Robust Matrix-Based Measures of Agreement Based on L-Statistics for Repeated Measures

• Activity Number: 614 - Statistical Methods for Longitudinal and Other Dependent Data
• Type: Contributed
• Date/Time: Thursday, August 1, 2019 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Elahe Tashakor* and Vernon Chinchilli
• Companies: Pennsylvania State University and Pennsylvania State University
• Keywords: L-statistics; agreement; robust estimation; multivariate

Abstract

The concordance correlation coefficient is commonly used to assess agreement between two raters or two methods of measuring a response when the data are measured on a continuous scale. It typically is used under the assumption that data are normally distributed. However, in many practical applications, data often are skewed and/or thick-tailed. Previously, we have proposed an approach that extends existing methods of robust estimators to produce more robust versions of the concordance correlation coefficient. In this paper we extend the application of this class of estimators to a multivariate situation, possibly repeated measurements, based on a matrix norm that possesses the properties needed to characterize the level of agreement between two p × 1 vectors of random variables. We provide two data examples to illustrate the methodology, and we discuss the results of computer simulation studies that evaluate statistical performance.