My experiment has 5 conditions (let's say Treatment A, B, C, D and E). There are 12 measurements (let's say Measurement 1-12). I'm running 12 ANOVAs on the measurements, I assume this means I should run with some multiple comparison correction, e.g., Bonferroni with p=0.05/12. Once I've determined which ANOVAs are significant, I'll run post-hoc Tukey HSDs on each of those. However, given that I am running up to 12 post-hoc Tukey HSDs, do I have to do multiple comparison correction beyond what the Tukey HSD already does (per ANOVA)? Do I need to divide the significance values of the Tukey HSDs by the number of Tukey HSDs I run?

  • $\begingroup$ If they are 12 repeated measurements of the same response variable, then probably yes. If they are just 12 different responses, then I think most people would just consider it 12 different families of tests. But I'm interested in others' comments. $\endgroup$ – rvl Jan 20 '15 at 3:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.