For unbiased DCorr, a t-distribution approximation is used to estimate the p-value. MGC, unbiased DCorr, and biased DCorr should have a simple permutation test to estimate each of their p-values.
The proposal here is to not only remove the t approximation, but to have the abstract p-value only be a simple permutation test, and let it be inherited by MGC, unbiased DCorr, and biased DCorr. For other tests, their methods can override the inherited p-value.
For unbiased DCorr, a t-distribution approximation is used to estimate the p-value. MGC, unbiased DCorr, and biased DCorr should have a simple permutation test to estimate each of their p-values.
The proposal here is to not only remove the t approximation, but to have the abstract p-value only be a simple permutation test, and let it be inherited by MGC, unbiased DCorr, and biased DCorr. For other tests, their methods can override the inherited p-value.