Dear Badreddine,
> I want to compare two groups, subject vs controls using SPM99 and use
> Ancova to correct for global differences. The only option were I can
> enter two groups in SPM 99 using PET?SPECT models is:
>
> Compare populations: 1 scan/subject (2 sample t-test)
> were the only options are:
> proportional scaling OR no global normalization.
>
> In SPM 96 the option single subject: replication of condition we had the
> possibility to choose Ancova to compare groups.
>
> Any thoughts about how to run in SPM99 a comparison of two groups (eg.
> healthy vs disease) using Ancova to correct for global differences.
Andrew has configured the interface for PET design specification to
make it easy to use and to reflect our general appreciation of what
consitutes the most appopriate analyses. Because we now consider
within-subject AnCova as the best approach to global normalization (and
comparing groups of subjects with SPECT is best effected with scaling
normalization) I suspect he has constrained the options. You could use
a single-subject design with 2 conditions to get the AnCova option.
I hope this helps - Karl
|