A Challenge to Greg Mankiw

by Chad Aldeman on August 31, 2009

in Undergraduate Education

Gregory Mankiw is a professor of economics at Harvard University, and he runs a fairly popular blog. Today he writes about the difficulty in selecting 15  applicants for his freshman seminar from a pool of 200 qualified applicants:

I am teaching a Harvard freshman seminar this semester (in addition to ec 10), and one of my first tasks is to choose the 15 students. About 200 applied. That means that getting into my seminar is about as hard as getting into Harvard–except that you first have to get into Harvard before you can even apply!

Having spent much of yesterday reading through the applications, I fully recognize how difficult and somewhat random such admissions processes are. I could fill almost the entire seminar with kids with perfect SAT scores (2400), but I won’t, as there is more to life than test scores. I am looking also for passion about the subject, interesting life experiences, and a balance among the group of students to promote good discussion. But judging that from a few brief essays is very, very hard. To those students I do not pick: I am sorry, and it is my loss as well for not having the opportunity to get to know you better.

Here’s a challenge to Professor Mankiw: if the process is random, and you have a hard time telling one perfect-SAT student from another, take the opportunity and show Harvard that they probably can’t either. Select what you think are the top 7 applicants, and then pull 8 randomly from the pile. Grade them fairly and anonymously all semester, and then, at the end, compare the final grades with your selections. Do the randomly chosen students do better or worse than the ones you hand-picked? If you can’t find a difference, show the results to your university and encourage them to adopt a random lottery system for the coveted freshmen spots. I’m betting there won’t be much of a difference, and Professor Mankiw’s freshmen seminar seems like the perfect place to test it.

{ 3 comments }

Liora December 11, 2009 at 7:56 pm

Interesting logic; but harvard freshman seminars are pass-fail.

Crimson Wife September 8, 2009 at 5:25 am

You’d have to set some sort of minimum qualification for a lottery. Not all of the applicants to Harvard are actually Harvard material.

But I’d guess that a lottery of the say, top 1/3 of the applicant pool would do basically about as well as the current system.

Matthew K. Tabor September 1, 2009 at 7:19 pm

Statistics 101: Your sample sizes are too small.

Comments on this entry are closed.

Previous post:

Next post: