Statistical Power

1Detailed Explanation

Power = 1 - β, where β is the Type II error rate. Conventionally, studies aim for 80% or 90% power. Power depends on: effect size (larger effects are easier to detect), sample size (larger samples have more power), significance level (stricter alpha requires larger samples), and one-tailed vs. two-tailed testing. Underpowered studies are unethical because they expose participants to an intervention that cannot definitively answer the research question. Post-hoc power calculations based on non-significant results are meaningless. A priori sample size calculation should be reported. Power analysis software (G*Power, nQuery) or simulation can determine required sample size.

2Examples

• A.A study with 80% power at alpha = 0.05 can detect the specified effect 80% of the time
• B.An a priori sample size calculation: 128 participants per arm needed for 80% power to detect d = 0.5

3Why It Matters in Research

Sample size justification is required by CONSORT and most ethics committees. Underpowered studies waste resources and participant contributions.

4Related Terms