Sample size and power planning
Enter an effect, α and target power. The calculator estimates sample size for a two-sided t-test with independent groups or paired measurements, and for a one-way ANOVA with equal group sizes. Calculations have been checked against SciPy.
- Effect magnitude |d| used
- 0.500
- Needed per group
- 64
- Total observations
- 128
- Power at the entered sample size
- 0.338
The independent-groups t-test assumes normal data and a common variance. The paired t-test assumes normally distributed differences; d uses the SD of differences. ANOVA assumes normality, equal variances and equal group sizes. Observations must be independent except for declared pairs. For Welch, repeated, factorial and nonparametric methods, this calculation is only a guide. Choose an effect from prior evidence or the smallest one that matters in your field, not from the same data you are analyzing.
Cohen benchmarks are conventional: d = 0.2, 0.5 and 0.8; f = 0.1, 0.25 and 0.4. These are not universal cutoffs. Prefer the smallest effect that matters in your field.