How to Use R for a Fully Nonparametric Comparison of Multivariate Samples

• Activity Number: 399
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Amanda Ellis*+ and Woody Burchett and Arne C Bathke and Solomon W. Harrar
• Companies: University of Kentucky and University of Kentucky and University of Kentucky and University of Montana
• Keywords: nonparametric ; Multivariate ; R

Abstract

Experimenters are often concerned with data that contain multiple responses. When the responses are ordinal, or constitute a mixture of quantitative, ordinal, and binary variables, a nonparametric approach is advisable. However, only recently, fully nonparametric methods have been developed for nonparametric inference on multivariate data, and these methods can be complex and difficult to compute. In what follows, we show how to use R to facilitate the analysis of one-way multivariate data. The tools are designed to allow investigators to perform nonparametric test in a user friendly manner, while still being concise, valid, and informative. They offer visual results of the data as well as a variety of newly developed nonparametric tests, including validated small sample approximations. In addition to an overall analysis, the user has the option to perform subset analyses for different combinations of responses.