Hi Erin,
Looking at spm_list.m, it gets the (equivalent) Z-scores back from the
p-values with the obvious formula:
Ze = spm_invNcdf(1 - Pz)
spm_t2z appears to have some quite complicated code in it, which one
might assume would be more accurate... I'm afraid I can't help with
the explanation for not using it. Possibly it's motivated by the fact
that different p-values (derived from e.g. cluster size) don't have an
equivalent t-value to use with spm_t2z; using the simpler (less
accurate?) formula is then perhaps more consistent... but I'm guessing.
Best,
Ged.
Mazerolle, Erin wrote:
> Hi list,
> I have a quick question about an old post I found about changing t
> scores to z scores
>
> This post:
> http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind04&L=SPM&P=R376942&I=-3
>
> Suggests using this command for SPM2:
>
> z_score = spm_t2z(t_score, degrees_of_freedom)
>
> t_score: comes from your SPM map
> degrees_of_freedom: at the bottom right of the SPM map
>
>
> When I try this in SPM for large t-scores, I get a different Z score
> than the table that is created when I hit the Volume button. For example
>
> T = 9.84
> Z in table = 7.79
> Z from command above = 7.33
>
>
> I am guessing that SPM5 uses a different m file to get the z scores that
> come out in the table. Does anyone know if that's true, and if so, what
> file it is?
>
> Thanks!
> Erin
>
|