A New Scale-Invariant Nonparametric Test for Two-Sample Bivariate Location Problem

• Activity Number: 340
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Sunil Mathur*+ and Sujay Datta
• Companies: University of Memphis and University of Akron
• Keywords: Location ; Test ; Bivariate ; Asymptotic ; Normal

Abstract

Diagnostic testing in medicine is crucial in determining interventions and treatment plans. It is important to analyze diagnostic tests accurately so that right decision can be made by clinicians. A scale-invariant test is proposed when treatment and control samples are available and change in condition between treatment and control group is investigated. The proposed test is shown to have asymptotically normal distribution. The power of the proposed test is compared with several of the existing tests using Monte Carlo simulation techniques under different bivariate population set-ups. The power study shows that the proposed test statistic performs very well as compared to its competitors for almost all the changes in location and for almost all the distributions considered in this study. Under non-normal population set-up, the proposed test statistic dominates all of its competitors considered here for the study for almost all shifts in location. The computation of proposed test statistic is shown using a real-life data set in which the development of the immune system in 36 HIV-positive new-born children is investigated.