December, 2011
Propensity score methods are being increasingly used as a less parametric alternative to traditional regression to balance observed differences across groups in medical and health policy research. Data collected in these disciplines are often multilevel in analytically relevant ways. The propensity score, however, has been developed and used primarily with unstructured data. We present and compare several propensity-score-weighted estimators in the context of hierarchically structured data, including marginal, cluster-weighted and doubly-robust estimators. Using both analytical derivations and Monte Carlo simulations, we illustrate bias arising when the usual assumptions of propensity score analysis do not hold for clustered data. We show that exploiting the multilevel structure, either parametrically or nonparametrically, in at least one stage of the propensity score analysis can greatly reduce these biases. These methods are applied to a study of racial disparities in breast cancer screening among beneficiaries in Medicare health plans.
Keywords: multilevel, propensity score, racial disparity, treatment effect, unmeasured confounders, weighting.
The manuscript is available in PDF formats.