T-Test of Two Mutually Exclusive Proportions

From Q
Jump to: navigation, search

Where g_1 and g_2 are the two proportions:

t=\frac{g_1-g_2}{\sqrt{d_{eff}(\frac{g_1+g_2 - (g_1-g_2)^2)}{n - 1})}}


p \approx 2\Pr(t_{n-1} \ge |t|),
b is 1 if Bessel's correction for Means is selected in Statistical Assumptions and 0 otherwise,
d_{eff} is Extra deff, and
n is the unweighted sample size.