After calculating the treatment effect under all possible treatment assignments, the next step is to: Count the number of cases when the treatment effect is at least as large as the observed case Count the number of cases in which the absolute value of the simulated treatment effect is less than the treatment effect actually observed, divide by the total number of cases, this is the relevant p-value Average the p-values from each of the possible treatment effects Count the number of cases in which the absolute value of the simulated treatment effect is greater than or equal to than the treatment effect actually observed, divide by the total number of cases, this is the relevant p-value