Analysis of Binary Longitudinal Data with Time-Varying Coefficients

 
   Lead Investigator:    Seonghyun Jeong
   Institution      :    North Carolina State University
   E-Mail           :    sjeong4@ncsu.edu
   Proposal ID      :    773



   Proposal Description:
   This research considers the analysis of longitudinal data where a binary          response variable is observed repeatedly for each subject over time. In           analyzing such data, the common assumption that regression coefficients are       constant over time may need to be relaxed to account for time-varying effects     of some subject characteristics. This research provides a Bayesian method to      estimate time-varying coefficient models with random effects to account for       subject specific effects of binary longitudinal data as well as                   between/within-subject variation. The method capitalizes on the method of         partial collapse and the basis search technique, thereby facilitating posterior   computation and accommodating spatially inhomogeneous curvature of                nonparametric functions, while avoiding overfitting. The proposed methods will    be illustrated with a simulated study and a real data set.