Abstract:
|
In biomedical research, the gold standard for comparing outcomes in two or more groups remains the randomized clinical trial (RCT). However, oftentimes it is not feasible to conduct RCTs for reasons ranging from medical ethics to financial considerations. An increasingly-popular alternative is to conduct observational studies instead. The propensity score (PS) plays a fundamental role in bias control in observational studies. When PS-matching is the analytical approach used, it is important to ensure that adequate baseline covariate balancing has been achieved. Based on both simulated and actual data, we discuss formal tools for covariate balancing assessment.
|