Cohen's d Calculator
Estimate effect size from summary statistics or directly from a t-statistic. Useful for A/B tests, experiments, and research reports.
What is a d calculator?
A d calculator computes Cohen's d, one of the most common standardized effect sizes. Instead of only asking whether two groups are statistically different, Cohen's d helps answer a more practical question: how large is the difference?
This matters in science, product analytics, education, healthcare, and business testing. A tiny p-value can still correspond to a very small real-world effect when sample sizes are large. Cohen's d gives you a scale that is easier to compare across studies and contexts.
How this calculator works
Method 1: Means + SDs + Sample Sizes
For two independent groups, the calculator first computes a pooled standard deviation and then divides the mean difference by that pooled spread:
d = (M₁ − M₂) / SDpooled
It also reports Hedges' g, a small-sample corrected version of d, and an approximate 95% confidence interval.
Method 2: t-statistic + Sample Sizes
If you already have a t-value from an independent-samples test, the calculator can estimate:
d = t × √(1/n₁ + 1/n₂)
This route is handy when your paper or software output includes t but not means/SDs.
Interpreting Cohen's d
A common rule of thumb uses absolute values of d:
- < 0.20: Negligible
- 0.20 to 0.49: Small
- 0.50 to 0.79: Medium
- 0.80+: Large
These are guidelines, not universal laws. In some domains, a d of 0.20 is meaningful; in others, even 0.50 may be modest. Always interpret effect size in context.
Worked example
Suppose a training program group has mean test score 78.4 (SD 10.2, n=35), while the control group has mean 72.1 (SD 9.5, n=33). Enter those values into the calculator and you will get a d around 0.64. That falls in the medium effect range, suggesting a meaningful shift in performance.
When to use d (and when to be careful)
Good use cases
- Comparing two independent groups on a continuous outcome
- Summarizing treatment or intervention magnitude
- Power analysis planning and meta-analysis inputs
Common pitfalls
- Using d with highly skewed outcomes without checking assumptions
- Mixing paired/repeated-measures designs with independent-group formulas
- Ignoring confidence intervals and only reporting point estimates
- Interpreting "small" as "unimportant" without domain context
Reporting template you can reuse
"The intervention group scored higher than control, d = 0.64 (Hedges' g = 0.63), indicating a medium effect."
If possible, also include confidence intervals, sample sizes, and the raw means and standard deviations. Transparent reporting makes your results easier to evaluate and replicate.
Final thoughts
A good d calculator helps you go beyond significance testing and focus on practical magnitude. Use it alongside p-values, confidence intervals, and domain knowledge for better decisions and stronger analyses.