The most confounding experience of my year spent at UNC was most likely the difference in scientific work in the USA as compared to Germany (if not to say Europe). Where during my studies in Potsdam, Berlin, Salzburg and Paris (my point is that this is not something peculiar to my home university) of political science in Europe formal analysis was virtually nonexistent (I believe there is one auditorium course in Potsdam on game theory), in my post-graduate classes at UNC every article centered on some kind of more or less complicated model. I always found this rather dubious, even if I felt I simply did not understand most models well enough to be able to criticize them. Stephen Walt has attacked this American obsession with formal analysis. Read the article, it is highly informative.
Honestly, I kind of liked game theory - I didn't excel at it, but then my first (and only) course was a post-graduate one, which might not have been such a good idea - but I feel that it is dangerous to concentrate too heavily on it a tool for analysis. As usual some middle ground between the American purely mathematical and the French purely argumentative path should be found. Yet, I most definitely want to do another course (get a book, I'm out of university after all) on this subject, simply to understand most American political scientists better.