Print

Print


Dear Pascal,

>I have a question regarding a conjunction analysis.
>I use SPM5 and I would like to do a RFX analysis of the conjunction 
>of 4 conditions over 20 subjects.
>So I would like to do a conjunction in each subject of the 4 
>conditions and than use this data of each
>of the 20 subjects in a RFX analysis. Do I have to use a flexible 
>factorial design ?  Any suggestions on
>how to set this up ?

I do not think you can do a (RFX) t-test on a conjunction.  This is 
because a conjunction is , itself a
statistical test. In other words ,you would not test a contrast 
conjunctions but test for a conjunction of contrasts.

I suspect the best thing to do it to create 4 contrasts for each 
subject and analyze these 20 x 4
contrasts with an ANOVA at the second (between-subject) level. You 
could then perform a conjunction
over the 4 condition-specific contrasts?

I hope this helps - Karl