[Eeglablist] Including interaction with condition with two datasets per participant

[Agata Dymarska]

Dear Cyril,

Thank you for you suggestion. I created a new STUDY as outlines, but I am still having trouble with the analysis which includes two datasets/sessions per participant.
After creating .daterp files, in the 1st level analysis, the programme produces errors: "couldn't find a direct match between .set and .daterp". It appears that the comparison of the two is not made based on participant name but on the order in which they are loaded, and even though they are loaded in chronological order, the comparisons are made between dataset 1-2, 1-3, 1-4, then 2-3, 2-4, etc. I'm not sure why that happens, because I even ensured that the subject names and dataset names are the same so there is no doubt which data comes from which subject. The error produced is pasted below:


generating temporary files, pulling relevant trials ...
pulling trials for sub-Hs481a-word_selection_receiver_renamed_ses-1_design1.set ...
Warning: couldn't find a direct match between .set and .daterp, loading C:\Users\eeg-analysis2\Study New\Strangers\Hs481a-word_selection_receiver_renamed_ses-01.daterp
making up statistical model for sub-Hs481a-word_selection_receiver_renamed_ses-1_design1.set ...
pulling trials for sub-Hs481a-word_selection_receiver_renamed_ses-2_design1.set ...
Warning: couldn't find a direct match between .set and .daterp, loading C:\Users\eeg-analysis2\Study New\Couples\Hs481a-word_selection_receiver_renamed_ses-02.daterp
making up statistical model for sub-Hs481a-word_selection_receiver_renamed_ses-2_design1.set ...
pulling trials for sub-Hs481b-word_selection_receiver_renamed_ses-1_design1.set ...


I'm not even sure whether this error has anything to do with the comparison of the two sessions or whether it is an earlier stage of an individual dataset analysis that is not working.
Can you please advise on this?

Thank you,
Agata



Hi Agata

alternatively you can set the two sessions in STUDY. Each session with
language and valence. STUDY/LIMO will automatically compute the
difference between sessions, ie you end up with contrasts language
difference and valence difference. This allows setting up and ANOVA 2x2
on those difference (of course each effect now reflects the
difference/interaction with relationship which is like having an ANOVA
with only interaction terms). The main differences with merging are (1)
does not assume the same variance between sessions (2) but does not
allow testing main effects across sessions (although you can do the 2x2
in each session to check consistency of results).

Cyril