Statistics for Policy Analysis II
The world is full of causal claims, and many policy debates boil down to disagreements about the validity of these claims. Because of this, most quantitative research attempts to understand causality. This course is designed to introduce students to the basic techniques used by researchers to identify causal relationships from data. In doing so, they will gain the ability to understand and produce high quality quantitative evidence on whatever topics interest them.
The course itself will consist of two main components. We will first spend some time carefully defining what researchers usually mean by "causality" and discussing the diffculties of estimating causal relationships in data. We will then use this discussion to motivate the technique that is often referred to as the "gold standard" of causal inference: random control trials (RCTs). We will spend a significant amount of time discussion complications that arise when analyzing a seemingly simple RCT, in part because RCTs are increasingly used in research and because most of the alternative techniques we learn attempt to "mimic" an RCT. The second main component concerns a range of techniques that researchers use when randomizing treatments are impossible and/or impractical. These include matching estimators, regression analysis, regression discontinuities (RDs), and the use of panel data.
By the end of the course, everyone should have the ability to:
- Read, understand, and assess the quality of quantitative causal research.
- Develop appropriate empirical strategies to tackle interesting causal questions.
- Implement these strategies using statistical software.