Hi,

if I want to test for one group of subjects which edges from my cPPI matrix for a contrast between 2 conditions of a task are different from 0, how do I do that in the NBS toolbox? If I fill in that I want to One Sample test with a threshold of 0 I get 'stop: threshold not found'. Outside a NBS context I would test for all edges whether they are significantly different from 0 with a one sample t-test and then apply an FDR corrected alpha for the p-values. 

Many thanks,
Selma

Hi Selma, 

The threshold of 0 does not correspond to testing if the mean is different from 0. With the one-sample t-test, the null hypothesis is always a mean of 0, irrespective of the threshold value. The threshold value allow you to set a minimum meaningful effect size and is analogous to the cluster forming threshold in cluster stats. 

I think that you may be confusing the threshold value with the hypothesis test of a mean of 0. They are different things. 

Perhaps try a t-stat threshold of 2.5 or 3. A threshold of 0 is not valid. 

best,
Andrew
 
Originally posted by Selma Lugtmeijer:

Hi,

if the threshold for a one sample test is like an effect size and you suggest a threshold of 2.5 or 3, what is the scale? I want to get the connections with a medium to large effect.

Kind regards,
Selma

Hi Selma, 

for the one-sample t-test, the threshold is in the same units as your connectivity measurement. So if your connectivity measurements range between 0 and 10 (for example), a threshold of 2.5 will consider connections with connectivity strengths above 2.5.

Note that for the two-sample t-test and F-test, the threshold is applied to the t-statistic and F-statistic, respectively. 

Andrew

Originally posted by Selma Lugtmeijer: