Corrected significance level in Tukey HSD post-hocs?

#1
Dear all,

I have a question that is really bugging me:

I'm testing if seed set in one plant species is impacted by the deposition of pollen from a different plant. To do this I prepared 7 different mixtures (7 treatments) of species' own pollen and foreign pollen and applied those to receptive flowers. A few weeks later I counted the seeds produced by each flower. The one-way ANOVA testing for differences in seed set between treatments comes out significant. I'm following up with TUKEY'S post-hoc test, which shows several significant comparisons between treatments.

Now my question:
Is the significance level already adjusted (corrected) for the big number (I think it's 21 if I'm not mistaken) of comparisons I am making or do I have to recalculate alpha. If the latter is true probably all my significant results will be down the drain given that I had so many treatments.


Many thanks for your help.
I really appreciate it.

Best wishes,

crazy_man
 

Dr.D

New Member
#2
Dont worry crazy man. If you have conducted a Tukey test(which is pretty good with guarding against Type 1 error), you have nothing to worry about. The 21 comparisons (given that you have 1 dependent variable) will be compared against an adjusted alpha based on Tukey's calculation. So those non-significant comparisons suggest that they did not meet the adjusted alpha (that is, they exceeded it), and the significant ones indicate that those comparisons fell below the newly adjusted alpha. So dont worry..no need to recalculate alpha..Tukey automatically did that with the 21 comparisons.
:yup: