Thursday, July 16, 2009

A Model of the 2010 USN&WR Law School Rankings

As in every year since 2005, I this year again built a model of the law school rankings published by the U.S. News & World Report ("USN&WR"). Figuring out the rankings—the "2010" rankings, as USN&WR's calls them—proved especially trying this time around. USN&WR changed several parts of its methodology this year and the ABA, which distributes statistical data on which my model depends, fell far behind its usual publication schedule. Finally, though, the model ended up generating scores gratifyingly close to those that USN&WR assigned law schools. Here's a snap-shot comparison of the results:

Chart of Accuracy of Model of USN&WR 2010 Law School Rankings

For details about how and why I modeled USN&WR's law school rankings, as well as for similar snap-shots, see these posts from 2005, 2006, 2007, and 2008.

Perhaps in later posts I'll offer some reflections on what this year's model of the USN&WR rankings teaches. For now, I'll just offer this happy observation: The close fit between USN&WR's scores and the model's scores suggests that law schools did not try game the rankings by telling USN&WR one thing and the ABA (the source of much of the data used in my model) another. Even a skeptic of law school rankings can find something to like in that.

[Crossposted at Agoraphilia and MoneyLaw.]


Anonymous said...

Does your model identify any problem with Brooklyn Law School, which reported its full time enrollment statistics and represented them as the enrollment statistics for its entire class -- as explained further here:

US News has failed to correct Brooklyn's rankings, but agrees that the ranking is incorrect due to this incorrect reporting.

Tom W. Bell said...

Anon: Because I am interested in copying USN&WR's results, I treated Brooklyn the same way that it did.

Occam's Razor said...

If you run the Brooklyn numbers with the full-time and part-time data combined, do you get a different scaled score that would change Brooklyn's rankings? If so, by how much?