Fit Statistics for Nested Models in Which Parameter Estimates are Obtained Using Generalized Method of Moments in the Presence of Time-Dependent Covariates

• Activity Number: 392
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Maryann Shane*+
• Companies: University of Northern Colorado
• Keywords: Generalized Method of Moments ; time-dependent covariates ; longitudinal data ; goodness-of-fit ; nested models ; information criterion

Abstract

The Generalized Estimating Equations (GEE) approach proposed by Liang and Zeger (1986) has become the most popular method in estimating parameters for longitudinal data. This technique accounts for correlation inherent among repeated observations, allowing the researcher to specify the nature of this working correlation. However, GEE presents major disadvantages when time-dependent covariates (TDCs), special types of predictors involving a feedback loop, are present (Pepe & Anderson, 1994; Fitzmaurice, 1995). Recently, Generalized Method of Moments (GMM) has been proposed as an alternative to GEE when estimating parameters of longitudinal data with time-dependent covariates (Lai & Small, 2007).

There has been a lack of attention paid to GMM fit statistics in the literature. This paper presents two statistics to assess the goodness-of-fit of models estimated using GMM in the presence of TDCs. Comparisons of identifying poor fit in nested models is compared to similar capabilities of the quasi-likelihood information criterion (QIC) using GEE.