Hello!

My question is whether it is possible to compare differences in Brainstorm. For example, there is a deviant-minus-standard difference (A1-B1) calculated for one deviant-standard condition and the same difference for a different deviant-standard condition (A2-B2). Is it possible to compare statistically the two differences (compare (A1-B1) to (A2-B2)) with a t-test, for example? This is a within-subject comparison. The tutorials only list comaprisons between two stimuli, e.g., A1 to B1.

Kind regards

With a non-parametric t-test you can compare sets of samples of any type of data. Why not these differences.

Note that: `(A1-B1)-(A2-B2) = (A1-A2)-(B1-B2)`

Whether it make sense to subtract the differences of baselines from the difference of active states makes sense or not in your experiment is your responsibility.

Thank you very much for the reply!

Thank you for the additional note as well. I did not make this explicit, but just in case another person finds this post useful - in our case the A and B cases are not baselines, but two different deviations from a hypothetical baseline. This is common in the mismatch negativity research, in which it is not always possible to have a baseline ("control") condition - this depends a lot on the type of stimuli used in the experiment. In our study we couldn't use one. This, of course, influences the conclusions, and we (at least try to) account for it.

Edit (because I am really bad at expressing thoughts): This comparison should virtually be looking for a specific interaction in a 2x2 ANOVA (A1 and B1, A2 and B2), and should be analogous to a difference wave comparison in ERP studies - whether one of the differences is larger than the other. This might be incorrect in case of source localization.