This spring I am teaching Cornell’s Comparative Methods course. The near-final syllabus is here (PDF).
(To those Cornell PhD students reading this: hi! I’ll see you next Wednesday.)
Compared to the previous time that I taught this course, I am doing four things differently. First, I am cutting down on the meta-debate about positivism, empiricism, and various alternatives. Students can have those discussions elsewhere, and time spent debating epistemology distracts from other course goals.
Second, I am taking a firmer line on what we think statistical analysis is good for. Now, the basic question is not “how do we summarize correlations between variables?” but rather “what would have to be true for a regression to be useful?”
Third, I am treating multimethod research more seriously, and critically. In fact, I suspect that the most challenging readings are those on how to integrate quantitative and qualitative methods.
Fourth, I am introducing a replication assignment. This assignment will differ in form and intent from what I understand to be the standard replication assignment in political science, described here. The purpose of the assignment is not to ascertain whether or not statistical tables can be reproduced, but to expose students to working with primary sources and building arguments from them. The assignment has a learning goal, not a disciplinary function. It is for this reason that I have no interest restricting the replication assignment to statistical work only. Less “science police,” more understanding and building on the research of others.
Perhaps other replication assignments have similar goals, but that is not how they are frequently portrayed.