Sample size calculator
Equal two-arm designs. Uses standard normal quantiles (α 0.05 → 1.96, α 0.01 → 2.58; 80% power → 0.84, 90% → 1.28). Dropout and clustering are not included.
n per arm (completers)
63
Total randomized ≈ 126 before dropout. Inflate for loss-to-follow-up in the SAP.
How to use the sample size calculator
Choose means or proportions, set alpha and power, enter the planned effect size, and read n per arm before dropout. Check diagnostic operating characteristics with the sensitivity and specificity calculator, plan review type with the IRB review-type helper, or quantify standardized effects with the effect size calculator.
Need a statistician-ready sample-size justification?
Equal-arm normal approximations omit dropout, clustering and adaptive designs. Book a Consultation with our clinical / regulatory team to turn effect-size assumptions into a SAP-ready justification for your device study.
Book a ConsultationFAQ
- What designs does this sample-size calculator cover?
- Equal two-arm comparisons of means or proportions using standard normal (Z) quantiles. It does not model unequal allocation, survival endpoints, non-inferiority margins, cluster randomization, interim looks or dropout inflation beyond a manual note on the results card.
- Which alpha and power quantiles does the tool use?
- For a two-sided alpha of 0.05 it uses Z≈1.96; for alpha ≤ 0.01 it uses Z≈2.58. For power it uses Z≈0.84 at 80% and Z≈1.28 at 90%. Those are the usual normal approximations taught in introductory biostatistics; a full SAP may use exact or simulation-based methods instead.
- Why does a statistician still need to sign the plan?
- Sample size drives Type I/II error, estimands, multiplicity and operating characteristics. Sponsors and IRBs expect a documented statistical analysis plan (or equivalent) with assumptions, not a web calculator printout. Use this page to explore assumptions before the statistician locks the design.
- How should I handle dropout and missing data?
- Inflate the completer n by your expected retention (for example divide by 1 − dropout fraction) and state the rule in the protocol. Missing-data assumptions belong in the SAP; this calculator does not impute them.
- Can I use this for diagnostic accuracy studies?
- Only as a rough two-proportion sketch. Diagnostic studies often size on sensitivity, specificity or AUC with different variance structures. Pair this tool with the diagnostic 2×2 calculator for operating characteristics, then confirm sizing with a statistician familiar with diagnostic designs.
- Does effect size come from a pilot or from literature?
- Either, but document the source. Over-optimistic deltas under-power the study; over-pessimistic deltas waste subjects. For device trials, clinical meaningfulness of the delta should match the intended use and primary endpoint definition.
Educational only. Not medical advice, not a regulatory determination, and not a substitute for qualified RA/QA, legal, clinical, or statistical counsel. Sample size calculator does not submit to FDA, Notified Bodies, or IRBs.
Related clinical study tools
Need a statistician-ready sample-size justification?
Equal-arm normal approximations omit dropout, clustering and adaptive designs. Book a Consultation with our clinical / regulatory team to turn effect-size assumptions into a SAP-ready justification for your device study.
Book a Consultation