 |
 |
Hi On 15 Oct 2014, at 11:45, Jun Miyata <[log in to unmask]> wrote: > Dear FSLNets experts (especially Prof. S. Smith), > > I have done dual regression and FSLNets on my data, and found significant group difference for partial correlation between IC A and IC B with FSLNets. I'm interested in the causal directionality between these two ICs, and checked the netmat11. However I'm not sure what is the correct way to analyze netmat11. I appreciate if you could tell me if my procedures are correct. I did the following way: > > 1. I identified the A->B and B->A in netmat11. As I entered 3 ICs (A, B, and C. This order is also the order of IC number) into FSLNets, netmat11 is (number of subjects) x 9 matrix. A->B is column 2 and B->A is column 4. > 2. I performed one sample t tests for column2 and column4. > 3. As column2 is significantly larger than 0 (and column4 less than 0), I conclude the directionality is A->B. To be quite sure, I would explicitly extract sets of two timeseries from your data and feed them into pwling - and see the header at the top of Aapo's code (pwlin.m) for an explanation of directionality. See nets_netmats.m for how to call pwling. To be doubly sure, you might create a very simple pairwise timeseries simulation to verify the above. > In addition, if I get non-significant results for this t test, can I say "the directionality is bi-directional"? not exactly - a more accurate version of that statement is that you don't have strong evidence for unidirectional causality. Cheers. > > Thanks in advance, > > Jun --------------------------------------------------------------------------- Stephen M. Smith, Professor of Biomedical Engineering Associate Director, Oxford University FMRIB Centre FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK +44 (0) 1865 222726 (fax 222717) [log in to unmask] http://www.fmrib.ox.ac.uk/~steve --------------------------------------------------------------------------- Stop the cultural destruction of Tibet