Randomization Sign-Flip p-Value

Problem

Run a two-sided sign-flip randomization test on paired differences using n_perm sign vectors drawn from the seeded Lehmer RNG (an even draw keeps the sign, an odd draw flips it). Return (count + 1)/(n_perm + 1), where count counts permuted absolute means at least as large as the observed absolute mean; return 1.0 for empty input or n_perm <= 0.

Hint

Under the null each pair's sign is exchangeable.

Test cases

case 1

in: [[1,2,3,4,5],100,42]

expected: 0.0594059405940594

case 2

in: [[1,-1,1,-1],100,7]

expected: 1

case 3

in: [[2,2,2],50,1]

expected: 0.2549019607843137

case 4

in: [[],10,1]

expected: 1

case 5

in: [[1],0,1]

expected: 1

Starter codePython

def sign_flip_p_value(diffs, n_perm, seed):
    # Your code here
    pass

5 test cases · run your code against them in the editor