I’m an undergrad, currently in my 3rd year doing my dissertation.
I have 2 groups (time and method), and all participants are in both groups, so it’s a 2-way repeated measures ANOVA.
The time variable has 4 levels (5, 10, 15 and 20 mins), and the method variable has 2 levels (A and B).
I am only wanting to compare A and B at each time point, but after doing the RM ANOVA on SPSS, I don’t know which follow up tests to use. I know that if I do post-hocs, then it compares everything, and therefore decreases the chance of getting significant results. I’ve also had a look at standard contrasts, but I can’t see one which would suit my situation.
To make things easier, I want to compare the following:-
(1,1) vs (2,1)
(1,2) vs (2,2)
(1,3) vs (2,3)
(1,4) vs (2,4)
with the first number representing the method, and the second number the time.
There’s not really a way around the decrease in power. You will have to do multiple comparisons.
If it were me, I would take the difference between methods for each participant (A-B or B-A) and then regress that on time. That would quantify the size of the linear time effect. Alternatively, you could simply do 4 dependent t-tests for each time point with a Bonferroni correction (alpha = 0.05/4 = 0.0125 or alpha = 0.10/4 = 0.025 for trends).
I’m sure there are better solutions, but that’s a quick and dirty way of doing it (and probably what I would do for my “main test”.
If the effect is nonlinear (plot the data), you could also regress the difference on the appropriate time transformation (if it’s nonlinear, it’s probably log-linear.