Rank-Tracking Probabilities with Applications in Longitudinal Studies

• Activity Number: 627
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Xin Tian*+ and Colin Wu
• Companies: National Heart, Lung, and Blood Institute and National Heart, Lung, and Blood Institute
• Keywords: Basis approximation ; Conditional distribution ; Rank-tracking probability ; Longitudinal studies ; Kernel estimate

Abstract

An important scientific objective of longitudinal studies involves tracking a subject's ability of having certain health status at a later time point given the subject's health status at an earlier time point. Proper definitions and estimates of tracking abilities have important implications for guiding and justifying long-term biomedical studies. We propose in this paper a class of "rank-tracking-probabilities" (RTP) to describe a subject's conditional probabilities of having certain health ranks relative to the population at two different time points, and show that the RTPs can be used as an important quantitative measure of tracking abilities of an outcome variable in longitudinal studies. Nonparametric estimation and inference methods for RTPs and their functions are developed for large longitudinal data using both local and global smoothing methods. Applying our methods to an epidemiological study of childhood obesity, we demonstrate that RTPs and their nonparametric estimators and inferences provide a comprehensive set of statistical tools for many scientific objectives that cannot be effectively investigated using the usual conditional-mean based longitudinal methods.