I am new to statistics and found XLToolbox. I was able to run ANOVA but the posthos/multiple tab is grayed. Why is that? I have five specific groups to run
I would appreciate any information on this topic.
Thanks
The posthoc tab will only be enabled if the differences between the groups are significant (p < 0.05). With groups whose means do not differ statistically significantly, the posthoc analysis does not make sense.
Kind regards
Daniel
Hello Daniel, very good tool indeed (I very appreciate mainly the chart export capabilities).
I have one objection about the post-hoc test, from the statistical point of view Tukey test (and virtually all other post hoc test, except the protected Fisher LSD) can be performed even if the overall ANOVA is nonsifnificant; they are “stand alone” procedures that can be used separately from ANOVA. One reason for this is that ANOVA and post hoc tests results don’t have to agree exactly and post hoc comparisons could be significant even if ANOVA is not. And second practical thing is that it could be useful to perform exploratory comparisons. I will appreciate such possibility.
Good book about this topic where it is explained more rigorously is for example in Hsu J.C. Multiple Comparisons, Theory and methods. Chapman & Hall, 1996
or very briefly in Statistical Misconceptions by Schuyler W. Huck.
Regards
Peter
Hi Peter,
I appreciate this feedback. The book is currently on loan from our university library; I’ll have a look at it some time.
It’s very easy to programmatically enable post-hoc tests in the Toolbox even if the ANOVA did not report statistical significance; it should be fixed in the next version.
Kind regards
Daniel
Version 6.52 is out now, it has post-hoc tests constitutively enabled.
Daniel