Power analysis for logistic regression (GLM)

Generalised linear models for a binary outcome (logistic regression, logit link). Use these when your response is yes/no, success/failure, or event/no-event.

Note

These pages are a recognition index — organised by the shape of the analysis, not by which MCPower feature they show off. If your outcome is continuous, you want OLS; if your data is grouped or clustered, see GLMM.

Examples

Simple logistic regression: one continuous predictor relapse ~ biomarker_level — one continuous predictor on a yes/no outcome, no covariates.
Logistic two-group comparison (binary predictor) remission ~ treatment — compare event rates between two groups (chi-square recast).
Logistic regression with a categorical predictor survived ~ habitat — a multi-level categorical predictor on a binary outcome.
Multiple logistic regression: covariate-adjusted employed ~ years_education + age + gender — a focal predictor on a binary outcome, covariate-adjusted.
Logistic continuous-by-continuous moderation relapse ~ biomarker_level * age — do two continuous predictors interact on a binary outcome?
Logistic treatment-by-moderator interaction (binary x continuous) remission ~ treatment * biomarker_level — a binary-by-continuous interaction on a yes/no outcome.
Logistic 2x2 factor-by-factor interaction voted ~ gender * urban — two binary factors interacting on a binary outcome.
Logistic three-way interaction germinated ~ light * moisture * temperature — a three-way interaction on a binary outcome.
Logistic regression: continuous predictor plus categorical control employed ~ experience_years + region — a continuous predictor plus a categorical control (parallel slopes).