Analysis, Validity and Inference of Observational Epidemiology, 2 op

This intensive, 4-day seminar course will introduce a variety of practical analytical approaches for addressing threats to validity in observational studies. Specific topics will include the limitations of traditional hypothesis testing in observational epidemiology, modern approaches to detecting and attenuating confounding, probabilistic correction for variable misclassification, and handling of missing data. These topics will be taught as non-technically as possible, but references and programming resources will be provided to allow for rigorous application of the methods. Throughout the course, emphasis will be placed on how to successfully incorporate these methods into real-world studies and subsequent publications. Examples using SAS R and Microsoft Excel will accompany many of the lectures. Students are expected to have a solid foundation in epidemiologic study design and biostatistics, including the interpretation of linear, logistic, and proportional hazards regression models, and will be asked to read two to three papers each night in preparation for the next day’s class. Examples using SAS R will focus on demonstrating the mechanics of implementing the methods, and will not assume familiarity with SAS R syntaxprogramming. Equivalent functions in Stata and R SAS will be pointed outdiscussed whenever applicable. Lecture content will be reinforced by afternoon laboratory sessions, in which students can apply the methods taught learned in lecture to the analysis of simulated data sets.