AN ALTERNATIVE RANK CORRELATION COEFFICIENT WİTH GRAPHICAL APPROXIMATION

Yaprak Arzu ÖZDEMİR, Meltem EKİZ
1.752 592

Abstract


 ABSTRACT

In this study, a rank correlation coefficient (v) proposed by Blest(2000), which does not require normality assumption in hypothesis testing of correlation coefficient for interval level or ratio scaled measured variables, is introduced. This rank correlation coefficient, which is examined based on graphical presentation and their comments, has a structure, which considers the interchanges in the early ranking of the data. When the sample size is five this introduced v rank correlation coefficient is considered and it has showed that it is more advantageous compared to other rank correlation coefficients since it gives different results even if for small changes in ranking. Furthermore, the critical values of v correlation coefficient are obtained for different significance levels and these values are presented in a table.


Keywords


Key Words: Ranks, Kendall’s τ rank correlation coefficient, Spearman’s ρ rank correlation coefficient

Full Text:

PDF