Dear Alex, Torben, Philipp, Ged,
Thanks for all of your replies. I should have known that there is no way
around what I wanted to avoid: trying it myself.
Torben was right in that there are a few scaling and offset issues, but that
templates are very similar (I will not post my results as I don't have time
to make a proper presentation).
Ged (and Torben off-list) made a really useful and relevant point about
introducing a nuisance covariate for the two "scanners", which, by the way,
in my case did NOT make the results less significant.
Thanks again,
Best wishes,
Helmut
----- Original Message -----
From: "Ged Ridgway" <[log in to unmask]>
To: <[log in to unmask]>
Cc: <[log in to unmask]>
Sent: Monday, July 02, 2007 2:03 PM
Subject: Re: [SPM] SPM2 vs. SPM5 normalized template
> Hi Helmut,
>
> I agree with all of Philipp's and Alex's comments; one minor thing I'd
> mention...
>
>> ***If*** 1=yes and 2=satisfactory, and if there are no format problems,
>> then off the top of my head I can’t see much of a problem (as shocking as
>> it sounds). Knowing you, you will of course check for any systematic
>> differences between the two preprocessing pathways… and I hope the two
>> groups don’t correspond to patients versus controls etc.
>
> Maybe it would be worth including a covariate to indicate processing
> pathway? Rather like the suggestion here
>
> http://en.wikibooks.org/wiki/SPM-VBM#Data_from_different_scanners
>
> Then your results will be (correctly) less significant the more correlated
> the processing pathway indicator variable is to the effect of interest.
>
> Best,
> Ged
>
|