Power Calculator

Interactive sample size and power curves for the most common study designs in clinical and life-science research

Purpose

The Power Calculator is the workhorse app of the site: a single interface that computes sample size, power, detectable effect, or significance level for any of the common designs a clinical or life-science researcher encounters, and draws the sensitivity of each quantity to the others as a curve.

User inputs

• Design family (means, proportions, correlations, ANOVA, regression, survival, equivalence)
• Specific test within the family
• Three of the four power quantities, with the fourth to solve for
• Effect-size input mode: standardised, raw, or from pilot summary statistics

Outputs

• The solved quantity with a bolded summary sentence (“you need \(n = 64\) per arm to detect a difference of 5.0 with 80% power at \(\alpha = 0.05\)”)
• A power curve: power as a function of total sample size, with the current point highlighted
• A secondary curve: power as a function of effect size for the current \(n\)
• The exact R code that would reproduce the calculation with pwr, pwrss, or WebPower
• A downloadable PDF report with all inputs, results, and curves, suitable for a grant application

Didactic value

Beyond its utility as a calculator, the app teaches the trade-off structure that governs every study design: doubling the sample size roughly halves the detectable effect, but the relationship is not linear; crossing from 80% to 90% power has a steeper cost than going from 70% to 80%; unbalanced allocation costs power relative to 1:1.

Embedded in

• Every tutorial under sample-size/
• clinical-biostatistics/rct-design.md
• clinical-biostatistics/non-inferiority-trials.md

Source code

Local: apps/05-power-calculator/

Run with:

shiny::runApp("apps/05-power-calculator")