I have two groups of participants ($A$ and $B$) who have been measured at two points in time ($T_0$ and $T_1$) for a continuous variable (a biomarker in the blood).
If I wanted to know if the change over time in the biomarker ($T_1-T_0$) was different between the group $A$ and $B$ I'd compare the change in each group with a $t$-test or similar. However, I want to know if any of the four means of $T_0(A)$, $T_1(A)$, $T_0(B)$, $T_1(B)$ are different from the others. If there were just four different groups I'd use a one way ANOVA, but the groups are related two-and-two. What can I use instead?
Best Answer
Use a two-factor (Factor 1 = group and Factor 2 = time) ANOVA with an interaction term between your two factors. Then conduct a post-hoc Tukey Honest Signficant Difference test to assess the mean differences. The interaction term in the ANOVA will allow you to compare all six mean differences with the post-hoc test.