Hello!

First, some background:

I've completed a ROI-to-ROI analysis with resting state data specifying 23 ROIs from the CONN atlas. In the second-level analysis, I want to test the association between my primary variable of interest (V1) and connectivity values within the ROI connectivity matrix, while also adjusting for 2 covariates (age and sex). I therefore have 4 subject effects in my general linear model, intercept (all subjects), V1, age and sex. My between-subjects contrast is [0 1 0 0] to test the main effect of V1.

When I go to Results Explorer, I define my connectivity matrix as "targets are source ROIs only", which gives me a 24x24 connectivity matrix (23 ROIs + the effect of rest). I further threshold ROI-to-ROI connections by intensity and use FDR analysis-level correction and enable permutation tests. As an example, I select the ACC as my seed ROI to see its connectivity to the other ROIs of interest and the relationship between the connections and V1. The test statistic for the ACC, overall, is F(4)(11) = 7.95, p-FDP = 0.0029.

My question: How are the df [F(4)(11)] generated for this f-test? It is the same regardless of how many seed ROIs I choose to view.

Thank you so much for any clarity that you can provide!

Best,
Allie

I actually have a similar question. I am interested in identifying similarity and differences between controls and patients in the correlation between early life trauma severity with function connectivity and controlled age, sex, and headmotion. For within group correlation(such as the control group),  I selected control, patient, control_trauma, patient trauma, age, sex, headmotion, and input contrast as [0 0 1 0 0 0 0]. In the results explorer table, I got

Seed networks.Language.IFG F(29)(105) = 2.88 0 0.0035

Seed IFG tri r F(29)(105) = 2.86 0 0.0035

IFG tri r SubCalC T(133) = -4.06 0.0001 0.0136
IFG tri r Ver45 T(133) = 3.74 0.0003 0.0222

I can understand there was significant positive correlation between IFG trir-subcalc connectivity and trauma severity. How do I read the overall significance of IFG tri  or Seed networks.Language.IFG? Does this mean the connectivity of IFG with all other ROIs I had chosen showed significant correlation with trauma? 

Thanks,


Lihong

Hey Lihong,

For your F-tests, looking at the overall significance of IFG tri or Seed networks.Language.IFG, the significant p-value indicates that at least one connection is significantly related to trauma among all the connections tested. The step-down t-tests reveal which connections, specifically, are significantly related to trauma.

Hope this helps.

Allie