Dunnett's Correction A/B/n Sample Size Estimator
Calculate the adjusted per-comparison alpha threshold and optimal traffic allocation when comparing k experimental treatments against a single baseline control using Dunnett’s procedure.
Count of test variants excluding the control (e.g. A/B/C/D is k = 3)
Maximum aggregate false positive risk across all variants
Planned sample size allocated to the control group
P-value threshold each variant must beat against control
How to Use This Calculator
- Enter the number of treatment variants (k) tested against control.
- Select your target family-wise error rate (0.05 default).
- Enter the planned sample size for the control group.
- Review the adjusted p-value threshold and optimal variant sample allocation.
Formula & Calculation Methodology
Because all treatments are compared against the same shared control, test statistics are correlated. Dunnett’s correction accounts for this correlation, providing more power than independent Bonferroni tests.
Worked Example
A/B/C/D test with 3 variants (k = 3) at α = 0.05 and 5,000 control sample
Testing 2 variants against control at α = 0.05
Frequently Asked Questions
Why not use standard Bonferroni correction (α / k)?↓
Bonferroni assumes all tests are completely independent. In A/B/n testing, comparisons share the exact same control group and are positively correlated; Dunnett’s test accounts for this, preserving higher statistical power.
Why should the control group have more traffic than each variant (N_c / N_v = √k)?↓
According to Dunnett’s optimal allocation theorem, because the control participates in every comparison, allocating more sample to control minimizes overall variance across all comparisons.
Does Dunnett’s test allow comparing variants to each other (B vs C)?↓
No. Dunnett’s test is strictly designed for comparisons against the single control (B vs A, C vs A). If you need all pairwise comparisons (B vs C), use Tukey’s HSD instead.
What is a limitation of testing too many variants at once?↓
As k increases, the required alpha threshold becomes very strict, necessitating dramatically larger overall sample sizes to achieve statistical significance.
Related Calculators
False Positive Risk Calculator for A/B Test Peeking
Open calculator →Sequential Testing Duration and Boundary Calculator
Open calculator →Minimum Reliably Detectable Effect (MRDE) Calculator
Open calculator →Sample Ratio Mismatch (SRM) Diagnostic Analyzer
Open calculator →Sample Size Calculator
Open calculator →Conversion Rate Calculator
Open calculator →Source Citation & Disclaimer
Informational tool only; not licensed tax, legal, or financial advice.