Opening your learning space
Starting the app …
Opening your learning space
Starting the app …
Cohen’s d standardizes the difference between two independent group means using their pooled variability.
Free to browse · flashcards or a multiple-choice quiz · import from Essential
Formula
d dimensionless · means and standard deviations share the measurement unit · n dimensionless
| Symbol | Meaning | Unit |
|---|---|---|
| Standardized mean difference in pooled-standard-deviation units | dimensionless | |
| x̄1, x̄2 | Means of the two independent groups | unit of the measured quantity |
| s1, s2 | Group sample standard deviations | unit of the measured quantity |
| n1, n2 | Group sample sizes | people or measurements |
| Pooled standard deviation assuming a common variance | unit of the measured quantity |
d = \frac{\bar{x}_1-\bar{x}_2}{s_p}, \quad s_p=\sqrt{\frac{(n_1-1)s_1^2+(n_2-1)s_2^2}{n_1+n_2-2}}Cohen’s d expresses a mean difference in units of spread, making results measured on different scales easier to compare. For two independent groups with plausibly similar variances, sp is the degrees-of-freedom-weighted pooled standard deviation. The measure is not a p-value and does not say whether an effect is statistically significant. The common 0.2, 0.5 and 0.8 labels are rough conventions, not universal thresholds for biological importance. With small samples, d is biased; Hedges’ g corrects this bias. Paired data require different standardizations.
For two independent groups where a pooled common spread is sensible; paired data or strongly different variances require other variants.
The raw mean difference is divided by a common unit of spread.
Pooled standard deviation
Only for d ≠ 0 and with the matching group order.
Two groups differ by 6 points, sp=9. Calculate d.
d=6/9≈0.67.
The mean difference is about two-thirds of the pooled standard deviation; this alone implies neither significance nor biological importance.
Reading d as a p-value or universal relevance threshold.
d describes standardized magnitude; uncertainty, context and study design remain necessary.
Formula cluster
Effect size and precision answer different questions and should be considered together.
Group 1 has mean 72, s1=10 and n1=25; group 2 has mean 66, s2=8 and n2=25. sp=√((24·100+24·64)/48)=√82≈9.06. Thus d=(72−66)/9.06≈0.66. Check: the 6-point difference is about two thirds of the pooled standard deviation.
Comparing biological studies on different measurement scales, reporting mean differences alongside p-values and preparing meta-analyses
Here is how to work through a typical Cohen's d Effect Size (d = (x̄₁ − x̄₂)/sₚ) task step by step:
Task
Two groups differ by 6 points, sp=9. Calculate d.
Solution path
d=6/9≈0.67.
The mean difference is about two-thirds of the pooled standard deviation; this alone implies neither significance nor biological importance.
First calculate the pooled standard deviation by weighting group variances by their degrees of freedom. With equal sample sizes n1=n2=25, s1=10 and s2=8, sp=√82≈9.06. For means 72 and 66, d=(72−66)/9.06≈0.66. The sign indicates direction according to group order; the magnitude gives the gap in pooled-spread units. The raw difference remains important for substantive interpretation. Pair standardized values with the raw difference, group spreads and sample sizes. This keeps units, data volume and practical meaning visible rather than hiding them behind a dimensionless number.
No. Commonly quoted benchmarks such as 0.2, 0.5 or 0.8 are rough conventions, not biological laws. A standardized effect of 0.66 may be unimportant or highly relevant depending on the measurement, consequences and field. Cohen’s d also does not indicate estimate precision or whether a hypothesis was statistically tested. Report means, spreads, sample sizes and a confidence interval where possible. Practical decisions depend on the absolute magnitude and its consequences. Pair standardized values with the raw difference, group spreads and sample sizes. This keeps units, data volume and practical meaning visible rather than hiding them behind a dimensionless number.
The classic formula fits two independent groups whose spreads can reasonably be combined as estimates of one population variance. It weights each variance by n−1. With strongly unequal variances, pooling can create a misleading reference scale; an alternative effect size or a model allowing unequal variances may be preferable. For paired measurements, the spread of the differences is relevant, not the pooled spread of independent groups. The design and chosen d variant should therefore be stated explicitly. Pair standardized values with the raw difference, group spreads and sample sizes. This keeps units, data volume and practical meaning visible rather than hiding them behind a dimensionless number.
Cohen’s d describes the standardized magnitude of a group difference. A p-value describes, under a statistical model, how unusual the data or a more extreme result would be if the null hypothesis held. A small p-value does not guarantee a large or important effect; with huge samples, a small difference can be significant. Conversely, a practically relevant difference may be imprecise and non-significant in a small sample. Effect size, uncertainty and test result answer different questions and should be reported together. Pair standardized values with the raw difference, group spreads and sample sizes. This keeps units, data volume and practical meaning visible rather than hiding them behind a dimensionless number.
Cohen’s d tends to slightly overestimate the magnitude of the standardized population difference on average, especially in small samples. Hedges’ g multiplies d by a degrees-of-freedom correction factor and reduces this small-sample bias. For large samples the difference is usually small. Document the variant and correction rather than mixing metrics without explanation. For basic interpretation, g remains a standardized mean difference and, like d, needs substantive context and an uncertainty estimate. Pair standardized values with the raw difference, group spreads and sample sizes. This keeps units, data volume and practical meaning visible rather than hiding them behind a dimensionless number.
Quanta exam set
Curated exam set for "Cohen's d Effect Size":
Question (front)
Which formula describes Cohen's d Effect Size?
Answer in your set
Question (front)
How do you rearrange Cohen's d Effect Size for Pooled standard deviation?
Answer in your set
Question (front)
Which common mistake happens with Cohen's d Effect Size?
Answer in your set
+ 8 more cards: units, variables, derivation, example, exam task
The formula set contains 11 cards. After importing, review them and start practicing; your ratings determine later review intervals.
This list names subject references and further reading. A page or section number appears when it is part of the reference; these are not verbatim quotations.
Create a curated FSRS exam set for d = (x̄₁ − x̄₂)/sₚ: formula recall, variables, derivation, rearrangement, worked example, common mistakes and exam context.
Curated formula set · LaTeX · FSRS spaced repetition · import from Essential