Generalized Estimating Equations for Vector Regression

Generalized estimating equations (GEE; Liang & Zeger, 1986) are extended to general vector regression settings. When the response vectors are of mixed type (e.g. continuous–binary response pairs), the proposed method is a semiparametric alternative to full-likelihood copula methods. When the response vectors are of the same type (e.g. physical measurements on left and right eyes), the proposed approach can be viewed as a “plug-in” to existing methods, such as the vglm function from the state-of-the-art vgam R package of Yee (2015). In either scenario, the proposed approach offers accurate inferences on model parameters regardless of whether the working model is correctly or incorrectly specified. The finite-sample performance of the proposed method is assessed using simulation studies based on a burn injury dataset (Song, 2007) and a sorbinil eye trial dataset (Rosner et. al, 2006). Simulation results are complemented by data analysis examples using the same two datasets. keywords: sandwich estimator of variance; vector regression