Locally Efficient Semiparametric Estimators for a Measurement Error Model in Longitudinal Studies

• Activity Number: 656
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Laura "Liqiu" Jiang*+
• Companies: Kendle International Inc.
• Keywords: Semiparametric ; Measurement error model ; Longitudinal

Abstract

In longitudinal studies, we are interested in the relationship between a binary response and a profile of repeated measurements, where the subject profile is a general linear model with random effects. If the profile is linear, the relationship of the binary response to the individual intercept and slope is of interest. The naive method to fit a regression model to estimate individual random effects can lead to biased results. Conditional score approaches for generalized linear models require no distributional assumption for the random effects and yields consistent inference regardless of the true distribution. However, the estimator can only be used for generalized linear models in canonical form with normally distributed measurement error. We overcome this limitation with locally efficient semiparametric estimators (Tsiatis and Ma; Biometrika 2004) for functional measurement error models for longitudinal studies. The distribution of random effects can be misspecified and the method still yields consistent inference. We show, via simulation, the estimator reduces bias and improves empirical probability coverage compared to the naive method.