Miikka Rokkanen

Dec 2014

Professor Rokkanen certainly has good intentions, but his section was not effective in conveying the material. Granted, this was his first semester teaching W3412, and the course will probably change in the future if he happens to teach it again, so take my review with a grain of salt. For clarity's sake, let me separate the professor from the course itself. ----- As a lecturer, Rokkanen is enthusiastic but oftentimes unclear. He does occasionally joke about his spoken English, but this really isn't a problem. The problem comes when he's trying -- and believe me, he's trying very earnestly -- to explain the intuition behind the methods. I personally found his explanations unhelpful, as they didn't so much provide intuition as rephrase what was on the slide with less jargon (case in point: his lecture on IV estimation). What that means is you'll probably have to read the Stock-Watson textbook a good amount, or be prepared to google a lot, if you want to hone your intuition, which really is the whole point of introductory metrics. His lecture slides are structured, organized, and overall decent, although I would recommend taking a look at Arkonac's slides, which tended to be less organized but provided more intuition. Remember: intuition is key and the ultimate goal of this course. Onto the course itself, e.g. the homework, the exams, the pacing, etc. It was odd -- and by odd I mean unfortunate -- that all of our problem sets and exams were borrowed from Arkonac's course. Practically speaking, I felt like we were less prepared in completing the problem sets than students from Arkonac's section. Each professor allocates their lecture time on different topics differently, and it was frustrating having to search for a derivation that was absent from his slides (incidentally, they were often included in Arkonac's). There are tons of reviews on Arkonac's problem sets, but let me add a few comments. The homework is long and tedious. Watching anyone hunched over their computer, filling out those dreaded tables while simultaneously trying to use STATA is a sad sight to behold. For the average student, I'd estimate them to take roughly 4-8 hours to complete. It sucks, it really does. True/False can be tricky, and grading is not especially vicious, but it's easy to lose points if you answer a question incorrectly... well, no sh*t. My advice is plain and simple: suck it up. The exams test you on concepts similar to what was on the problem sets, and the hours of completing them will help internalize what you can probably expect to show up at least somewhere on the midterm/final. The feeling in your stomach (believe me, it's not a good feeling) when you're working on the exam is the same feeling you get when you're working on the problem set. Do the homework. A few last remarks on the curriculum. For those of you with no background in linear regression, there's a ton of material that is covered in this course, and 100% of it is predicated on the mutual understanding between the department and yourself that you're up-to-speed on basic probability and statistics. Know what conditional probability is. Know what an estimator is, really. It's nothing too advanced, but coming in with poor preparation can cause you much more trouble along the way. Would I recommend? I would recommend Rokkanen for introductory metrics if his problem sets and exams are, in the future, tailored to his own lectures and expectations. If not, choose Arkonac.