Compute Cohen’s d, Hedges’ g and Glass’s Δ from two groups — with confidence intervals, overlap statistics and interpretation.
How this is calculated
What is Effect Size Calculator
An effect size tells you how large a difference is, independent of sample size. While a p-value only signals whether a difference is statistically detectable, an effect size answers the question researchers actually care about: how much do the two groups differ? This Effect Size Calculator converts two groups’ descriptive statistics into the standardized measures most often reported in journals, theses, and meta-analyses.
How to use it
Enter the mean, standard deviation, and sample size for each of your two groups. Group 1 is treated as the treatment or experimental group and Group 2 as the control or comparison group. The results update automatically as you type, and you can adjust the confidence level for the intervals using the slider.
What it calculates
Cohen’s d is the standardized mean difference based on the pooled standard deviation of both groups. It is the most widely reported effect size for comparing two means.
Hedges’ g applies a small-sample correction to Cohen’s d. Cohen’s d slightly overestimates the population effect when samples are small, so Hedges’ g multiplies d by a correction factor and is preferred when group sizes are modest.
Glass’s delta divides the mean difference by the control group’s standard deviation alone, rather than the pooled value. It is useful when the treatment is expected to change variability as well as the mean, so the control SD is the more meaningful yardstick.
The point-biserial correlation r expresses the same effect on the correlation scale, which is convenient for some reporting standards and for converting between effect-size metrics.
The calculator also reports three intuitive overlap measures. The probability of superiority (the common-language effect size) is the chance that a randomly chosen person from Group 1 scores higher than a randomly chosen person from Group 2. Cohen’s U3 is the percentage of the control group that falls below the average treatment score. The overlap statistic is the shared area under the two distribution curves.
Interpreting the result
Cohen (1988) proposed rough benchmarks: around 0.2 is a small effect, 0.5 is medium, and 0.8 or above is large, with values under 0.2 typically considered negligible. These are conventions, not hard rules. A “small” effect can be highly important in some fields and trivial in others, so always interpret the number against the context of your discipline and prior research.
Reporting tips
Always report which effect size you used and which standard deviation it is based on, alongside the confidence interval and your sample sizes. Confidence intervals communicate the precision of the estimate, which is why the calculator displays them by default. For meta-analytic work or small samples, Hedges’ g is usually the safer choice.
Frequently asked questions
What is a good effect size? There is no universal answer. Use Cohen’s benchmarks as a starting point but judge importance relative to the typical effects in your research area.
Should I report Cohen’s d or Hedges’ g? Report Hedges’ g when your samples are small, since it corrects the upward bias in d. With large samples the two values are nearly identical.
Can I get an effect size from a t-test? Yes. If you only have the t value and group sizes, d can be derived as t multiplied by the square root of (1/n1 + 1/n2); this calculator works directly from the means and standard deviations, which is the most transparent approach.
