B0201
Title: Gamma regression with censored outcomes and missing data
Authors: Jean-Francois Dupuy - INSA de Rennes (France) [presenting]
Abstract: Gamma regression is a member of the family of generalized linear models that have proved useful in several domains for modeling a positive-valued outcome as a function of explanatory variables. The case of a right-censored outcome in Gamma regression was recently examined in the literature (right-censoring occurs when the exact outcome is not observed, but is known to be greater than or equal to the observed outcome value). We go a step further and investigate estimation in the Gamma regression model when the outcome is right-censored and censoring indicators are missing at random (MAR). We propose and investigate an augmented inverse probability weighted (AIPW) estimator adapted to this setting. We describe its asymptotic properties (this estimator is consistent and asymptotically normal) and its double robustness property. We also describe a simulation study that investigates the finite sample performance of the proposed estimate.
