Tuesday, April 11, 2006
I’ve been a huge fan of J. Gordon Hylton ever since, as a youngster, I read his great work on African American lawyers in Reconstruction Virginia. I highly recommend it to you all. Gordon and my colleague David Callies and some other folks have an innovative property casebook, which I recommend to you for insights into teaching and property theory, even if you’re teaching out of another book. And if you are using another book, I recommend it to you for consideration for adoption.
When I saw, in Jason Czarnezki's post over at Empirical Legal Studies, that Gordon has a new ranking system–-“US News Without the Clutter”–-I was, to say the least, interested. (Paul Caron's picked up this story.) Gordon makes the point, which I agree with completely–-that law school rankings ought to focus on quality of students and quality of faculty. The key question for me is what’s the intellectual experience at a law school? Is it a place on fire with ideas? If so, that’s a place that deserves a good ranking, IMHO. I think that’s what ought to matter to prospective students, as well.
Gordon, then, looks at two things: student quality (as measured LSAT midpoint between 25th and 75th percentiles) and faculty quality (as measured by US News’ peer assessment scores). Good idea–I tend to think that much of the rest of that stuff is (1) manipulable (particularly self-reports on graduates’ employment) and (2) irrelevant to the intellectual experience of students and faculty at the school.
I’m enamored of what should henceforth be known in the trade as the “Hylton Rankings.”
I thought his method–adding the mid-point LSAT (after subtracting 130 from it) of each school to its peer assessment score (multiplied by 10)--was worth a little more investigation. The standard deviation, as you will notice below, for the LSAT-midpoint is much larger than that for peer assessment; therefore, the combined score gives more weight to the former more than to the latter. The means and standard deviations for peer assessment and LSAT midpoint are as follows:
LSAT mid-point Peer assessment
maximum 173.0 maximum 4.9
median 157.50 median 2.30
minimum 146.5 minimum 1.3
Mean 158.11 Mean 2.51
SD 5.24 SD 0.85
So, I tweaked the Hylton rankings slightly. (I also added in U.Conn, which Gordon had left out.) I calculated standard scores with a mean of 50 and a standard deviation of 10 for each of the two variables (peer assessment and LSAT midpoint), added the scores for each school, and divided by 2. Thus, the composite score gave equal weight to the new variables. Here is a list of schools whose “Hylton ranks” and “modified-Hylton ranks” differ by more than |3|. (There are 46 such schools; the new ranks are higher in 25 and lower in 21.)
Schools Whose Hylton Ranks and Modified-Hylton Ranks Differ by More than |3|
|Difference||School||New score||New Rank||Hylton Rank|
The tweaked Hylton rankings and the original Hylton rankings
are almost identical. The correlation between Hylton’s ranks and the
new ranks is .998. The almost perfect correlation between the two sets
of ranks might suggest that there's no need to tweak the scores.
One more thing: the correlation between peer assessment and LSAT midpoint is .91. All of which suggests that the people asked to compile the US News peer assessment scores are making judgments that are closely related to law student quality (as measured by LSAT midpoint). I will have some more thoughts on this and a related issue of law review citations soon.
UPDATE: See also the related post, Hylton Rankings II and there's more talk of rankings of law schools based on their journals here and yet more talk of rankings of law reviews here.
Comments are held for approval, so they will not appear immediately.