We have resting state data from 30 participants for which we want to examine Graph Theory metrics in a network of 18 language-related ROIs that we have defined. 

In order to determine the first-level threshold we should use we produced a graph of the global and local efficiencies as a function of the threshold, by leaving the threshold box empty (as suggested in this post). A screenshot is attached for both the cost and correlations.

As the legend is illegible I would like to ensure that the solid line represents my data, the dotted line represents the Lattice, and the dashed line represents the Random. Is this correct?

If I understand correctly, the graph on the right describes the calculation of the formula (GE_data-GE_lattice+LE_data-LE_random), and the chosen threshold should maximize this value. Can you help me understand which threshold value answers that according to my data? 

1. What does it mean about my data that the graph is negative, and the maximum value is close to zero? would you still recommend using this point as the threshold?

2. I see that many studies use a range of thresholds. How should I use this graph to determine the range of thresholds to use?

3. When using a range of thresholds how should I report the results across the different thresholds? Is it ok to average Graph Theory metrics (e.g. global and local efficiency) across thresholds?

4. Is there a way to implement a range of thresholds in the conn GUI? 

I really appreciate any help you can provide.