Hi,

I have a 2 (between: treatment vs. control) x 2 (within: time 1 vs.time 2) x 2 (within: target vs. other). Participants in each group are measured at 2 time points, and at each time point I get ratings of a target and ratings of others. I am interested in the 3-way interaction, and decomposing it to see the two way (fully within) interaction for each group. In other words my research question is: Does the difference in ratings of the target vs. others change across time points, and is this difference in differences different for the two groups? Although I will also need to test significant differences in target vs. other ratings at each level of the other variables. I preferably want to do this in G*power or a similar interface. I do not have adequate past research to get the necessary input for a sim to analyze power that way.

Can anyone help with an a priori power analysis to determine sample size for this design? I've tried F-test: ANOVA: repeated measures in G*power, but the sample sizes for the within-between interaction seem low, and I'm not sure putting 2 measurements (since it's a within-subjects difference of differences) is entirely correct.

Thank you in advance!