We strive to make the models in SPM more accurate with each release. Better
models should lead to findings that are closer to the truth. In general, if
you use a different model of your data (whether for pre-processing, GLM or
DCM etc), then you should expect to find different results.
The differences you see are probably largely a result of this change, which
was introduced in SPM2:
SPM2 now uses restricted maximum likelihood REML estimates of variance
components (pooled over voxels) to allow departures from i.i.d. assumptions.
The ensuing non-sphericity estimates are then used to form ML estimates using
weighted least squares (WLS). This increases the scope of data that can be
analysed with SPM and provides more efficient and precise [hyper-]parameter
estimates.
As a general rule, we suggest using the most recent version of SPM (maybe
waiting a month or so after the initial release, so that any glitches can be
ironed out). Reviewers probably (I hope) look down on submitted manuscripts
that use old SPM versions, although they may have some historical value.
Best regards,
-John
On Friday 13 March 2009 13:51, Julie E McEntee wrote:
> Hello List,
>
> The SPM website states, "SPM5 is a major update to the SPM software,
> containing substantial theoretical, algorithmic, structural and interface
> enhancements over previous versions." I have followed the link to the
> 'document' that "describes the new features," but it is not very specific,
> especially in reference to fMRI data analysis.
>
> Can anyone direct me to a source/paper that outlines the differences
> between the versions of SPM? In particular, our lab is re-analyzing data
> previously analyzed using SPM99 and getting very different results. I was
> hoping that information as to the changes made between versions of the
> program may explain the results that we are getting using SPM5.
>
> Thank-you for your help,
>
> Julie
>
> Julie E. McEntee, M.A., C.C.R.P.
> Senior Research Program Coordinator
> Department of Psychiatry- Neuroimaging
> Johns Hopkins University School of Medicine
> 600 N. Wolfe St./ Phipps 300
> Baltimore, MD 21287
> Phone: 410-502-0468
> Fax: 410-614-3676
|