Demystifying causal inference: ingredients of a recipe

In the last few decades, scholars have contributed to a flourishing literature on casual inference and the demand for its application in areas like programme evaluation has increased. Our suggestion is that the following ingredients are useful for demystifying causal inference in introductory courses: (1) using the potential outcomes and causal graph frameworks, (2) covering applications with real data that use key methods for causal inference: experiments, regression discontinuity etc., (3) using Monte Carlo simulation, and (4) using data graphs. The first two ingredients are components of the scholarship in causal inference, while the latter two are more general ingredients of statistical and econometric pedagogy. We discuss the case for these ingredients, drawing on the substantive and pedagogical literature, our experience, and student opinions.