On Zero inflated models with applications to maternal healthcare utilization

Abstract

We consider the problem of modelling count data with excess zeros and over-dispersion which are commonly encountered in various disciplines  that limit the use of traditional models for count outcomes. Our research work applies the Zero-inflated Poisson and Negative Binomial models in  modelling Maternal Health Care (MHC) utilization in Nigeria, employing the Andersen’s behavioural model to examine the effect of predisposing,  enabling, and need factors on MHC utilization. The performance of these models are compared to the traditional Poisson and negative binomial  models. The Vuong test and AIC suggests that the Zero-inflated Negative Binomial model provided the most significant improvement over traditional  models for count outcomes.