Sample-size sketch
Explore an approximate per-group sample size for a two-sided, balanced comparison of two independent means.
The result
Method & assumptions
Normal approximation for a single primary continuous endpoint. It ignores clustering, unequal allocation, small-sample t corrections, multiplicity, and non-normal outcomes. A statistician should design a real study. This is not a clinical-trial protocol.
Sanity check
For d=0.5, 80% power and α=0.05, the approximation gives 63 evaluable observations per group; 10% dropout raises recruitment to 70 per group.
n per group ≈ ceil[2(z1−α/2 + zpower)² / d²] Recruitment per group = ceil[n / (1−dropout)]
Implementation version 1.0 · 2026-10-15. Deterministic formulas or editorial rules, with no AI model or external service processing your inputs.
Need a complete research scope? This tool is one part of the workflow. Prepare a project brief or compare Karam’s proposed research services.