A Comparison of Rank-Based Methods for Mixed Models

• Activity Number: 631
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): John Kloke*+ and Joseph W. McKean and M. Mushqur Rashid
• Companies: Bucknell University and Western Michigan University and FDA
• Keywords: Nonparametric ; Random effects ; Rank regression scores ; Robust ; Wilcoxon

Abstract

In recent years several rank-based approaches for analyzing cluster correlated data have been developed. Among these are two procedures for estimating the fixed effects in a mixed model. One such procedure is based on the joint rankings of all the residuals (Jaeckel type). The inference for this procedure involves robust estimations of correlation due to the random effect. A second procedure is based on minimizing a sum of rank-based convex functions over the clusters. The rankings of residuals for this second procedure are performed with in each cluster (Friedman type). In this talk we first provide a brief overview of these methods. Next we present the results of a large simulation study.