Hi Shary,
It might be explicitly mentioned, but you could also just explain
that as the threshold |t_crit| gets sufficiently high, the nominal
p-value provided by SPM becomes a reasonable (at least for a very
fine lattice) approximation for the probability of one or more
voxels above |t_crit| (i.e., the Poisson distribution of the Euler
characteristic of a random field argument). And since this is a
probability of one or more voxels being > |t_crit| (and not ">
|t_crit| or < -|t_crit|"), it is a one-tailed probability. Then I'd
cite the seminal papers that present the random field results.
Hope this helps.
Eric
Quoting "[log in to unmask]" <[log in to unmask]>:
> Dear colleagues,
>
> Is there any reference in which one tailed t-test in SPM is
> explained at least in one sentence? I need to refer to it in our
> paper and can't find one.
>
> Thanks,
> Shary
>
>
|