Effect Size Calculator Explained: Interpreting Small, Medium, and Large Effects

How to Use an Effect Size Calculator: Step-by-Step Guide for Researchers

1. Choose the correct effect size measure

• Cohen’s d: two-group mean differences (independent or paired).
• Hedges’ g: like Cohen’s d but corrects small-sample bias.
• Pearson’s r: correlation strength between two continuous variables.
• Eta-squared / Partial eta-squared: effect size for ANOVA.
• Odds ratio / Risk ratio: binary outcomes in contingency tables.

2. Gather required inputs

• Means and SDs: group means and standard deviations (for d/g).
• Sample sizes: n for each group (or total n).
• t or F statistics: use if raw means/SDs aren’t available.
• Contingency table counts: cell frequencies for odds/risk ratios.
• Correlation coefficient r and n: for Pearson’s r conversions.

3. Select the calculation method (formula or test statistic)

• If you have means, SDs, and ns: compute Cohen’s d = (M1−M2) / Spooled, where Spooled = sqrt[ ((n1−1)S1^2 + (n2−1)S2^2) / (n1+n2−2) ].
• If you have t and n: d = tsqrt(1/n1 + 1/n2).
• For Hedges’ g: apply small-sample correction J = 1 − (3 / (4df − 1)); g = J * d.
• For ANOVA: eta-squared = SSbetween / SStotal; partial eta-squared = SSbetween / (SSbetween + SSwithin).
• For correlations: Fisher Z-transform for CI; convert r to d if needed: d = 2r / sqrt(1−r^2).

4. Enter values in the calculator

• Input numbers in the matching fields (means, SDs, ns, t, F, cell counts, or r).
• Choose independent vs paired designs where applicable.
• Select two-tailed vs one-tailed if calculator offers CIs or p-related outputs.

5. Interpret the output

• Effect size value: magnitude (positive/negative indicates direction for mean differences or correlation).
• Common benchmarks (use cautiously):
  • Cohen’s d: small ≈ 0.2, medium ≈ 0.5, large ≈ 0.8.
  • Pearson’s r: small ≈ 0.1, medium ≈ 0.3, large ≈ 0.5.
• Confidence intervals: assess precision—wide CIs indicate uncertainty.
• Bias correction: prefer Hedges’ g when sample sizes are small.

6. Report results clearly

• State the effect size measure, value, and CI (e.g., “Cohen’s d = 0.45, 95% CI [0.12, 0.78]”).
• Mention sample sizes and whether data are independent or paired.
• If converting between metrics, note the conversion method used.

7. Practical tips

• Use raw means/SDs when available rather than relying on test statistics.
• For skewed or non-normal data, consider robust effect sizes or bootstrapped CIs.
• Always report both p-values and effect sizes; effect sizes convey practical importance.
• When comparing studies, ensure the same effect metric and comparable sample contexts.

Quick worked example

• Independent groups: M1=5.2, SD1=1.1, n1=30; M2=4.6, SD2=1.3, n2=28.
  • Spooled = sqrt(((29)(1.1^2)+(27)(1.3^2))/(56)) ≈ 1.205
  • d = (5.2−4.6)/1.205 ≈ 0.50 (medium)

Resources and calculators

• Use verified tools that allow selection of measure and report CIs and corrections.