State-of-the-Art Strategies for Addressing Selection Bias When Comparing Two or More Treatment Groups

The estimation of causal effects is one of the primary activities of most longitudinal research studies. Controlled experiments are held as the gold standard for estimating such effects. However, experiments are often infeasible and only observational data, in which participation in a program or intervention is out of the control of the researchers, are available for analysis. RAND statisticians and their colleagues have developed tools to implement new causal effect estimation methods that use observational data.

This course introduces causal modeling using the potential outcomes framework and propensity score weights to estimate causal effects from observational data. Presenters demonstrate how propensity score weights can be utilized to estimate intervention effects and how to evaluate balance before and after propensity score weighting and how to fit models with available software in Stata. Tools are also available in SAS, R, and Shiny. Those who take the course learn how to implement propensity score weighting using state-of-the-art methods and gain insights into some of the practical issues around evaluating the quality of propensity score weights for two or more treatment groups and time-varying treatments.

The recordings are from a full-day course presented at Cardiff University, and the content has not undergone formal quality assurance review. Funding for the course was generously supported by grants 1R01DA034065 and 1R01DA045049 from the National Institute of Drug Abuse.