Sample Size Calculator for Animal Studies

Find how many animals you need for a two-group comparison, a before/after design, a one-way ANOVA or a comparison of two percentages, optionally with a second factor such as sex, and get a justification paragraph ready for your animal use protocol. Everything runs in your browser; nothing you enter is stored or sent anywhere.

Study design

Add a second factor, such as sex (two-way ANOVA)

Expected effect

Use pilot data or published studies. The result is only as good as this estimate.

Mean, control groupMean, treated groupStandard deviationWithin each group (pooled)Units (optional)

Settings

Significance level (α)PowerTest directionExpected attrition (%)Animals lost to surgery, misplacement, exclusion

Result

Animals per group

34 in total across 2 groups

With 10% attrition

animals per group · 38 total to order

Effect size

Cohen's d, large

Actual power with 17 animals per group: 80.7%.

Power by animals per group (dashed line: target 80%)

Justification for your protocol

Paste this into your animal use protocol or grant. Edit the source of your estimate first.

Your estimate is based on

Sample size was determined by an a priori power analysis for a two-tailed independent-samples t-test (α = 0.05, power = 80%). Based on pilot data from our laboratory, we expect a difference of 25 between groups (100 vs. 75) with a standard deviation of 25 (Cohen's d = 1). This requires 17 animals per group (34 in total). To allow for an expected attrition of 10%, we will use 19 animals per group (38 in total).

How the numbers are calculated

t-tests and ANOVA use exact power calculations based on the noncentral t and F distributions, the same method as the R package pwr. We checked the results against published reference values: for example, a medium effect (d = 0.5) needs 64 animals per group for 80% power.

Two percentages are compared with the normal approximation and a continuity correction (Fleiss), which is slightly conservative.

Effect sizes: Cohen’s d is the difference between group means divided by the standard deviation. For before/after designs, dz is the mean change divided by the standard deviation of the changes. For ANOVA, Cohen’s f is the standard deviation of the group means divided by the within-group standard deviation. With a second factor such as sex, the design becomes a two-way ANOVA: f is worked out separately for the main effects and the interaction from the group means, and each is tested with the two-way ANOVA F test, assuming equal group sizes.

A power analysis is only as good as the effect size you expect, so base it on pilot data or published studies. Detecting an interaction, such as whether a treatment works differently in males and females, usually takes several times more animals than detecting a main effect. For designs this calculator doesn’t cover, such as several repeated time points or mixed models, consult a statistician.