Cohen's d Effect Size Calculator

How to use

1. 1Pick an input mode at the top. Summary statistics is the fastest if you already have the mean, SD, and n for each group. Raw data is useful when you have the underlying observations and want the means and SDs computed for you.
2. 2Label your groups. Group 1 is the numerator of Cohen's d, so its mean is what gets compared against the Group 2 mean. Group 2 doubles as the control for Glass's delta.
3. 3Enter the data. In summary mode, type each group's mean, standard deviation, and sample size. Set the SD type selector to match how your SD was computed (Excel STDEV.S or STDEV.P) so the pooled formula uses the right convention. In raw mode, paste the values for each group separated by commas, semicolons, line breaks, or whitespace.
4. 4Read the effect size card. Cohen's d, Hedges' g, Glass's delta, eta squared, omega squared, and the point-biserial correlation are all shown with the 95% CI for d. The colored badge translates the magnitude into the Cohen (1988) and Sawilowsky (2009) scales.
5. 5Check the Practical meaning panel for CLES, U3, and overlap. These convert d into the kind of sentence anyone can understand without statistical training.
6. 6Click Copy summary to paste a clean multi-line report of the means, SDs, every effect size, the CI, and the interpretation into a paper, a Slack thread, or a lab notebook.

About this tool

Cohen's d Effect Size Calculator computes the standardized mean difference between two independent groups and reports every related effect size measure in a single readout, entirely in your browser. Two input modes are supported: a summary-statistics mode that accepts the mean, standard deviation, and sample size of each group, and a raw-data mode that accepts the underlying observations one group per textarea and computes the descriptive statistics for you. The result panel shows Cohen's d using the pooled sample standard deviation, Hedges' g with the J(df) small-sample correction (J = 1 - 3 / (4 df - 1)), Glass's delta which divides by the control-group SD alone for situations where the intervention is expected to change variability as well as the mean, the population variant of d that uses the n-weighted SD, eta squared and a less-biased omega squared, the point-biserial correlation between group membership and outcome, the mean difference and pooled standard deviation that feed into d, and the 95% confidence interval for d using the Hedges and Olkin (1985) approximate variance. Practical-meaning summaries are reported alongside the raw numbers: the Common Language Effect Size (probability that a randomly chosen Group 1 observation exceeds a randomly chosen Group 2 observation), Cohen's U3 (the percentile of Group 2's mean within Group 1's distribution), and the overlap area of two equal-variance normal curves with the calculated d. Both the original Cohen (1988) small / medium / large thresholds and the extended Sawilowsky (2009) scale (very small / small / medium / large / very large / huge) are shown so the reader can match the convention used in their field. The SD type selector handles the common mismatch between sample SD (n - 1, the default in R, SPSS, JASP, and Excel STDEV.S) and population SD (n, Excel STDEV.P) without the user having to convert by hand. A swap-groups control flips Group 1 and Group 2 in one click, which is useful when the sign of d came out the wrong way for the write-up. Useful for psychology, education, medical, and social-science researchers reporting effect sizes alongside p-values, growth and product teams interpreting A/B test results in standardized units, graduate students checking textbook problems, meta-analysts converting between effect size measures, and anyone who needs more than a t-test to describe how separated two groups really are. The visitor and group data you enter never leaves your browser.

Free to use. Works in your browser. No signup, no login.