Applying the Causal Roadmap to Single Ti

This course will introduce students to the theoretical and practical aspects of applying a “causal roadmap” to research questions in epidemiology using both single timepoint and longitudinal data. A causal roadmap approach to empirical investigation is intended to strengthen transparency and clarity in the research process and typically consists of several steps including: 1) formulating a research question, 2) translating it into a causal quantity, 3) listing the assumptions required to identify this causal quantity from the data, 4) choosing an estimation approach, and 5) doing the analysis. We will learn single timepoint and longitudinal g-computation/ standardization, inverse-probability-of-treatment weighting (IPTW), and doubly robust estimation approaches (e.g., targeted minimum loss-based estimation (TMLE)). The final class will include integrating machine learning into the estimation approach. Each module will include hands-on exercises in R in which we will apply the estimation approaches to data. Data for each analysis exercise will be provided by the instructor. For the final project, students can choose to use data provided by the instructor or data for which they already have access.