spss >> Post hoc test for Friedman ANOVA

by eggy_gee » Fri, 20 Apr 2007 11:52:32 GMT

Hi. Anyone know if there is a way to do a post-hoc test for the
Friedman's ANOVA. I have sig. differences over 3 levels but need to
know which level is diff. from each other. i.e like the contrasts post-
hoc for the one-way ANOVA for repeated measures. Any ideas?


spss >> Post hoc test for Friedman ANOVA

by Bruce Weaver » Fri, 20 Apr 2007 19:19:57 GMT



The book by Siegel & Castellan describes a large-sample z-test for
pair-wise contrasts, IIRC. But I expect you would get pretty similar
results using the Wilcoxon signed ranks test.

--
Bruce Weaver
XXXX@XXXXX.COM
www.angelfire.com/wv/bwhomedir

spss >> Post hoc test for Friedman ANOVA

by John Reece » Sun, 29 Apr 2007 19:30:46 GMT

Just do a series of Wilcoxon paired-samples tests with Bonferroni adjusted
significance levels, or use a sequential rejection method for slightly more
power. That does the job. Of course, if you're going to do this, you have to
ask the question of why you are doing the Friedman test in the first place.

John

spss >> Post hoc test for Friedman ANOVA

by Bruce Weaver » Sun, 29 Apr 2007 22:36:14 GMT


John, I don't think there is any need to do a Bonferroni adjustment if
the pairwise contrasts are contingent on first obtaining a significant
omnibus test--i.e., if a Fisher's LSD type approach is applied. When
the variable has only 3 levels, Fisher's LSD controls family-wise alpha
at the per contrast level. Dave Howell discusses this in his textbook
(Statistical Methods for Psychology).

Cheers,
Bruce

--
Bruce Weaver
XXXX@XXXXX.COM
www.angelfire.com/wv/bwhomedir