Marginal Regression Models for Mixed Bivariate Responses

A new framework for marginal regression model with bivariate responses from different distributions was proposed in this study. It adopts a Generalized Estimating Equation (GEE) approach of model estimation. A framework for mixture of response variables from different distributions was proposed for “Normal and Poisson”, “Normal and Bernoulli”, and “Poisson and Bernoulli”. Application on the proposal framework was examined in measuring the effect of certain hospital inputs on hospital performance in three selected tertiary health institutions.