Inverse probability weighting assigns higher weight to underrepresented combinations of treatment and covariates. If young patients are under-treated relative to their share in the population, IPW up-weights them so the weighted sample reflects the target population.

IPW uses the same propensity score as matching but reweights the full sample instead of discarding unmatched units.

When to prefer IPW

IPW preserves sample size and handles continuous or multi-valued treatments more naturally than one-to-one matching. It can be unstable if propensity scores are very close to 0 or 1 — trimming or doubly robust methods help.