I conducted a repeated measure ANOVA with 1 repeated factor with 3 levels. The result is significant F= 3.013, p = 0.5, sphericity assumed. I concluded, that these levels are different.
To compare levels I used Bonferroni/Dunns pairwise comparison, which show not a single significant differences, highest p-value = 0.63. If I check means and the graph the direction in with out a doubt, but I need to report it so this is not sufficient.
What kind of test you would recommend? I'm thinking about Bonferroni-Holms. Any opinions?
From this short description your omnibus test is significant but subsequent pairwise comparisions not significant. Just report this, why do you need to run a different test. If this is the scenario, it is straightforward enough and readers should understand.
Reply With Quote
Posting Permissions