This course is the second semester in the statistics sequence for political science and public policy offered in the Political Science Department at MIT. The intellectual thrust of the course is a presentation of statistical models for estimating causal effects of variables. The model of an effect is a conditional mean (though we might imagine other effect). The notion of causality is the effect of one variable on another holding all else constant. The course develops this idea through three statistical models. First, I present the multivariate regression model as a means of holding other variables constant within the analysis. Specific issues are considered, such as measurement error and qualitative dependent variables. The most important difficulty is the problem of omitted variables. After presenting the multivariate regression model the course moves into two methodological approaches for implementing quasi-or natural experiments to avoid the problem of omitted variables. The second part of the course presents panel models and designs as a way of removing the effects of omitted variables. The third part of the course presents instrumental variables as a way of making a variable of interest independent of the omitted variables. I will also teach you rudimentary matrix algebra. That is essential for reading and understanding more advanced statistics texts and it helps you develop strong capacity to think about data designs and problems. The requirements of this course are:
Weekly problem sets. These will be a mix of empirical problems and theoretical problems.
Midterm and final exams.
Final paper. This should be a reasonably short research note or study; say 15 to 20 pages. We may do this one of two ways. First, each of us could establish a particular project of interest. Second, the class as a group could try to tackle a general problem, with each of us doing papers on a common theme. I would recommend studies of electoral competition. The data are rich and readily available.