A Note on Multivariate Antedependence Models

• Activity Number: 179
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Chulmin Kim*+ and Dale Zimmerman
• Companies: Rochester Institute of Technology and University of Iowa
• Keywords: Antedependence Models ; Longitudinal Data ; Covariance Structure

Abstract

In many of medical and biological studies, one or more attributes are measured on each subject over repeated time, yielding longitudinal data. The importance of covariance modeling has been noted for analyzing longitudinal data. The goal of covariance modeling, like that the mean modeling, is to obtain as parsimonious a presentation of the covariance as possible, yet one that fits the data well. Antedependence (AD) models can be very useful for covariance structure for longitudinal data. AD models are generalization of Autoregressive (AR) models that allow the variances and same-lag correlations to vary over time. We generalize the univariate AD models to multivariate AD (MAD) models and study some of their properties. An example is given to illustrate the usefulness of those properties of MAD.