Within Subjects comparisons based on Grouping variable

xyb3nce

I'm in a bit of a pickle. I ran an RMANOVA with one repeated measures factor (target stimulus type) and a between-subjects (grouping) factor (0 or 1). The RM factor has three separate variables, all signifying a mean response by subject for a distinct condition. (There were three conditions to which each subjects were exposed: target stimulus types 1, 2 and 3. Each variable (e.g., Variable 1) lists the mean answers to these target stimuli for each subject.)

Besides the main effects, I want to compare the mean scores for these (dependent) variables by group. Let's say, I expect that the mean scores for target stimulus type 1 will be lower than type 2 in Group 1.

Since post-hoc tests can't be done (I have a couple of pre-recorded theory-driven hypotheses), I figure I need to do contrasts. However, the JASP UI doesn't support this.

Also, an important issue to mention is that besides the between-subjects effect, the within-subjects effect and the interaction were insignificant. Nevertheless, I still need to do the comparisons. The easy way out would be to run separate t-tests or two RMANOVAs with just members of one group -- this, however, would not work out statistically, and I need to handle all my data and predictions in one model (preferably).

Could anyone help me out?

patc3

There's several contrasts/post-hoc comparisons that might be of interest to you, notably:

• under Contrasts, you can pick "custom" and enter your own contrast coefficients (I think this is what you're looking for)

• under Post Hoc Tests you can obtain post-hoc pairwise comparisons (though I understand this might not be what you're looking for)

• under "Marginal Means" you can obtain estimated marginal means

• under Simple Main Effects you can obtain comparisons within a given level of the other factor (though in your case you say the interaction is not significant, so I don't think this is as much of interest to you here)