Power & sample size

A power calculator specifically for linear and binary logistic regression models, with either a continuous or categorical predictor.

There’s a million power and sample size calculators out there for all manner of basic statistical tests, but I haven’t found any for basic regressions that are intuitive to use, so I made one.

What you can explore

Set effect size, significance level, target power, and planned sample size. Adjust the design effect, analysis share, and correlation with covariates to reflect assumptions.

You can work in odds ratios and log-odds coefficients when messing with binary logistic models.

How it works

The calculator uses Fisher information and large-sample Wald approximations to estimate power/needed effect size. With 2+-level categorical predictors, results are separate category-versus-base contrasts, not an omnibus test. Consequential study designs should be confirmed by simulation.

Calculations run in your browser. No account or upload is required.