One of the most difficult aspects of social science is the fact that we don't often get to test policy in a lab, we have to use real-world data and try and isolate effects from all of the noise. Fortunately we have become pretty sophisticated at doing so and even more adept at designing policy implementation in a way that gives us some experimental evidence. Today in class we will discuss data analysis and different ways of teasing out causality and the necessary caution one must have in interpreting results.
As an example we will think about perhaps the most commonly used example: the education - earnings relationship.
We will also study the empirical evidence on class size and discuss the STAR study and it limitations.
In other words, what do these two graphs tell us and how much should we trust conclusions based on these pieces of evidence?