Dear Dario,
Guillaume is right. You have defined too many main factors. Simply define "subject" as main factor and "group" and "time" as interaction. There is also an example for exactly your design in the CAT12 manual on page 24.
Best,
Christian
On Wed, 18 Jan 2017 12:42:34 +0000, Guillaume Flandin <[log in to unmask]> wrote:
>Dear Dario,
>
>I don't use the TFCE toolbox so I cannot be specific regarding the error
>you get but in your flexible factorial design, you should only include
>'main effect' of subject and 'interaction' group x time. Testing for a
>group by time interaction in this design will be with an F-test [1 -1 -1 1].
>
>Best regards,
>Guillaume.
>
>
>On 18/01/17 10:08, Dario Zaremba wrote:
>> Dear SPM experts,
>>
>> we have a flexible factorial design on longitudinal CAT12 data in SPM12 and want to run TFCE toolbox on it. We get the following error:
>> Error using cg_tfce_estimate>correct_xX (line 1234)
>> Weird design was found that cannot be analyzed correctly.
>>
>> Our design is as follows:
>> - subject factor
>> - group factor (patients n=41 vs. controls n=14)
>> - time factor (baseline vs. 2yr-follow-up)
>> - covariate TIV
>> - covariate age
>>
>> We included following main effects & interactions
>> - main effect subject
>> - main effect group
>> - main effect time
>> - interaction group x time
>>
>> The contrast of interest is a F-contrast on the interaction group x time:
>> 0 0 0 0 1 -1 -1 1
>> 0 0 0 0 -1 1 1 -1
>>
>> Is it possible to estimate TFCE on this SPM.mat and specific contrast?
>>
>> Thank you very much in advance,
>> Dario
>>
>
>--
>Guillaume Flandin, PhD
>Wellcome Trust Centre for Neuroimaging
>University College London
>12 Queen Square
>London WC1N 3BG
|