Power of Test
The power of a statistical test is P(reject H₀ | H₀ is false) = 1 - β, where β is the probability of Type II error. Power represents the probability of correctly detecting a true effect. Power depends on effect size, sample size, significance level, and the test used. Researchers typically aim for power ≥ 0.80.
Real World
When Spotify A/B tests a new recommendation algorithm across 10,000 users, their data science team calculates required sample size upfront to ensure at least 80% power to detect even a 1% increase in listening time.
Exam Focus
List the three factors that increase power (effect size, sample size, significance level) and the direction of each change.
How well did you know this?