For ANOVA with post hoc multiple comparisions as well as a number of planned contrast tests, shouldn't the p values be corrected by the total number of tests for both ANOVA's post hoc test and the contasts?
right now , it seemed the ps for contrast are not corrected, and the post hoc comparisons are corrected by the number of "Post Hoc Tests"
correction of multiple ANOVA contrast test
Re: correction of multiple ANOVA contrast test
i think what we're doing here is considered "standard", but let us know if that isn't the case.
jonathon
jonathon
Re: correction of multiple ANOVA contrast test
If, for example, you were inspecting an issue of a journal, with each article reporting between three and seven p values (a total of 40 p values), and if you wanted the probability of making a least one Type I error (in the set of 40 test) to be no greater than 0.05, you could apply the Bonferroni correction for the set of 40 tests. But it seems a bit hopeless, no?