d·r·η²

Effect Size Calculator

Calculate Cohen's d, Pearson r, and eta-squared (η²) with magnitude labels, step-by-step working, and automatic cross-conversions between all three metrics.

Cohen's d Pearson r Eta-Squared η² Small / Medium / Large
Share this tool

Effect Size Type

Two Independent Groups

Group 1

Group 2

Paired Samples or One-Sample

M − μ₀ for one-sample; mean of diff scores for paired

Correlation Coefficient

Enter the raw correlation coefficient from your analysis

ANOVA Sums of Squares

Effect (between-groups) SS from ANOVA table

Error/residual SS from ANOVA table

Quick Load:
Calculating effect size…
d·r·η²

Select an effect size type, enter the values, and click Calculate Effect Size.

Use Quick Load to see a worked example instantly.

Effect Size on a Distribution Curve

A larger Cohen's d means less overlap between two groups — the effect is easier to observe in practice.

Group 1 Group 2 d = effect size Group 1 (control) Group 2 (treatment) ← Larger d = less overlap = greater, more detectable difference

What Is Effect Size?

Effect size is a standardised measure of the magnitude of a phenomenon — how large an effect is, independent of sample size. Unlike a p-value (which only tells you whether an effect exists), effect size tells you how meaningful or practically important the effect is.

Two studies can both produce p < 0.05 yet have very different effect sizes. A drug trial with 10,000 participants might detect a statistically significant improvement that is clinically meaningless (d = 0.02), while a smaller study might find a large, practically important effect (d = 0.90). Effect size bridges the gap between statistical significance and real-world importance.

The American Psychological Association (APA 7th edition) requires effect sizes to be reported alongside all inferential statistics. Cohen's d, Pearson r, and eta-squared (η²) are the three most widely used metrics across social science, psychology, medicine, and education research.

Cohen's d: Standardised Mean Difference

Cohen's d expresses the difference between two group means in standard deviation units. It is the most common effect size for t-tests and mean comparisons.

d = (M₁ − M₂) / SD_pooled

For paired samples or one-sample tests, use:

d = M_diff / SD_diff
Magnitude|d| rangeTypical meaning
Negligible|d| < 0.2No practically meaningful difference
Small0.2 ≤ |d| < 0.5Detectable but modest difference
Medium0.5 ≤ |d| < 0.8Noticeable difference in context
Large|d| ≥ 0.8Substantial, easily observable difference

How to Calculate Pooled Standard Deviation

For independent samples, the pooled SD weights each group's variance by its degrees of freedom:

SD_pooled = sqrt(((n₁−1)×SD₁² + (n₂−1)×SD₂²) / (n₁+n₂−2))

Example: Group 1 (M=50, SD=10, n=30) vs Group 2 (M=42, SD=12, n=28):

SD_pooled = sqrt((29×100 + 27×144) / 56) = sqrt(121.21) ≈ 11.01
d = (50 − 42) / 11.01 ≈ 0.727 → Medium effect

Pearson r as an Effect Size

When a study reports a correlation, the Pearson r coefficient is itself an effect size. Cohen's benchmarks for r differ from those for d:

Magnitude|r| rangeTypical meaning
Negligible|r| < 0.10No meaningful relationship
Small0.10 ≤ |r| < 0.30Weak relationship
Medium0.30 ≤ |r| < 0.50Moderate relationship
Large|r| ≥ 0.50Strong relationship

r² (eta-squared) tells you the proportion of variance in one variable explained by the other — r = 0.35 means 12.25% of variance is shared between the two variables.

