Statistical Power

Power is the probability of correctly rejecting a false null hypothesis. It depends on effect size, α, variance, and sample size.

Concept

Higher sample size or effect size increases power; lower α reduces power.

Example

With small effect size, doubling n may be required to raise power from 0.6 to 0.8.

FAQs

A priori vs post-hoc?
A priori computes required n; post-hoc computes achieved power.

Multiple testing?
Adjust α (e.g., Bonferroni) which affects power.