Hi Kevin,
Take a look at the “uncorrected” option for thresholding (see Post-stats tab in Feat). Once you collect enough data to use corrected thresholds, you likely won’t want to lower the cluster-forming threshold below z = 2.3 (or even 3.1; http://www.ncbi.nlm.nih.gov/pubmed/24412399).
Cheers,
David
On Sep 10, 2014, at 11:21 AM, Kevin Japardi <[log in to unmask]> wrote:
> Hello all,
>
> There has been something confusing me for some time now.
>
> For some of our projects, we've been running our group analyses in FSL Feat, with varied regions of activation. In some instances, people in our lab recommended to run a set of "uncorrected" analysis gfeats, using a cluster p threshold == 1 (z-stat of 1.7). (Such that our results aren't corrected for multiple comparisons.)
>
> Can someone clarify how this is "uncorrected"? Aren't I just increasing our cluster threshold from (normally) 0.05 to 1, thus probably resulting in very little activation due to regions of activations not surpassing the cluster threshold?
>
> If this isn't a true way of running an uncorrected analysis, any thoughts on how to do so?
>
> Thank you,
> Kevin Japardi
|