Bayesian Inference in Semiparametric Mixed Models for Longitudinal Data

• Published in: Biometrics Vol. 66; no. 1; pp. 70 - 78
• Main Authors: Li, Yisheng, Lin, Xihong, Müller, Peter
• Format: Journal Article
• Language: English
• Published: Malden, USA Blackwell Publishing Inc 01.03.2010; Wiley-Blackwell; Blackwell Publishing Ltd
• Subjects: Algorithms; Bayes Theorem; Bayesian analysis; Bayesian theory; BIOMETRIC METHODOLOGY; Biometrics; biometry; Biometry - methods; Computer Simulation; data collection; Data Interpretation, Statistical; Data smoothing; Dirichlet process prior; Epidemiologic Methods; Identifiability; Inference; Information Storage and Retrieval - methods; Longitudinal data; Longitudinal Studies; Matrices; Modeling; Models, Statistical; Multilevel models; Nonparametric models; Parametric models; Postprocessing; Random effects; Regression analysis; Research methodology; shrinkage; Smoothing spline; Statistical variance; Uniform shrinkage prior; variance; Variance components
• ISSN: 0006-341X; 1541-0420; 1541-0420
• DOI: 10.1111/j.1541-0420.2009.01227.x

We consider Bayesian inference in semiparametric mixed models (SPMMs) for longitudinal data. SPMMs are a class of models that use a nonparametric function to model a time effect, a parametric function to model other covariate effects, and parametric or nonparametric random effects to account for the within-subject correlation. We model the nonparametric function using a Bayesian formulation of a cubic smoothing spline, and the random effect distribution using a normal distribution and alternatively a nonparametric Dirichlet process (DP) prior. When the random effect distribution is assumed to be normal, we propose a uniform shrinkage prior (USP) for the variance components and the smoothing parameter. When the random effect distribution is modeled nonparametrically, we use a DP prior with a normal base measure and propose a USP for the hyperparameters of the DP base measure. We argue that the commonly assumed DP prior implies a nonzero mean of the random effect distribution, even when a base measure with mean zero is specified. This implies weak identifiability for the fixed effects, and can therefore lead to biased estimators and poor inference for the regression coefficients and the spline estimator of the nonparametric function. We propose an adjustment using a postprocessing technique. We show that under mild conditions the posterior is proper under the proposed USP, a flat prior for the fixed effect parameters, and an improper prior for the residual variance. We illustrate the proposed approach using a longitudinal hormone dataset, and carry out extensive simulation studies to compare its finite sample performance with existing methods.