This paper introduces a regression extension of the $eta$-model for random graphs, deriving Cramér-Rao bounds, comparing estimators via simulations, and demonstrating applications in social network analysis.

Contribution

It generalizes the $eta$-model by incorporating regression functions, providing theoretical bounds, and applying the methods to real dynamic social network data.

Findings

Derived Cramér-Rao bounds for various $eta$-models.

Compared maximum likelihood estimators to bounds through simulations.

Demonstrated model applicability on healthcare worker communication networks.

Abstract

We develop a maximum-likelihood based method for regression in a setting where the dependent variable is a random graph and covariates are available on a graph-level. The model generalizes the well-known $\beta$-model for random graphs by replacing the constant model parameters with regression functions. Cram\'er-Rao bounds are derived for the undirected $\beta$-model, the directed $\beta$-model, and the generalized $\beta$-model. The corresponding maximum likelihood estimators are compared to the bounds by means of simulations. Moreover, examples are given on how to use the presented maximum likelihood estimators to test for directionality and significance. Last, the applicability of the model is demonstrated using dynamic social network data describing communication among healthcare workers.