Just one brief comment...
At 11:18 AM +0000 2/12/02, Will Penny wrote:
>Eini Martikainen wrote:
>
>> I sent this question last week, but didn't get any answer, so here we go
>> again:
>>
>> Dear List,
>>
>> I'm analysing fMRI-data and I'd like to perform rfx-analyze. I'm confused
>> which basis function I should use. I have three groups each with several
>> conditions. If I want to compare three/four conditions in ONE group, how
>> should I do that? I understood I can't use Anova because the data are not
>> spherical. Should I use some PET/SPECT model? Which one?
>>
>
>In SPM-99, if you assume sphericity you can implement this ANOVA using
>one of the PET models.
>
>There is currently no way to get round the sphericity assumption.
>
>In the development version of SPM we are developing an EM/REML-based
>algorithm for coping with non-sphericity. This should be released later this
>year.
>
>
>>
>> Also if I want to compare just two conditions from the same subject should
>> I use two-sample t-test or paired t-test? The conditions are not exactly
>> paired (like before/after something).
>>
>
>Then I think the two-sample test is the most appropriate.
If these conditions come from the same subject then the paired t-test
is still appropriate. The observations are not independent (as
assumed by the two sample t-test) if they come from the same
individual. You might as well allow the within subject variability
to play its part appropriately.
>Best wishes,
>
>Will.
>
>>
>> Thank you!
>>
>> Eini Martikainen
>> Department of Applied Physics
>> University of Kuopio
>> Finland
>
>--
>William D. Penny
>Wellcome Department of Imaging Neuroscience
>University College London
>12 Queen Square
>London WC1N 3BG
>
>Tel: 020 7833 7478
>FAX: 020 7813 1420
>Email: [log in to unmask]
>URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|