S.O.S. - effect size indexes for nonparametric tests

[ladylaura]

Hi there folks,

I am carrying out a study (repeated measures design) employing nonparametric statistics ( Friedman's and Wilcoxin signed rank tests as post hoc comparisons) . I need to calculate effect size which SPSS does not process for these particular tests. I'm finding the literature very high brow and am confused with the choice of effect size indexes out there. I think that Cliff's delta is what I should be calculating but am unsure. Does anyone know what is routinely used in these circumstances and is there a computer program that can calculate this for me. Any help you could give would be most appreciated. I can't move on with the project until I have effect size calculated. Thanks.

Yours in crisis,
Lady Laura

 

[Arekushia]

effect size for Wilcoxon signed ranks test

Hi Laura,

I have just been trying to figure this out for myself, and it's actually pretty easy. SPSS gives you enough output for you to do it "manually" through the formula function in Excel, or with a calculator.

The SPSS output will give you a z score for each variable analysed. You then need to divide the z score by the square root of N (total number of participants in both groups). The result is the effect size.

Hope that helps!
Areku