π’ Input Parameters
Number of users/samples in each group
Small=0.2, Medium=0.5, Large=0.8
Type I error probability, typically 0.05
π About Statistical Power
Statistical Power = 1 - Ξ², the probability of detecting an effect when it truly exists. Typically, power β₯ 0.80 is desired.
Formula based on normal distribution: Power = Ξ¦(z - z_Ξ±), where z = d Γ β(n/2), d is effect size, n is sample size per group.
Power depends on three factors: larger sample size β higher power; larger effect size β higher power; larger Ξ± β higher power.