A statistical method that pairs treated and untreated subjects with similar baseline characteristics to mimic randomisation.
Propensity matching (short for propensity score matching) estimates each subject's probability of receiving a given exposure or treatment based on measured covariates, then pairs treated and untreated subjects with similar scores to create more comparable groups. It is used in observational studies to reduce confounding when a randomised controlled trial isn't feasible or ethical. It can only balance groups on the covariates that were measured and included in the model, so unmeasured confounding remains a risk and causal claims should be made cautiously.
This guide was auto-drafted and is pending editorial review.