Introduction to Causal Inference

Catalog description

Many social science and scientific inquiries can be framed as causal questions. Does a new cancer treatment cause a reduction in mortality? Do financial grants cause students to do better in college? Does a new public policy cause an increase in voter turnout? When tackling these questions, we frequently come across the phrase "correlation does not imply causation." If that's the case, then what does imply causation? In this course, we will discuss causal inference methods for measuring causal effects of different interventions (e.g., drug treatments, financial grants, and public policies). First, we will discuss how experiments and #8212;-where interventions are randomized among subjects and #8212;-can imply causation when an appropriate experimental design and statistical analysis is used. Then, we will discuss how observational studies and #8212;-where interventions are not randomized and #8212;-can also imply causation when approaches like propensity score methods, matching, and doubly robust estimation are employed. Finally, we will discuss instrumental variables and regression discontinuity designs and #8212;-which are frequently used in medicine and public policy for establishing causal inferences. Throughout we will use R to conduct causal analyses. A working knowledge of regression is encouraged, but regression will also be discussed and taught during much of the course. Prerequisites: 36-218 Min. grade C or 36-219 Min. grade C or 21-325 Min. grade C or 21-425 Min. grade C or 36-235 Min. grade C or 15-259 Min. grade C or 36-225 Min. grade C