Dear Victor
I don't have SPM on this computer and I don't remember the line number
in spm_spm. But you could be right that the F-contrast for the effects
of interst is actually specified via SPM.xX.iG and SPM.xX.iB in which
case you should be OK. I will look that up soon if you like.
Best regards
Torben
Victor Vorobyev wrote:
> Dear Torben,
>
> thanks for your reply!
>
> The effect of interest looked ok in sense that only relevant column
> were marked as "of interest" above my design matrix.
>
> Eventually, in order to put my mind in peace, I performed two
> single-subject analyses, explicitly specifying only SPM.xX.iG in the
> first one, and both SPM.xX.iG and SPM.xX.iC in the second one.
> Both analyses included realignment parameters and AR(1) correction.
>
> I found no difference between results of those two analyses when
> looking at the result tables as provided by F(effect of interest) and
> some T-contrasts.
>
> So, it looks for me like it is rather SPM.xX.iG than SPM.xX.iC that
> finally influences what is "of interest".
> At least in my particular case specifying SPM.xX.iC happen to be
> redundant.
> It makes me happy since I do not need to re-process my data :)
>
> Best regards,
>
> Victor
>
>
>> Most likely yes. You can check it yourself by selecting the effects of
>> interest in the results section.
>>
>>
>> best
>> torben
>>
>>
>> Victor Vorobyev wrote:
>>
>>> Dear SPM users,
>>>
>>> could anyone tell me how the effect of interest is defined in SPM2:
>>> is it the one specified by SPM.xX.iC, or is it everything that is NOT
>>> specified by SPM.xX.iB and SPM.xX.iG ?
>>>
>>> The reason I am asking is that I used batch mode to process my group
>>> fMRI data as a fixed-effect model with realignment parameters included
>>> as regressors of no interest. Therefore I specified SPM.xX.iG to mark
>>> the effect of no interest, but somehow forgot to change SPM.xX.iC (that
>>> now contain all columns including those specified in SPM.xX.iG).
>>>
>>> Unfortunately I noticed that my mistake only after I had completed all
>>> the long-lasting computations, so I am wondering did I make a critical
>>> mistake or not? Specifically, did it lead to wrong AR(1) estimation?
>>>
>>> thank you!
>>>
>>> Victor
>>>
>>>
>>> *********************************
>>> Victor Vorobyev, Ph.D.
>>> Centre for Cognitive Neuroscience
>>> University of Turku
>>> Finland
>>> *********************************
>>
>>
>>
>
>
|