While the design standards for randomized trials are well-established and widely understood within the fields of epidemiology and medicine, the rigorous design of studies utilizing non-randomized (observational) data is not as common. In this talk, we will introduce the target trial framework (Hernan and Robins, 2016), which encourages researchers to first conceptualize an ideal randomized trial design for collecting data to estimate a causal effect of interest. Following this, the observational data are scrutinized and prepared for analysis with the aim of approximating the target trial as closely as possible. Understanding the value of the target trial framework hinges on recognizing that the causal effect estimated using a randomized design is equivalent to that estimated by a well-designed observational study. However, it’s important to note that the assumptions enabling causal estimation in observational studies are distinct from those in experimental designs. This presentation will review these assumptions and demonstrate how researchers can design observational studies to enhance the credibility of their causal inferences.
