Modeling Distal Scalar Response with Zero-Inflated Count Longitudinal Covariates

• Activity Number: 28
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Hanyu Yang*+ and Runze Li and Anne Buu and Robert A. Zucker
• Companies: Penn State and Penn State and University of Michigan and University of Michigan
• Keywords: functional data ; zero-inflated count ; measurement error ; time-varying effect ; health risk behavior

Abstract

We propose a generalized time-varying effect model that handles the observed longitudinal covariate processes of zero-inflated counts, and accommodates to response variable from any distribution of exponential family. A two-stage approach is developed, using smoothing spline techniques, to characterize individual developmental trajectories, and delineate their time-varying effects on distal outcomes. Longitudinal data from a well-known study of youth at high risk for substance abuse are analyzed. The result demonstrates that the proposed model is useful in informing critical developmental periods of prevention or intervention. A set of simulation experiments is conducted to assess the performance of the model under different circumstances. Our simulation study also suggests that, when count data contain excess zeros, the two-stage model involving regular Poisson cannot estimate either the longitudinal covariate process or its time-varying effect well. This result, therefore, emphasizes the importance of modeling zero-inflation in observed longitudinal covariates.