Hi Owen,
>Hi
>I am currently analysing a task and have reached the second level,
>unfortunately the task design could be coded as a 2x3x3x2 anova, but neither
>my brain nor my computer could cope coding all the contrasts for the main
>effects and interactions, even with Rik's magic recipe.
>
>
oh that's gonna be easy to analyze
>so i have question that i pray someone can help with:
>
>does it make sense to use the output(contrast image) of a paired t-test in
>an second Anova?
>
>
well, I'm not sure sure but I don't think so (if you speak about using
results coming from a paired t test at the second level)
>is there anyway to break down such a beast into bite-sized chunks that mear
>mortals such as myself can get our teeth into
>
>
I would say that you can try a canonical decomposition - I think you can
start to look at that in SPSS or Statistica and then try to reproduce in
SPM ..
Maybe you can also start be reducing the design in 2x3x3 computing a
difference at the first level ?
good luck
cyril
|