M. Yglesias has an interesting post on the merits of elite universities. He's responding to G. Easterbrook's article on the non-merits of Harvard, which in turn bases its argument on A. Krueger and S. B. Dale's paper "Estimating the Payoff to Attending a More Selective College: An Application of Selection on Observables and Unobservables" (working paper PDF) (this work was announced in 1998, but didn't get published in peer-reviewed form until the Nov. 2002 issue of the Quarterly Journal of Economics.)
Whew. OK, bibliography part I out of the way. Now, let me point you to another, related paper, by J. E. Olson of St. John's University, called "The Cost Effectiveness of American Higher Education: The United States can afford its colleges and universities", available in Higher Education: Handbook of Theory and Research, vol. XII. (Not online, unfortunately, though you should be able to obtain an electronic abstract by emailing Agathon Press.)
I came across this paper while I was doing a literature search in an actual bricks-and-mortar, dead-trees library as a first-year grad student. Among other interesting things, the paper points out a number of explanatory models which account for the difference in economic outcomes for people who have greater education --- of which my old notes on the paper summarize three:
- "By providing credentials, colleges randomly eliminate some applicants for employers so that employers have to interview fewer prospective hires. This saves employers money. There is no real difference in ability between credentialed and uncredentialed applicants."
- "Colleges provide an 'obstacle course' for students to navigate. Students who successfully navigate the obstacle course will have, on average, greater intellectual abilities than students who do not, even though the specific skills tested bear little relationship to job skills. Employers are justified in hiring college graduates for this reason."
- "Like credentialing, except that once a person is credentialed, (s)he will feel obligated to perform up to the level of that credential, and will therefore make a better hire. For example, a Harvard graduate might have higher expectations for him/herself, and therefore will perform better." (This explanation has its roots in a theory from psychology called 'expectation states'.)
I believe these three models are fairly standard in the education literature. They were constructed to explain differences in outcomes for people who attend college and post-collegiate education versus those who do not; but they easily be applied to those who attend elite vs. non-elite universities.
In this context, Matthew's explanation suggests that he believes in the "screening" function of higher education. More precisely, he believes in the obstacle course that high school students must navigate in order to get admission to elite higher education. Easterbrook, on the other hand, seems to believe that the elite vs. non-elite distinction is simply melting away. I leave it to you to decide which you find more convincing.
Final note: Yglesias writes
Harvard's admissions department does a very good job of selecting students from a wide applicant pool, but the faculty does no better at educating the students than does the faculty at any number of less-selective institutions. There is, perhaps, nothing especially surprising about this when one considers that teaching ability is not a factor in hiring decisions at highly-selective universities which, ex ante, should make one suspicious of the notion that the professors there are particularly good at teaching.
I just want to second this statement --- at least, for some top universities. From firsthand observation, I can say that UW's computer science dept. currently does care about teaching ability in its hiring process (though it is, naturally, only one of several considerations). But there are other top computer science departments where it is transparently obvious, from the past history of faculty hires and tenure decisions, that remaining or becoming the nation's top computer science research department is the one and only consideration. Their students are stellar enough that the teaching quality hardly matters. And, of course, by sheer luck, sometimes these top departments still hire great teachers. But that's a side effect, not a first-order goal.
If you really want the best education possible, then you should find a department that paradoxically doesn't have a top research ranking. In fact, it's only a small exaggeration to say that you should find a department whose professors publish no research papers, because such a department might actually make hiring and tenure decisions based on demonstrated teaching ability instead of research output.