When a randomized trial is not possible for evaluating the effectiveness of a new treatment, several alternatives have been proposed. Two of these methods are difference-in-differences (DID) analysis and the stability controlled quasi-experiment (SCQE). DID allows for estimation of causal effects with an assumption of "parallel trends": the trend in average non-treatment outcomes are the same between treated and comparison groups. SCQE relies on an assumption of the outcome's "baseline trend": the change between cohorts is the same in the overall non-treatment outcome. We compare these two methods under a range of baseline trend assumptions. We also evaluate the methods' reliabilities in protecting against false inferences and over-confidence. Our application is the effect of a placebo health policy, which is known to have no true effect, on 30-day mortality rate among patients treated for heart attack, heart failure, and pneumonia in U.S. hospitals.
Cookie SettingseScholarship uses cookies to ensure you have the best experience on our website. You can manage which cookies you want us to use.Our Privacy Statement includes more details on the cookies we use and how we protect your privacy.