How Cohen's d Is Calculated
Cohen's d measures the standardised difference between two group means. It tells you how many standard deviations apart the two groups are, so you can judge whether a difference is practically meaningful, not just statistically significant.
The formula is:
d = (M1 - M2) / SDpooled
Where the pooled standard deviation is:
SDpooled = sqrt( ((n1 - 1) × SD1² + (n2 - 1) × SD2²) / (n1 + n2 - 2) )
This pools the variance from both groups, weighted by their degrees of freedom, to give a single baseline spread against which to measure the mean difference.
Worked Example
Using the calculator defaults: Group 1 has a mean of 105 and a standard deviation of 15 with 30 participants. Group 2 has a mean of 95 and a standard deviation of 15 with 30 participants.
Mean difference = 105 - 95 = 10
Pooled variance = ((29 × 225) + (29 × 225)) / 58 = 13050 / 58 = 225 (because both SDs are equal, the pooled variance equals 225)
Pooled SD = sqrt(225) = 15
Cohen's d = 10 / 15 = 0.67
A d of 0.67 falls in the medium range (0.5 to 0.8), meaning the two groups are roughly two-thirds of a standard deviation apart. This matches the calculator output for the default values.
Interpreting Cohen's d
| |d| Value | Classification | Practical meaning |
|---|
| 0.0 to 0.19 | Negligible | Difference barely visible in data; likely noise |
| 0.2 to 0.49 | Small | Subtle but real; detectable with large samples |
| 0.5 to 0.79 | Medium | Noticeable to an attentive observer |
| 0.8 and above | Large | Clearly visible difference between groups |
These conventions come from Jacob Cohen's 1988 book Statistical Power Analysis for the Behavioral Sciences. Cohen himself acknowledged they were rough benchmarks and encouraged researchers to consider the specific context of their field. In some medical settings a d of 0.2 is clinically important; in others a d of 1.0 might still be insufficient to change practice.
Cohen's d vs Other Effect Sizes
Cohen's d is one of the most widely used effect size measures for comparing two group means. Other common measures include Hedges' g (which applies a small-sample correction to d), Glass's delta (which uses only the control group SD as the denominator), and eta-squared or partial eta-squared (used with ANOVA). For proportions and categorical data, odds ratios or phi are more appropriate. This calculator focuses on Cohen's d for continuous outcomes measured in two groups.
Why Effect Size Matters
A result can be statistically significant (low p-value) but practically trivial. With very large samples, even a tiny mean difference will be statistically significant. Effect size gives you the practical magnitude regardless of sample size. Reporting effect size alongside p-values and confidence intervals is now required or strongly recommended by most journals and the APA Publication Manual.
Related Calculators
Sources and method: Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum Associates. Pooled standard deviation formula per Cohen (1988) and APA Publication Manual (7th ed.). Hedges' g correction factor J = 1 - (3 / (4 × df - 1)) where df = n1 + n2 - 2.
This calculator computes Cohen's d for two independent groups using the pooled standard deviation. It assumes both groups are drawn from populations with equal (or approximately equal) variance. For paired designs, use the standard deviation of the difference scores as the denominator instead. Results are for educational purposes and should be interpreted alongside p-values, confidence intervals and subject-matter knowledge.
