The pvalue to never be exactly 0 in permutations (because observed data belong to permutation set). There are multiple ways to do it
Obv. the first one is the simplest one, (does it reduce to adding 1/(nperm+1) to your p-value ?)
But isn't there a julia library for permutation tests? That would be even more versatile. I am also asking, because there are several methods to try to estimate the tails of permutation distributions, in order to reduce the number of permutations needed. Maybe those are already implemented there as well
The pvalue to never be exactly 0 in permutations (because observed data belong to permutation set). There are multiple ways to do it
Obv. the first one is the simplest one, (does it reduce to adding 1/(nperm+1) to your p-value ?)
But isn't there a julia library for permutation tests? That would be even more versatile. I am also asking, because there are several methods to try to estimate the tails of permutation distributions, in order to reduce the number of permutations needed. Maybe those are already implemented there as well
_Originally posted by @behinger in https://github.com/palday/MixedModelsPermutations.jl/pull/12#discussion_r577126516_