Calculateus

Cohen's d Effect Size Calculator

Calculate Cohen's d - a standardized measure of the size of the difference between two group means.

Result

Cohen's d
0.444
Effect Size
Small

About the Cohen's d Calculator

The Cohen's d Effect Size Calculator standardizes the difference between two group means into a single number expressed in standard deviation units. It answers a different question than a significance test does, namely how large a difference actually is in practical terms, not just whether it's statistically detectable.

How It Works

Enter the mean of each group and a pooled standard deviation that represents the combined spread of both groups. The calculator subtracts the second mean from the first and divides by the pooled standard deviation to get Cohen's d, then labels the result's magnitude using Cohen's original benchmarks based on the absolute value of d.

d = (Mean 1 - Mean 2) / Pooled Standard Deviation; magnitude is labeled Negligible below 0.2, Small below 0.5, Medium below 0.8, and Large at 0.8 or above, using the absolute value of d.

Formula & Methodology

By hand, subtract group 2's mean from group 1's mean, then divide that difference by the pooled standard deviation. Take the absolute value of the result and compare it to 0.2, 0.5, and 0.8 to find its magnitude label.

Examples

A moderate test score gap

With a group 1 mean of 85, a group 2 mean of 79, and a pooled standard deviation of 13.5, Cohen's d comes out to (85 - 79) / 13.5 = 0.44, which the calculator labels a Small effect since it falls below the 0.5 threshold.

A large difference in a training outcome

For a group 1 mean of 72, a group 2 mean of 58, and a pooled standard deviation of 10, Cohen's d is (72 - 58) / 10 = 1.4, which the calculator labels a Large effect since it's well above the 0.8 threshold.

Advantages

  • Expresses a mean difference in standardized units, making it comparable across studies or variables measured on different scales.
  • Automatically classifies the result using Cohen's established magnitude labels, saving you from memorizing the thresholds.
  • Separates the question of effect size from the question of statistical significance, which a p-value alone cannot do.

Common Mistakes

  • Treating a statistically significant result as automatically meaningful, when a large sample size can make even a negligible Cohen's d reach significance.
  • Using the standard deviation of just one group instead of a properly pooled standard deviation across both groups.
  • Applying Cohen's generic benchmarks of 0.2, 0.5, and 0.8 without considering that typical effect sizes vary a lot by research field and context.

Edge Cases to Watch For

  • Pooled standard deviation must be greater than zero, or the calculator returns an error instead of dividing by zero.
  • The sign of d shows direction, whether group 1 or group 2 scored higher, while the magnitude label is always based on the absolute value, so a d of -0.6 is still labeled Medium.
  • The magnitude thresholds of 0.2, 0.5, and 0.8 are Cohen's original rough conventions, not universal statistical cutoffs, and what counts as a meaningful effect can vary by field.

Common Use Cases

  • Researchers reporting effect sizes alongside p-values in psychology, education, or medical study results.
  • Analysts comparing the practical size of a difference between a treatment and control group after a two-sample t-test.
  • Meta-analysts standardizing results across multiple studies that used different measurement scales.
Written & fact-checked by the Calculateus TeamLast updated August 5, 2026How we verify our formulas

Frequently asked questions

How do I interpret Cohen's d?

Jacob Cohen's original (rough) benchmarks are 0.2 for a small effect, 0.5 for medium, and 0.8 for large - unlike a p-value, which only tells you whether a difference is statistically detectable, Cohen's d tells you how large that difference actually is in standardized terms, which matters for practical significance.

Conclusion

Cohen's d turns a raw difference between two means into a standardized, comparable measure of effect size. Reported alongside significance testing, it helps separate whether a difference is real from whether it's actually large enough to matter in practice.