Regression power calculator

Fisher-information and Wald approximations for one predictor effect.

1. Outcome model

Outcome type Predictor type Two-sided alpha Target power Design effect Planned total N Reference prevalence (%)

Prevalence at x = 0 for a continuous predictor, or prevalence in the base category.

2. Continuous predictor

Analysis share of total N (%) Predictor SD Odds ratio per x unit Predictor R² with covariates

The predictor is approximated as normally distributed and centered at x = 0. Enter the effect and predictor SD on the same scale.

Dummy R² with covariatesEnter effects as odds ratios

Category	Share of total N (%)	Odds ratio vs base

The first row is the base. Each additional row is a separate category-versus-base coefficient. With 3+ categories, results are contrast-specific and the largest required N is highlighted; this is not an omnibus multi-df test.

Results

Required total N

—

Power at planned N

—

Expected events

—

Approximation only: large-sample Wald power, Fisher information, and one design-effect multiplier. Confirm consequential designs by simulation.