Friday, April 6, 2007

Is There a Bias in the USNWR Lawyer/Judge Assessment Against Schools in "Fly-Over" Land?

Posted by Jeff Lipshaw

Honestly, I understand now why Bill Henderson and the people over at Empirical Legal Studies have so much fun.

Okay, over at MoneyLaw, somebody commented that "there appears to be a 'coastal bias' going on. Schools on the coasts probably get more employers interviewing students and more exposure than most 475419_57666777 schools in the 'fly over zone.' How many lawyers/profs/judges in the 'fly over zone' get to answer this survey?"

Casting aside all the philosophical BS in which I am usually engaged, I got to work assessing this claim.  I did so by taking map of the United States and assigning what I refer to as "Coastal Bacon Numbers" to each state.  This refers to the degree of separation from either the Pacific or Atlantic Coast (I didn't count either the Great Lakes or the Gulf Coast) by the number of states you have to go through to get to a coast.  (The allusion is to six degrees of Kevin Bacon numbers, something Alan and Nancy are far more Coastalbacon concerned about than I.)  So if you are in Massachusetts or California, you have a Coastal Bacon number of 1.  The highest Coastal Bacon number is 5, and Minnesota is the only state that gets it.*

What you see above is a scatter diagram (click on it to see the detail) showing lawyer/judge assessment scores on the X-axis, and the average Coastal Bacon number of the schools at that rating on the Y-axis.  Again, I would need to channel Bill Henderson to do a regression analysis on this, but it appears to me that there is no correlation between coastal bias and lawyer/judge assessment.

My agent is available for invitations to participate as co-blogger over at ELS.

UPDATE:  For a competent analysis of the data, see Bill Henderson's post at ELS.

-----------------------

* I arbitrarily gave the District of Columbia a Coastal Bacon number of 1, even though it doesn't border on the Atlantic, because it embodies the East Coast Ivy League Prep School Liberal Establishment.  I also gave Hawaii a 2 because you aren't really on the West Coast until you get to California.  On that measure I should have given a 2 to Oregon and a 3 to Washington, but I didn't because it's really cool to be in Seattle.  I also excluded the top 16 schools in lawyer/judge assessment because it seemed to me they were unquestionably "national" and didn't fit my original bell curves anyway.

http://lawprofessors.typepad.com/legal_profession/2007/04/is_there_a_bias.html

Law & Society | Permalink

TrackBack URL for this entry:

http://www.typepad.com/services/trackback/6a00d8341bfae553ef00d8346105c169e2

Listed below are links to weblogs that reference Is There a Bias in the USNWR Lawyer/Judge Assessment Against Schools in "Fly-Over" Land?:

» Is there a Coastal Bias in the Lawyer/Judge US News Survey? from Empirical Legal Studies
Over at the Legal Professions Blog, my good friend Jeff Lipshaw, equipped with an excel spreadsheet, a map, and a ruler, has taken on the perennial claim of that the Lawyer/Judge input variable in U.S. News contains a coastal bias. [Read More]

Tracked on Apr 6, 2007 2:37:20 PM

Comments

I don't think excluding the top 16 schools makes much sense. If, e.g., they are all in Bacon Score 1 states, doesn't that say something about the hypothesis? If, e.g., they are evenly spread across the Bacon Score continuum (1-5), doesn't that say something quite different about the hypothesis? Perhaps you could post another, supplemental, figure?

Posted by: Joe Miller | Apr 6, 2007 7:56:51 AM

Looks downward sloping to me. Even the outliers slope downward.

Posted by: Michael Risch | Apr 6, 2007 9:56:52 AM

Joe, the average Coastal Bacon for the schools below the top 16 in lawyer-judge score is 2.18. The average Coastal Bacon for the top 16 is 1.75. The mean Coastal Bacon for all law schools ranked is 2.14.

Michael, I would be delighted to send my Excel spreadsheet with the underlying data to anyone who knows how to run a regression analysis with it.

Posted by: Jeff Lipshaw | Apr 6, 2007 11:02:41 AM

Alright, Excel is telling me that the best fit line for just the data in the scatter diagram is

y=.3x+1

I think that means slightly upward sloping.

What an amateur. Bill, where are you?

Posted by: Jeff Lipshaw | Apr 6, 2007 11:31:42 AM

I'm not sure that the Coastal Bacon measurement works as a measurement of coastal bias. Arkansas and Missouri, for example, would have scores of 3, but only because North Carolina and Tennessee are such long states. Iowa, on the other hand (which sits just above Missouri) has a score of 4, and Minnesota (just above that) the only 5. Do those with a coastal bias really distinguish among those 4 in-a-line states or are they all fly-over land? I think you should do it in miles rather than number of states. But this is sure fun!

Posted by: Suzanna Sherry | Apr 6, 2007 11:36:53 AM

Suzanna, your point is well-taken. But I think a better Coastal Bacon number would be some function of air travel time from JFK or SFO combined with driving time from the nearest airport to the campus.

Posted by: Jeff Lipshaw | Apr 6, 2007 11:43:52 AM

Michael told me offline my regression was backwards. I have reversed it, with the effect that the line is now

y=.3x+2

Posted by: Jeff Lipshaw | Apr 6, 2007 11:53:04 AM

Great post (and comments), and Bill Henderson's follow-up bears you out. He does note, however, you might "correct for some minor heteroscedasticity" but feels the bottom line stays the same. I strongly advise Jeff not to flaunt his heteroscedasticity with a minor.

Posted by: Childress | Apr 6, 2007 8:03:57 PM

Post a comment