Skip to content
Omnikalc

Cohen’s h calculator

Enter two proportions in percent, a significance level and a power to get Cohen’s h, how large the effect is and the sample size needed in each group to detect it.

Updated on
Your inputs
%

Value out of range

%

Value out of range

Results

Cohen’s h

0.2189

Absolute effect size |h|
0.2189
Small effect
Sample size per group
328
Feasible sample size
Total sample size
656
Difference in percentage points
10.0 pts
Power used
80 %

The Cohen’s h calculator gives the effect size of the difference between two proportions and the sample size per group needed to detect it with the power and the significance level you choose.

How the calculation works

  • Enter the two proportions in percent.
  • Choose the significance level and the power.
  • The calculator gives Cohen’s h, its size category, the sample size per group and the total.

Example

A new design is expected to convert 35 % of visitors against 25 % for the old one. The angles are 2 × arcsin(√0.35) = 72.5° and 2 × arcsin(√0.25) = 60°, so h = 0.2189, a small effect. With a 5 % two-sided test and 80 % power, n = 2 × 2.80² / 0.2189² gives 328 per group, or 656 in total.

Why it matters

The calculator helps researchers who plan studies with binary outcomes, marketing and product teams who size an A/B test, and students who report effect sizes.

Good to know

The arcsine approach is a good approximation for proportions between about 5 % and 95 % and for groups of at least 20. For very rare events, an exact method is better. The sample size assumes equal groups; with unequal groups, the total grows. A small effect usually demands hundreds of people per group, so check that the expected difference is realistic. Report h together with the proportions and a confidence interval, not alone. If you compare a proportion with a fixed value rather than with another group, the single-sample version halves the variance and divides the sample size by two.

The Cohen’s d calculator covers differences in means, the statistical power calculator links effect, sample and power, the A/B test sample size calculator targets conversion rates and the effect size converter moves between d, r and odds ratios.

Frequently asked questions

What is Cohen’s h?

It is an effect size for the difference between two proportions. The proportions are first transformed with an arcsine square-root, φ = 2 × arcsin(√p), which makes the variance independent of p, and h = φ1 − φ2. Cohen’s conventions are 0.2 for small, 0.5 for medium and 0.8 for large.

Why not just use the difference in percentage points?

A gap of 10 points means more near 0 % or 100 % than near 50 %, because the variance of a proportion is smaller at the extremes. The arcsine transformation puts all the comparisons on the same scale, so 5 % against 15 % is not treated like 45 % against 55 %.

How is the sample size computed?

For two independent groups of equal size, n = 2 × (z for the significance level + z for the power)² / h², rounded up. For a 5 % two-sided test with 80 % power, the sum of the two z values is 2.80.

Was this calculator helpful?

Thanks for your feedback!

Results are estimates provided for information purposes. They do not replace professional advice.