 |
 |
Hi everyone, Something is puzzling me….it relates to effect size estimate for the Mann Whitney test. In the text “Green, S. B. & Salkind, N. J. (2005). Using SPSS for Windows and Macintosh: Analyzing and understanding data (fourth edition). New Jersey:Pearson (p386)”, the effect size for a Mann Whitney test is defined as (chisquare value for the Kruskal Wallis test)/(N-1) -this is equivalent to eta squared for the ANOVA of the ranked data. Now, in the paper “Effect Size Estimates: Current Use, Calculations, and Interpretation” by Fritz, Morris and Richler (2012) [http://www.researchgate.net/publication/51554230_Effect_Size_Estimates_Current_Use_Calculations_and_Interpretation] , eta squared for the Mann Whitney test is correctly quoted as: (z/sqrt(N))^2 = (chisquare value for the Kruskal Wallis test)/(N-1) But it is implied that the “effect size” is the square root of this value, i.e. z/sqrt(N). Your views on this issue are much appreciated. All the very best, Kim You may leave the list at any time by sending the command SIGNOFF allstat to [log in to unmask], leaving the subject line blank.