Cohen’s d expresses the difference between two group means in standard-deviation units, so readers can judge how large an effect is, not just whether it is significant. Enter the two groups’ means, standard deviations and sample sizes (from your own results or a published paper) to get Cohen’s d, Hedges’ g, Glass’s Δ, a 95% confidence interval and an interpretation.
How to use this calculator
- Enter the mean, standard deviation and sample size of Group 1.
- Enter the same three values for Group 2.
- Click Calculate.
- Report d (or Hedges’ g for small samples) with its 95% confidence interval.
Formula
sₚ = √[((n₁ − 1)s₁² + (n₂ − 1)s₂²) / (n₁ + n₂ − 2)] d = (M₁ − M₂) / sₚ Hedges' g = d × (1 − 3 / (4(n₁ + n₂) − 9)) SE(d) = √[(n₁ + n₂)/(n₁n₂) + d² / (2(n₁ + n₂))], 95% CI = d ± 1.96·SE
Step-by-step example
- Experimental group: M = 78.4, SD = 9.2, n = 35. Control group: M = 72.1, SD = 10.1, n = 33.
- Pooled SD = 9.647.
- d = (78.4 − 72.1) / 9.647 = 0.65 (medium effect).
- Hedges’ g = 0.65; 95% CI [0.16, 1.14].
How to interpret the results
| |d| | Interpretation (Cohen, 1988) |
|---|---|
| < 0.20 | Negligible |
| 0.20 – 0.49 | Small |
| 0.50 – 0.79 | Medium |
| ≥ 0.80 | Large |
These benchmarks are rough guides; compare your effect with typical effects in your own field where possible. A confidence interval that excludes zero indicates a significant difference at the .05 level.
How to report it in APA 7 style
The intervention group scored higher than the control group, with a medium effect size, Cohen’s d = 0.65, 95% CI [0.16, 1.14].
Statistical symbols (t, F, p, r, M, SD, N, d) should be italicised in your thesis.
Assumptions
- The two groups are independent.
- The outcome is continuous and roughly normally distributed.
- Group variances are similar (otherwise consider Glass’s Δ using the control group’s SD).
How to do it in SPSS, R or Excel
R: effectsize::cohens_d(y ~ group). SPSS 27+: Independent-Samples T Test → tick Estimate effect sizes. Excel: apply the pooled SD formula above.
Roman Urdu mein khulasa
Cohen’s d batata hai ke do groups ka farq kitna bara hai, standard deviation ke hisaab se. Sirf p-value se ye pata nahi chalta ke farq practically ahem hai ya nahi. Dono groups ka mean, SD aur n likhein. 0.2 chhota, 0.5 darmiyana aur 0.8 bara effect maana jata hai. Chhote samples (20 se kam) mein Hedges’ g report karna behtar hai.
Frequently asked questions
What is a good Cohen’s d?
By Cohen’s (1988) conventions, d = 0.2 is small, 0.5 medium and 0.8 large. What counts as meaningful depends on your field; in education, effects around 0.4 are often considered practically important.
What is the difference between Cohen’s d and Hedges’ g?
Hedges’ g applies a small-sample correction to d. With large samples the two are almost identical; with fewer than about 20 per group, g is preferred.
Can Cohen’s d be negative?
Yes. The sign only shows which group has the higher mean. Many authors report the absolute value and state the direction in words.
Can I calculate Cohen’s d from a t value?
Yes. For two independent groups, d = t × √(1/n₁ + 1/n₂). The independent t test calculator on ScholarCalc reports d automatically.
Why report an effect size if my result is significant?
APA 7 requires effect sizes because p-values depend heavily on sample size. An effect size shows how large the difference actually is.
Which effect size should I use for paired data?
For pre-post designs use d_z (mean difference divided by the SD of differences), which the paired t test calculator reports.
Related calculators
- Independent Samples t Test Calculator
- Power Analysis Calculator
- Paired Samples t Test Calculator
- Not sure which test to use? Try the test chooser
Cite this calculator
If ScholarCalc helped your analysis, you can cite it in APA 7 format:
ScholarCalc. (2026). Cohen's d Effect Size Calculator [Web application]. https://scholarcalc.com/cohens-d-calculator/
Written and maintained by the ScholarCalc team. Last reviewed: November 2026.
Accuracy: results are computed in your browser and were validated against Python scipy, statsmodels and pingouin on test datasets. See Methodology and accuracy.