Fixed-Effects Models for Longitudinal Binary Data with Drop-Outs Missing at Random

• Activity Number: 398
• Type: Invited
• Date/Time: Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Paul Rathouz*+
• Companies: The University of Chicago
• Keywords: subject-specific model ; missing data ; conditional logistic regression ; double robustness ; semi-parametric efficiency

Abstract

We consider the problem of attrition under a logistic regression model for longitudinal binary data in which each subject has his own intercept parameter, eliminated via conditional logistic regression. This fixed-effects model exploits the longitudinal data by allowing subjects to act as their own controls. By conditioning on the drop-out process, we develop a valid but inefficient conditional likelihood using the complete-record data. Noting the drop-out process is ancillary in this model, we use a projection argument to develop a score with improved efficiency over the conditional likelihood score and embed both of these scores in a more general class of estimating functions. We propose a member of this class that approximates the projected score while being more computationally feasible. We present a small simulation and an example analysis from aging research.