Thanks for pointing that out. I have updated the code to include the variable name meanRTacrossruns instead of just RT.
The data is from an n-back task. For each subject, we have mean reaction time (RT) for each level of the task (0back, 1back, 2back, 3back), and a variability map for activation for each subject and level of the task. We are trying to examine the relation between within-subject variability in BOLD signal and mean RT for the different stimulus types, as well as how mean RT is associated with the difference in variability between different levels of the stimulus type factor (i.e., 3back vs. 0back).
We are thinking that the error (that still happens after updating the "RT") is related to the data table...
Here is a larger portion of the data table:
Subj StimType meanRTacrossrun meanACCacrossrun InputFile \
1 0back 616.9448 0.9732 '0back_variability_map+tlrc' \
1 1back X X '1back_variability_map+tlrc' \
1 2back X X '2back_variability_map+tlrc' \
1 3back X X '3back_variability_map+tlrc' \
...
48 0back X X '0back_variability_map+tlrc' \
...
But, it could be that the -ranEff or -SS_type that we have chosen are not logical with what we are trying to do? We structured our script based on example 2 on this webpage: [
afni.nimh.nih.gov]
In our data, meanRTacrossruns is a within-subject quantitative variable, and generally, the meanRTacrossruns value increases at each increasing level of the n-back task.
We believe that marginal sum of squares (the default, 3) is best, but are certainly open to suggestions.
Thanks for your help.