The analysis of causal effects has always been in the center of interest of social scientist, but conventional wisdom held that correlation is not causation, meaning that using observational data, only association may be established but not causation. This position is changing and the course will explain, why the mode of data collection (observational study or designed experiment) is crucial in the possibility of determining causal effects and will discuss the two most important proposals to establish causality, based on observational data. Rubin’s method is based on the counterfactual framework and uses propensity score based matching to estimate the average causal effect. while Pearl’s approach is based on directed graphical models to describe and test causal relationships.
This is an Advanced Methods course, open to ALL tracks.