What this answers

This page answers "which sample-size calculator on this site actually matches my study design?" Every design listed here needs a different formula and different required inputs (an assumed effect size, a target precision, a baseline rate), and no single generic calculator can honestly cover them all at once.

How the routing works

Each option maps to one canonical engine, listed in full below. No calculation happens on this page itself; it exists purely so you land on the correct method the first time rather than guessing which formula applies to your situation.

The seven supported designs

A single mean uses Sample Size for a Mean. A single proportion uses Sample Size for a Proportion. Two independent group means use Sample Size for Two Means. Paired before/after measurements use Sample Size for Paired Means. Two independent proportions use Sample Size for Two Proportions. A target correlation uses Sample Size for Correlation. Three or more group means use Sample Size for One-Way ANOVA.

Assumption audit

Calculated from your data: nothing on this page itself; the destination planner performs its own calculation once you arrive.

Evidence to review: whether you have a genuine, defensible estimate of the expected effect size or baseline rate before planning; every design listed here requires that assumption as an input, and a poor assumption produces a confidently wrong sample size.

You must verify: your actual study design before selecting an option; a design mismatch here (for example, planning paired data as if it were two independent groups) leads to the wrong formula entirely, not just a slightly different number.

Common mistakes

Choosing "two independent group means" when your actual design is matched pairs, or the reverse, is the most common design error this router exists to prevent. A second common mistake is planning a study without a target power at all, since power (not just alpha) drives the required sample size just as strongly.

Limitations

This router covers the most common single-outcome designs. More elaborate designs, such as those needing a specific attrition-adjusted or cluster-randomized calculation beyond the options above, are not covered by any single engine on this site and may need dedicated statistical software or a consulting statistician.