As you can see the screen shot below (point no.4) that I read from here. I would like to know which function and parameter that should adjust in order to retain the size of largest cluster permutation based on the sum of t-test value or by average of t-test value of a cluster ?
Using t_power=1 you will get a sum of t values within a cluster. If you want the average t value within the cluster you can just take the cluster “score” and divide it by the number of points. Both of these should match what you get when just using the cluster indices/mask to operate on the t values directly (via sum or mean or whatever).
