Loading the page...
Preparing tools and content for you. This usually takes a second.
Preparing tools and content for you. This usually takes a second.
Fetching calculator categories and tools for this section.
Free effect size calculator for standardized mean differences. Compare independent groups with pooled SD, or quantify a one-sample mean vs μ₀—with optional Hedges’ g and Cohen-style magnitude labels.
Last updated: April 13, 2026
Need a reporting toolkit for your research product? Get a Quote
Uses the pooled SD from both samples (equal-variance style Cohen’s d). Sign: M₁ − M₂ relative to the pooled scale.
Cohen’s d
-0.5
Medium (≈ 0.5–0.8)
Hedges’ g
-0.4901
Bias-corrected when df ≥ 2
Pooled SD (sp)
15
Degrees of freedom (for g)
38
Cohen’s benchmarks (rough guides)
|d| ≈ 0.2 small, 0.5 medium, 0.8 large — context matters; do not treat them as universal cutoffs.
Enter both means, SDs, and sample sizes. The tool reports sp, d = (M₁−M₂)/sp, and g when the correction is stable.
Use your sample mean, reference mean, sample SD, and n—ideal after a one-sample t-test or power planning exercise.
When df ≥ 2, g = J·d with J ≈ 1 − 3/(4df − 1). For very small df, g may be omitted to avoid misleading corrections.
Positive vs negative d shows which mean is larger (two-sample) or whether you sit above/below μ₀ (one-sample). Labels use |d| for magnitude.
Automatic “small / medium / large” hints follow common textbook cutoffs—always contextualize in your discipline.
After computing d here, compare with the site’s t-test calculator and variance summaries for a full picture of signal and spread.
Two groups: M₁ = 100, M₂ = 107.5, s₁ = s₂ = 15, n₁ = n₂ = 20
Cohen’s d ≈ -0.5
Pooled SD = 15, so the mean gap is half a pooled SD—a “medium” benchmark in Cohen’s rough scheme.
Inputs are validated as finite numbers; sample sizes must be integers of at least 2. Two-sample d uses the pooled variance estimator with df = n₁+n₂−2. One-sample d divides the mean difference by the sample SD with df = n−1. Hedges’ correction follows a standard approximation when df is large enough to behave sensibly. Explore more on our Math & Science calculators index.
Two-sample: sp² = ((n₁−1)s₁² + (n₂−1)s₂²)/(n₁+n₂−2), d = (M₁−M₂)/spOne-sample: d = (M − μ₀)/sHedges: g = J·d, J ≈ 1 − 3/(4df−1) when df ≥ 2Continue with t-test calculator and variance calculator.
Get Custom Calculator for Your PlatformShare it with anyone reporting t-tests or meta-analyses
Suggested hashtags: #EffectSize #CohensD #Statistics #Research #MetaAnalysis