On Resisting Outliers in Bivariate Data Using Weighted Measures of Rank Correlations

Research Paper (postgraduate) from the year 2015 in the subject Statistics, , language: English, abstract: This paper aims to compare some known weighted rank correlation measures with recent weighted measures in resisting outliers. The stability of type I error and the stability of power for these measures is introduced through a simulation study applied on some bivariate distributions containing outliers.

In this paper, we will compare between 13 weighted rank correlation coefficients for different sample sizes. The comparison will be in resisting outliers through a simulation study to compare the stability of type I error and to compare the stability of the Power for these measures. The simulation study was of size 10,000 repeat and was done for 13 different bivariate distributions for different sample sizes.