What test is better for comparing theese means?

Question

I have many groups of four replicates of groups comming from two sets of treatments. I found interaction between the two varibles (treatments) via a 2 way ANOVA. Now a need to see which means are different and which ones are the same.... Should i use Tukey's HSD or Fisher's LSD? Why can't i do a Student's T test?

Answer 1

You wouldn't use a t test because with the number of comparisons you have going on, you're running the risk of finding a correlation that appears to be significant but it's only by chance. And you wouldn't use Fisher's LSD because that assumes that you have the same variance across all groups, which you can't assume. So that leaves Tukey's HSD.