TaxProf Blog

Editor: Paul L. Caron
Pepperdine University School of Law

A Member of the Law Professor Blogs Network

Monday, November 19, 2012

Henderson: How to Increase Your Law School's Academic Reputation in U.S. News

US News (2013)The Legal Whiteboard:  How to Increase Your Law School's Academic Reputation, by William D. Henderson (Indiana):

I have uncovered four factors that are associated with statistically significant increases and decreases of USN Academic Reputation.  To illustrate, consider the scatterplot below, which plots the 1993 ordinal rank of USN Academic Reputation against the 2012 ordinal rank [click on graph to enlarge].

Acadrep
BarFour sets of dot (Red, Blue, Orange, and Green), each representing distinctive shared features of law schools, are distinctly above or below the regression line.  These patterns suggests that changes in USN Academic Reputation over time are probably not the result of random chance. But we will get to significance of the Red, Blue, Orange, and Green dots shortly.  

The primary takeaway from the above scatterplot is that 2012 USN Academic Reputation is overwhelmingly a function of 1993 USN Academic Reputation.   Over 88% of the variation is explained by a school's starting point 20 years earlier.  ...

That said, the scatterplot does not show a perfect correlation; slightly less than 12% of the variation is still in play to be explained by influences other than starting position.  A small handful of schools have made progress over these 20 years (these are the school above the regression line) and a handful have fallen backwards (those below the line).

BarThe Red circles, Blue rectangles, Orange diamonds, and Green circles represent four law school level attributes.  The Reds have been big gainers in reputation, and so have the Blues. In contrast, the Oranges have all experienced big declines; and as as a group, so have the Greens.  When the attributes of the Red, Blue, Orange, and Green Schools are factored into the regression, all four are statistically signficant. (Red, p =.000; Blue, p = .001; Orange, p = .012; Green, p = .000) and the explained variation increases 4% to 92.3%.  As far as linear model goes, this is quite an impressive result. ...

RedChanged the Law School Name (average gain: +39.0) ...

BlueConservatives Welcome (average gain: +28.3) ...

    Name changes and conversativism are the factors associated with an increases in USN Academic Reputation.  What are negative factors?

    OrangeScandals and negative NY Times coverage (average drop: -24.3) ...

     

    GreenSchools in the Rust Belt (average drop: -13.0) ...

    After [considering] ... starting position, name changes, conservativism, scandals, and Rust Belt status, ... [t]he statistics suggests that there is really no variation left to explain. ... Law faculty are comprised of very smart people, yet we organize virtually all of our hiring, strategic plans, and marketing efforts in an effort to make gains in a reputational game that cannot be won. 

    http://taxprof.typepad.com/taxprof_blog/2012/11/henderson-how-to-.html

    Legal Education | Permalink

    TrackBack URL for this entry:

    http://www.typepad.com/services/trackback/6a00d8341c4eab53ef017d3def6c81970c

    Listed below are links to weblogs that reference Henderson: How to Increase Your Law School's Academic Reputation in U.S. News:

    Comments

    Other law schools with an avowed conservative emphasis are Liberty, Ave Maria, and maybe Chapman. How would their inclusion affect the hypothesis?

    Posted by: Anonymous | Nov 19, 2012 11:29:37 AM

    Nice use of color! And this is a great example of how to combine formal analysis with subjective info about outliers. Indeed, I wouldn't be surprised if Prof. Henderson first did the regression and then looked at the outliers and thought about what they had in common--- which would be an example of letting the data tell you something.

    The advantage of being conservative is also an example of something else: the advantage of being Different. Almost all law schools simply compete head-to-head, trying to be good in exactly the same dimensions. Naturally, none of them then changes ranking except by some oddity such as losing funding or having a publicity disaster.

    Posted by: Eric Rasmusen | Nov 19, 2012 3:07:17 PM

    The labels of the x- and y-axes are reversed. x-axis should be 1993 and y-axis should be 2012

    Posted by: Anonymous | Nov 19, 2012 8:05:49 PM

    To Anonymous, I appreciate your pointers. That said, none of the schools you mentioned were ranked in 1993. Re the Y/X axis, you are right technically. But usually your "high" values are to the right, but this one is to the left, with 1 being better than 150. So, to make it more intuitive for the non-specialist, I flipped the X/Y axis so "outpeform" was above the line and "underperform" was below the line.

    In my experience, it is better to work with the schemas embedded in the brain than to work against them.

    Bill H.

    Posted by: William Henderson | Nov 20, 2012 5:01:42 AM

    I was the second Anonymous, posting about the axis labels. Apologies for neglecting to distinguish myself from the first Anonymous; for this post I've renamed myself accordingly. You're quite right about the counterintuitiveness of lower numbers being better rankings (a detail that I overlooked completely).

    Here's another approach that you may wish to consider if you present the data again: flip the axes back so that 1993 is horizontal, and then rotate the graph 180 degrees so that it "begins" at (200,200) instead of (0,0). This way, better rankings are rightwards and upwards, and improvers still show above the diagonal. (At least for the schemas embedded in my brain, this works best of all.)

    I too appreciated your use of colors. And having both red and blue as improvers makes an amusingly ambiguous political statement these days even if, as I suppose, you had none in mind.

    Posted by: Anonymous2 | Nov 20, 2012 2:28:55 PM

    I was the second Anonymous, posting about the axis labels. Apologies for neglecting to distinguish myself from the first Anonymous; for this post I've renamed myself accordingly. You're quite right about the counterintuitiveness of lower numbers being better rankings (a detail that I overlooked completely).

    Here's another approach that you may wish to consider if you present the data again: flip the axes back so that 1993 is horizontal, and then rotate the graph 180 degrees so that it "begins" at (200,200) instead of (0,0). This way, better rankings are rightwards and upwards, and improvers still show above the diagonal. (At least for the schemas embedded in my brain, this works best of all.)

    I too appreciated your use of colors. And having both red and blue as improvers makes an amusingly ambiguous political statement these days even if, as I suppose, you had none in mind.

    Posted by: Anonymous2 | Nov 20, 2012 2:31:50 PM