Effect Size Calculator
Cohen d and Hedges g from two means, two standard deviations and two sample sizes, with the pooled SD.
What the effect size calculator does
It standardises the difference between two group means. The pooled standard deviation is √(((n₁ − 1)s₁² + (n₂ − 1)s₂²)/(n₁ + n₂ − 2)); Cohen's d is (mean 1 − mean 2) ÷ the pooled SD; and Hedges' g multiplies d by the small-sample correction J = 1 − 3/(4(n₁ + n₂) − 9).
Two groups with means 105 and 100, both standard deviations 15 and thirty observations each have a pooled SD of 15, so d = 5 ÷ 15 = 0.3333. With 60 observations in total J is 0.987 and g is 0.329, a little smaller than d.
The sign is the order of the means: mean 1 above mean 2 gives a positive d. Swapping the groups flips the sign, not the size. The page prints the pooled SD and the degrees of freedom so the denominators can be checked.
It assigns no small, medium or large band. Those bands are a convention that varies by field, and a page that printed one would be making a judgement rather than doing arithmetic. The result line says so explicitly.
Each sample size is a whole number of at least 2, because the pooled variance divides by n₁ + n₂ − 2, and each standard deviation must be above zero. A missing value, a fractional count, a count above one million and a zero standard deviation are refused by name.
Figures print to up to four decimal places with trailing zeros trimmed.
How to use it
Enter Mean 1, Standard deviation 1 and Sample size 1 for the first group.
Enter Mean 2, Standard deviation 2 and Sample size 2 for the second.
Press Run. The result prints both means, the pooled SD with its degrees of freedom, Cohen d and Hedges g.
Worked example
Input with the defaults:
105, s 15, n 30 and 100, s 15, n 30
Output:
Pooled SD: 15 (df = 58)
Cohen d: 0.3333 ((mean 1 − mean 2) ÷ pooled SD)
Hedges g: 0.329 (d × J, J = 0.987)
Why it runs in your browser
The pooled variance and the two effect sizes are computed in this tab. Your summary statistics are never uploaded.
The page standardises the difference between two group means. Two groups with means 105 and 100, both standard deviations 15 and thirty observations each, have a pooled standard deviation of 15, so Cohen's d is 0.3333.
Hedges' g applies a small-sample correction to d: J = 1 − 3/(4(n₁ + n₂) − 9), so with sixty observations in total g is 0.329. Both figures are printed with the pooled SD and the degrees of freedom.
The page assigns no small, medium or large band. Those bands are a convention that changes with the field, and the result says so rather than pretending the arithmetic chose one.
What a standardised difference buys you
A raw difference of five points means one thing when the spread is two and another when it is fifty. Dividing by the pooled standard deviation puts the difference on a scale with no units, so two studies measuring different things can be compared on the same axis.
The pooled SD weights each group's variance by its degrees of freedom: √(((n₁ − 1)s₁² + (n₂ − 1)s₂²)/(n₁ + n₂ − 2)). That is why the page asks for both sample sizes and prints the degrees of freedom it used.
Why the page prints g as well as d
Cohen's d divides by the pooled SD estimated from the sample, which is biased upward in small samples. Hedges' g multiplies by J, a correction that shrinks toward zero as the total sample grows: at sixty observations J is 0.987, and at a thousand it is indistinguishable from 1.
Both are printed because papers report both. A reader matching a published figure needs to know which one it was, and the correction factor is printed in the result rather than left to a footnote.
How to use Effect Size Calculator
Enter Mean 1, Standard deviation 1 and Sample size 1 for the first group.
Enter Mean 2, Standard deviation 2 and Sample size 2 for the second group.
Press Run. The result prints both means, the pooled SD with df, Cohen d and Hedges g.
What to Expect
Best for
- Putting a difference between two groups on a unit-free scale
- Comparing an effect across studies with different instruments
- Reporting both d and g with the correction shown
Not ideal for
- A paired or repeated-measures design, which needs a different denominator
- More than two groups, which is an ANOVA question
- Deciding whether an effect is practically important
What this tool keeps
- Both means, both standard deviations and both sample sizes
- The sign of the difference
What may need cleanup
- Group summaries reported as standard errors, which must be converted to standard deviations first
- A design with unequal variances, where a different standardiser may be wanted
Common errors
- Entering a standard error where a standard deviation belongs
- Reading the magnitude bands as if they were computed by the page
Example use cases
- Reporting an intervention effect beside a t-test
- Comparing two published effects on one scale
- Teaching why a standardised difference is unit-free
Sample input
Group 1: mean 105, SD 15, n 30. Group 2: mean 100, SD 15, n 30.
Sample output
A pooled SD of 15 with 58 degrees of freedom, Cohen d of 0.3333, J of 0.987 and Hedges g of 0.329.
Who this is for
- Researchers reporting an effect beside a significance test
- Students learning standardised mean differences
Frequently asked questions
Should I report d or g?
Report the one your field reports, and say which. g is d with a small-sample correction, so they converge as the sample grows; at sixty observations the gap here is 0.004. Printing both lets a reader match either convention.
Why is there no small, medium or large label?
Because those thresholds are conventions, not arithmetic, and they differ between fields. A d of 0.3 can be trivial in one context and important in another. The page computes the number and leaves the judgement to the reader.
What does a negative d mean?
That mean 1 is below mean 2. The sign follows the order you entered the groups; swapping them flips the sign and leaves the magnitude unchanged.
Can I use standard errors instead of standard deviations?
Not directly. The pooled formula needs standard deviations. Convert a standard error back with s = SE × √n for the group it came from, then enter that s.
Build custom tools for your store
Join the SDK program to create useful ecommerce utilities for your store.