Eta-Squared (η²) for ANOVA

Eta-squared (η²) is the proportion of total variance in the outcome that is attributable to the between-groups factor. It is read directly from the ANOVA table:

η² = SS_between / (SS_between + SS_within)

Cohen's benchmarks for η² are lower than those for d or r because raw proportions of variance are inherently smaller numbers:

Magnitudeη² rangeTypical meaning
Negligibleη² < 0.01Less than 1% variance explained
Small0.01 ≤ η² < 0.061–6% variance explained
Medium0.06 ≤ η² < 0.146–14% variance explained
Largeη² ≥ 0.1414%+ variance explained

Converting Between Effect Size Metrics

Different journals and disciplines prefer different effect size metrics. The calculator automatically converts between all three. The formulas used are:

d → r

r = d / sqrt(d² + 4)

Cohen (1988) equal-n approximation

r → d

d = 2r / sqrt(1 − r²)

r → η²

η² = r²

Exact when η² is the squared correlation

η² → r → d

r = sqrt(η²) then d = 2r / sqrt(1−r²)

Why Effect Size Matters in Research

A p-value answers "Is this effect real?" — effect size answers "Is this effect worth caring about?" Both questions matter. Large samples can produce tiny p-values for trivial effects; small samples can miss large effects entirely.

Power analysis also requires effect size. Before collecting data, researchers must estimate the expected effect size to determine how large a sample they need. Reporting effect sizes in published studies allows future researchers to use those estimates in their own power calculations — compounding the value of each study.

Meta-analyses aggregate effect sizes across studies to reach more reliable conclusions than any single study can provide. Without standardised effect sizes, meta-analysis is impossible.

Effect Size Calculator FAQ

What is effect size in statistics?
Effect size is a standardised, unit-free measure of the magnitude of an observed phenomenon. Unlike a p-value, which only indicates whether an effect is statistically detectable, effect size tells you how large the effect is in practical terms — making it essential for interpreting the real-world importance of research findings.
What is a good effect size for Cohen's d?
Cohen proposed benchmarks: d = 0.2 (small), d = 0.5 (medium), d = 0.8 (large). These are rough guidelines, not rigid rules. What counts as a meaningful effect depends on the field and context. In medical research, even d = 0.2 can be clinically important if the outcome is serious. In educational research, d = 0.4 is often considered meaningful.
What is the formula for Cohen's d?
For two independent groups: d = (M₁ − M₂) / SD_pooled, where SD_pooled = sqrt(((n₁−1)SD₁² + (n₂−1)SD₂²) / (n₁+n₂−2)). For paired samples or one-sample tests: d = M_diff / SD_diff. Both formulas express the mean difference in standard deviation units.
What is the difference between Cohen's d for independent and paired samples?
For independent samples, d uses a pooled standard deviation that averages variance across both groups, weighted by degrees of freedom. For paired samples, d uses the standard deviation of the difference scores — which is usually smaller because correlated measures share variance, producing a larger d for the same raw difference.
How do I convert Cohen's d to Pearson r?
Use r = d / sqrt(d² + 4). This is Cohen's (1988) equal-n approximation. For example, d = 0.5 gives r = 0.5 / sqrt(4.25) ≈ 0.243. The inverse is d = 2r / sqrt(1 − r²). These conversions assume equal group sizes; when groups are unequal, a sample-size-weighted version is more precise.
What does eta-squared measure?
Eta-squared (η²) measures the proportion of total variance in the dependent variable that is explained by the independent variable (factor). η² = 0.10 means the factor accounts for 10% of all variance in the outcome. It ranges from 0 to 1 and is extracted directly from ANOVA sums of squares.
Is effect size the same as statistical significance?
No. Statistical significance (p-value) tells you the probability of observing your data if the null hypothesis were true. Effect size tells you how large the observed effect is. A study can be statistically significant with a trivial effect size (large sample, tiny real difference), or statistically non-significant with a large effect size (small sample, genuine large difference).
What effect size should I report in APA format?
APA 7th edition requires reporting effect sizes for all inferential statistics. For t-tests, report Cohen's d. For ANOVA, report η² or partial η². For correlation analyses, Pearson r is itself the effect size. Always report which benchmark convention you used (e.g., Cohen's 1988 benchmarks) and acknowledge that benchmarks are context-dependent.