Cochran-Armitage Trend

Problem

Return the standardized Cochran-Armitage trend statistic sum(w_i*(x_i - n_i*p))/sqrt(p*(1-p)*(sum(w_i^2*n_i) - (sum(w_i*n_i))^2/N)) for cases, group totals, and numeric dose scores. Return 0.0 for empty or mismatched input, zero totals, an overall proportion at 0 or 1, or zero trend variance.

Hint

The statistic is the dose-weighted covariance of cases and scores, standardized.

Test cases

case 1

in: [[10,20,30],[100,100,100],[0,1,2]]

expected: 3.5355339059327373

case 2

in: [[5,10,15],[50,50,50],[0,1,2]]

expected: 2.5

case 3

in: [[20,20,20],[100,100,100],[0,1,2]]

expected: 0

case 4

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

expected: 0

case 5

in: [[10,20],[100,100],[0,1]]

expected: 1.9802950859533488

Starter codePython

def cochran_armitage(cases, totals, scores):
    # Your code here
    pass

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