This guide provides resources related to power analysis and sample size calculations that are applicable to a broad audience from a clinical researcher looking for introductory education material to a statistician looking for a specialized calculator. Resources include:
Introductory Reference Material
Validated Calculators (Free Online Tools)
Note: Ensure reproducibility of your calculations; take pictures or screenshots and properly cite.
UCSF offer online and easy to use calculators of sample size for a variety of analyses with binary, continuous, and time to event measurements, as well as pediatric growth and posterior probability of disease.
Online and easy to use calculator of sample size for analyses of binary, means, and time to event measurements.
Online tools to estimate sample size for superiority, equivalence, and non-inferiority clinical trials for binary and continuous measurements. On the webpage, scroll up to the ‘POWER CALCULATORS’ menu at the top right.
U of Colorado Denver offers this online sample size tool with guided steps for a wide range of general linear mixed models with normally distributed errors.
Downloadable Software (Free and For Purchase)
Note: Ensure reproducibility of your calculations; take pictures or screenshots and properly cite.
Free downloadable software that is easy to use with a helpful manual. Supports broad range of calculations for proportions, means, and linear, logistic, and Poisson regression.
Free downloadable software with a manual to calculate sample size for binary, continuous, and time to event measurements.
For purchase software that is easy to use for power and precision analysis. Supports a very broad range of calculations for various hypotheses and scenarios involving proportions, means, correlations, and survival times, and linear, polynomial, and logistic regression.
For purchase software that is easy to use with helpful manual for more than 1,000 statistical tests and confidence interval scenarios.
For purchase software that is easy to use with a manual. Supports broad range of calculations for proportions, means, correlations, survival times, Bland-Altman limits of agreement, and ROC curves.
Free software that requires statistical programming in R. Supports calculations for proportions, means, and correlations.
For purchase software that requires statistical programming in SAS. Supports a variety of calculations for analyses involving binomial proportions, means, correlations, multiple regression, and survival curves.
This Hoenig and Heisey (2001) paper in The American Statistician discusses the fundamental flaws with post-experiment power calculations.
This Levine and Ensom (2001) paper in Pharmacotherapy explains how the logic underlying post-hoc power analysis is fundamentally flawed.
This Jiroutek and Turner (2017) paper in The Journal of Clinical Hypertension is easy-to-read and describes why retrospective power analyses are useless and nonsensical.