Effect Size Calculator: Cohen's d, Hedges' g, Odds Ratio and Risk Ratio
Turn means, SDs, t-values or 2×2 tables into effect sizes with 95% confidence intervals. It’s ideal for your results section, a power analysis or extracting data for a meta-analysis.
| Mean | SD | n | |
|---|---|---|---|
| Group 1 (treatment) | |||
| Group 2 (control) |
Have a standard error instead of an SD? SD = SE × √n.
| Event (yes) | No event | |
|---|---|---|
| Treatment / exposed | ||
| Control / unexposed |
If any cell is 0, 0.5 is added to every cell (Haldane–Anscombe correction).
How to use the effect size calculator
- Pick what you have: means and SDs, a t value, a 2×2 table of counts, or an effect size you want to convert.
- Type your numbers. Results update as you type.
- Read the effect size, its confidence interval and its magnitude (small, medium or large).
- Copy the APA-style sentence into your results, or copy the effect size and its variance into your meta-analysis spreadsheet.
What is an effect size?
A p-value tells you whether an effect is likely to be real. An effect size tells you how big it is. A drug can give a “significant” p-value with 10,000 patients and still make almost no difference in practice. That’s why the APA manual and most journals now ask you to report effect sizes with confidence intervals. Meta-analyses depend on them too: you cannot pool p-values, but you can pool effect sizes.
Which effect size should I use?
| Your data | Effect size | Small · Medium · Large |
|---|---|---|
| Two group means (continuous outcome) | Cohen’s d or Hedges’ g | 0.2 · 0.5 · 0.8 |
| Before–after or paired measurements | Cohen’s dz | 0.2 · 0.5 · 0.8 |
| Two variables (correlation) | Pearson r | 0.1 · 0.3 · 0.5 |
| Yes/no outcome in two groups | Odds ratio or risk ratio | OR ≈ 1.5 · 2.5 · 4.3 |
| ANOVA | η² or partial η² | 0.01 · 0.06 · 0.14 |
Formulas
Frequently asked questions
What is the difference between Cohen's d and Hedges' g?
Both are standardized mean differences. Hedges’ g multiplies d by a small-sample correction factor J, so it is slightly smaller and unbiased. With more than about 50 participants the two are almost identical. Meta-analyses usually report g.
What counts as a large effect size?
Cohen’s rough benchmarks for d are 0.2 (small), 0.5 (medium) and 0.8 (large). For r they are 0.1, 0.3 and 0.5. These are only rules of thumb. A d of 0.2 can be very important for a cheap public health intervention, so always interpret effects in your field’s context.
Can I calculate an effect size from a p-value?
Yes, indirectly. Convert the p-value to a t value with our p-value calculator (critical value tab with the exact p), then use the ‘From t statistic’ tab here with the group sizes.
Odds ratio or risk ratio: which should I report?
Risk ratios are easier to understand (‘twice the risk’). Odds ratios are needed for case-control studies and logistic regression. When the event is rare (under about 10%), the two are very similar.
What does NNT mean?
The number needed to treat is how many people must receive the treatment for one extra person to benefit, calculated as 1 ÷ the absolute risk difference. A lower NNT means a more effective treatment.
Is my data uploaded?
No. Everything is calculated in your browser, and nothing is sent to any server.