Calculate Effect Size: Cohen’s d and Other Measures

Effect size describes how large a difference, relationship, or pattern is, rather than simply whether a statistical test calls it significant. This calculator estimates Cohen’s d from two group means and standard deviations, using the pooled standard deviation; enter each group’s mean, standard deviation, and sample size to get d and its pooled spread.

Formula

For two independent groups, Cohen’s d is the difference between their means divided by the pooled standard deviation:

d = (M₂ − M₁) / sₚ

sₚ = √[((n₁ − 1)s₁² + (n₂ − 1)s₂²) / (n₁ + n₂ − 2)]

• M₁ and M₂ are the group means, in the same units as the measured outcome.
• s₁ and s₂ are the groups’ sample standard deviations, also in outcome units.
• n₁ and n₂ are the numbers of observations in the groups.
• sₚ is the pooled standard deviation, in outcome units. Cohen’s d is dimensionless.

The pooled spread weights each group’s variance by its degrees of freedom. The displayed d keeps the direction: a positive result means Group 2’s mean is higher, while a negative result means it is lower. The calculator requires positive sample sizes and nonnegative standard deviations; for this formula, each group needs at least one observation and the pooled denominator must be defined.

Worked Example

Use the calculator’s defaults: Group 1 has mean 10, standard deviation 2, and n = 25; Group 2 has mean 12, standard deviation 2, and n = 25. The mean difference is 12 − 10 = 2.

First calculate the pooled standard deviation:

sₚ = √[((25 − 1) × 2² + (25 − 1) × 2²) / (25 + 25 − 2)]
= √[(24 × 4 + 24 × 4) / 48]
= √4 = 2.

Then divide the difference by the pooled standard deviation: d = (12 − 10) / 2 = 1.000. The result is positive because Group 2 has the higher mean. The pooled standard deviation is 2.000 in the outcome’s original units.

What Effect Size Tells You

A p-value addresses how compatible the data are with a specified null hypothesis under a statistical model; it does not tell you how important or large an observed difference is. Effect size provides a magnitude estimate. Reporting both can help readers distinguish a small difference estimated precisely from a larger difference estimated uncertainly.

Cohen’s d expresses a mean difference in standard-deviation units. For example, d = 1 means the group means differ by one pooled standard deviation. That statement is not the same as saying one group scored one point higher: the raw difference depends on the outcome’s units and spread.

Rules of thumb often describe absolute d values near 0.2, 0.5, and 0.8 as small, medium, and large. These are rough conventions, not universal thresholds. A difference that is practically important depends on the field, outcome, measurement quality, and consequences of the difference. A tiny standardized effect can matter in some settings; a conventionally “large” one may not answer the practical question.

When to Use Cohen’s d

This version is appropriate for comparing the means of two independent groups when a standardized mean difference is useful. It is commonly used to summarize experimental or observational group comparisons and to compare results measured on different scales, provided the constructs and study designs are meaningfully comparable.

• Use the pooled-standard-deviation form when the groups are independent and a common within-group spread is a reasonable summary.
• For measurements taken from the same people twice, use a paired or repeated-measures effect-size definition. The variability of within-person differences matters, so this independent-groups formula is not interchangeable.
• For a single group compared with a known reference, a one-sample standardized difference may be suitable, with the reference and standard deviation specified.
• For relationships between variables, proportions, or categorical outcomes, use an effect size designed for that question, such as a correlation or a difference in proportions.

Choosing and Interpreting a Measure

Effect size is a family of measures, not one universal number. Cohen’s d standardizes a difference in means. Hedges’ g is a closely related standardized mean difference with a small-sample correction, often useful when estimating effects from smaller samples. Glass’s delta instead divides the mean difference by a selected group’s standard deviation, which can be relevant when an intervention may change the variability and the control group is the intended reference.

These measures can produce different values because they make different choices about the denominator and bias correction. State which measure you report and how it was calculated. For a complete account, include the raw difference in meaningful units as well as the standardized effect when possible, and report a confidence interval to show uncertainty. The calculator gives a point estimate, not a confidence interval or a test of significance.

When comparing studies, check that the outcome direction, group definitions, design, and standardization choices match. A positive d in one study may represent improvement while a positive d in another represents worsening if the outcomes are coded differently.

Common Mistakes and Limitations

• Mixing units: Both means must use the same unit, and both standard deviations must use that unit too. Convert inches to centimeters, pounds to kilograms, or other measures before entering values if they differ between groups. The ratio itself has no unit, but inconsistent inputs make it meaningless.
• Using the wrong standard deviation: Enter sample standard deviations, not standard errors. A standard error measures uncertainty in a mean and is generally smaller; substituting it changes the effect estimate.
• Ignoring direction: Reversing the order of the groups changes the sign, not the magnitude. State the group order so readers can interpret the result.
• Rounding too early: Keep full precision through the calculation and round the final result for reporting. The calculator displays d to three decimal places.
• Assuming pooled spread is always suitable: The pooled formula is most defensible when group variability is reasonably comparable. Strongly unequal variances or very different group sizes may call for another standardization or an analysis tailored to the study.

Cohen’s d does not describe the full shape of either group’s distribution, prove causation, or establish practical importance. A standardized mean difference can conceal differences in measurement reliability and population variability. Interpret it alongside the study design, raw data summaries, uncertainty, and subject-matter context.

Frequently Asked Questions

How do you calculate effect size from two means?
For independent groups, subtract one mean from the other and divide by the pooled standard deviation. The calculator uses d = (M₂ − M₁) / sₚ, where the pooled spread accounts for both groups’ standard deviations and sample sizes.

What does an effect size of 0.5 mean?
If it is Cohen’s d, the group means are separated by half a pooled standard deviation. Whether that is a meaningful difference depends on the outcome and context; 0.5 is not a universal cutoff for importance.

Can effect size be negative?
Yes. A negative Cohen’s d means the first group’s mean is higher than the second under the calculator’s formula, because it calculates Group 2 minus Group 1. The sign indicates direction; the absolute value indicates standardized magnitude.

Is Cohen’s d the same as a p-value?
No. Cohen’s d estimates the size and direction of a standardized mean difference. A p-value is a different quantity related to a statistical test and its assumptions. Neither alone communicates the full result or its practical importance.

Can I use this formula for paired data?
Not as a default. For paired observations, the effect-size calculation should reflect the within-person differences and the chosen repeated-measures convention. Use a paired-data method rather than treating the measurements as independent groups.

Related

• Back-to-Back Stem-and-Leaf Plot Calculator Guide
• Average Triathlon Time Calculator: Distance and Pace
• Gravel Driveway Cost Calculator: Estimate Material Needed

in m1 | Group 1 mean | - | 10 | 
in s1 | Group 1 standard deviation | - | 2 | 0
in n1 | Group 1 sample size | - | 25 | 1
in m2 | Group 2 mean | - | 12 | 
in s2 | Group 2 standard deviation | - | 2 | 0
in n2 | Group 2 sample size | - | 25 | 1
out d | Cohen's d | - | (m2 - m1) / sqrt(((n1 - 1) * s1 ^ 2 + (n2 - 1) * s2 ^ 2) / (n1 + n2 - 2)) | 3
out sp | Pooled standard deviation | - | sqrt(((n1 - 1) * s1 ^ 2 + (n2 - 1) * s2 ^ 2) / (n1 + n2 - 2)) | 3