Power studies before collecting a single datapoint

Steps

1. 1Estimate effect size from prior literature or pilot data honestly
2. 2Set α=0.05, power=0.80 minimum; compute required N per group
3. 3Account for expected attrition/contamination in recruitment numbers
4. 4Preregister hypotheses, exclusions, and analysis plan
5. 5Define stopping rules and interim analysis boundaries upfront
6. 6Archive design doc citable regardless of outcome

Common Pitfalls

• ▲N=10 convenience samples dressed as exploratory
• ▲HARKing after unblinded looks at trends

Commands

$ npx skills add aniruddhaadak80/skills --skill study-design-power-analysis-first

$ npx skills add aniruddhaadak80/skills --skill study-design-power-analysis-first -g

Tags