Alternatively a p-value can be found by using a standard permutation test, where the p-value is found by (k0+1)/(k+1), and k0 is the number of the k permutations that are at least as extreme as x. In such a testing framework, the permuting routine can be stopped if the estimated p-value (and its precision) reaches a certain level; e.g. if the p-value, minus three times the estimated standard deviation of the p-value, is above the significance threshold. Such a permutation strategy may be as fast as the approximation strategy, since fewer than 1000 permutations are needed to reject the hypothesis of association in many cases.
