Hi Andrew, 

I've just started using the NBS tool this week and have run into no issues with using it -- great tool and documentation!

I am writing because I am a little concerned about the results I am getting. I'm using 483 subjects (resting state) and relatively small correlation matrices for each subject with just 19 nodes. I am interested in simple group contrasts: [1,-1] and [-1,1]. When I run these contrasts I get one stable network for each contrast (meaning as I vary the threshold the network remains the same until the threshold is too stringent and the network doesn't survive). What I find odd is that it appears that these two networks include every possible edge in my matrix... in other words, one network is 18 nodes and 48 edges and the other is all 19 nodes and 123 edges. This adds up to the max # of edges, 171. I also ran the contrasts with the FDR option and found the same networks.

Am I wrong to think this is odd? As a new user, I would appreciate your thoughts on the likelihood of such an outcome.

Thanks much, 
Tessa

Hi Tessa,
I would be cautious if every possible edge in your matrix is found to be significant. This is possible, but quite unlikely in practice. It might be worth double checking you design matrices. You could also pick an arbitrary connection and check to see whether the connectivity values do in fact look significantly different between your two groups. What are the two groups you are comparing?

Andrew

Originally posted by Tessa Harrison:

Thanks very much for your quick response Andrew! Unfortunately, it turned out to be a bug in the script I had used to rename the CMs for loading into NBS. Everything is making more sense now. 

Thanks again,
Tessa