Sample size & power
Four quantities locked in one equation — sample size, effect, significance and power. Fix any three and read the fourth, and see the trade-off move.
Advanced
Sample size per group
—
Total sample size
—
Enter the three known quantities to solve for the fourth.
This is for planning a study before it runs. Computing "observed power" from a completed trial's own result is circular and not meaningful — power is an a-priori design property.
Power vs sample size
Power rises steeply, then plateaus — buying the last few percent costs many more participants.
Required n vs effect size
Small effects are expensive — halving the effect roughly quadruples the sample size.
How it's computed
Uses the pooled-variance normal approximation; exact methods and continuity correction can shift n by a few percent, so software may differ slightly.
This is the number of events required — the patient number depends on how many are expected to have an event over follow-up.