Bootstrap CI for ATE

Problem

Return the percentile bootstrap confidence interval for the average treatment effect. Resample the treated and control outcomes separately with replacement using the seeded Lehmer RNG, compute the difference of means each time, and return the alpha/2 and 1-alpha/2 percentiles with linear interpolation. Return [0.0, 0.0] when either group is empty, lengths differ, or n_boot <= 0.

Hint

Resample within arms to respect the treatment design.

Test cases

case 1

in: [[10,12,9,11,5,6,4,7],[1,1,1,1,0,0,0,0],200,7,0.05]

expected: [3.4937500000000004,6.256250000000001]

case 2

in: [[1,2],[1,0],100,3,0.1]

expected: [-1,-1]

case 3

in: [[1],[1],50,1,0.05]

expected: [0,0]

case 4

in: [[],[],10,1,0.05]

expected: [0,0]

case 5

in: [[1,2,3,4],[1,0,1,0],100,42,0.1]

expected: [-3,1]

Starter codePython

def bootstrap_ate_ci(y, treatment, n_boot, seed, alpha):
    # Your code here
    pass

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