t test on Gc path coefficients

Submitted by ma.shalchy on

Dear Chao-Gan

Hello
Thank you so much for your time. I have read your paper on  GCA and  need your help. After obtaining the path coefficients how could I perform a one sample ttest since I do not know the population mean?
Secondly, what is used to represent the connection strength between areas? t value of the related path coefficients?

Many thanks,
mahsa

YAN Chao-Gan

Tue, 12/30/2014 - 15:22

Permalink

Hi Mahsa,

For signed path coefficients, you can perform t-tests on them directly, as in Zang, Z.X., Yan, C.G., Dong, Z.Y., Huang, J., Zang, Y.F., 2012. Granger causality analysis implementation on MATLAB: A graphic user interface toolkit for fMRI data processing. J Neurosci Methods 203, 418-426.

The t-values can be used to represent the strength.

Best,

Chao-Gan

ma.shalchy

Wed, 12/31/2014 - 04:39

Permalink

Dear Chao-Gan

Thank you so much for replying. I have difficulty interpreting the coefficients. i.e if tvalue of ROI1 vs ROI2 is -0.8 and ROI2 vs ROI1 coefficient is +0.03 how should I assess the overall effective connectivity? In papers I do not see a negative path coefficient! Should I reverse the direction in negative coefficients? Should I find a net value by summing the negative and posituve values? What if both the values are positve?

Many thanks in advance,
mahsa

Hi Mahsa,

I think you can intepret the positive and negative values as excitatory and inhibitory. I don't think you need to sum the negative and positive values, as they are in two different directions, ROI1->ROI2; ROI2->ROI1. You can imaging that between two regions, they can have bi-directional connections.

Best,

Chao-Gan

Dear Chao-Gan

Thank you so very much for your detailed explanation. One more question about the one sample t-test...
Should I use a one-tailed t-test or two-tailed one?

Many thanks in advence.

 

Forums
DPABI/DPABISurf/DPARSF