Matching With Doses in an Observational Study of a Media Campaign Against Drug Abuse

Abstract
Multivariate matching with doses of treatment differs from the treatment-control matching in three ways. First, pairs must not only balance covariates, but also must differ markedly in dose. Second...