Cy. Wang et al., Regression analysis when covariates are regression parameters of a random effects model for observed longitudinal measurements, BIOMETRICS, 56(2), 2000, pp. 487-495
We consider regression analysis when covariate variables are the underlying
regression coefficients of another linear mixed model. A naive approach is
to use each subject's repeated measurements, which are assumed to follow a
linear mixed model, and obtain subject-specific estimated coefficients to
replace the covariate variables. However, directly replacing the unobserved
covariates in the primary regression by these estimated coefficients may r
esult in a significantly biased estimator. The aforementioned problem can b
e evaluated as a generalization of the classical additive error model where
repeated measures are considered as replicates. To correct for these biase
s, we investigate a pseudo-expected estimating equation (EEE) estimator, a
regression calibration (RC) estimator, and a refined version of the RC esti
mator. For linear regression, the first two estimators are identical under
certain conditions. However, when the primary regression model is a nonline
ar model, the RC estimator is usually biased. We thus consider a refined re
gression calibration estimator whose performance is close to that of the ps
eudo-FEE estimator but does not require numerical integration. The RC estim
ator is also extended to the proportional hazards regression model. In addi
tion to the distribution theory, we evaluate the methods through simulation
studies. The methods are applied to analyze a real dataset from a child gr
owth study.