Monday, April 24, 2006
Seth J. Chandler (Houston) has reverse engineered the U.S. News law school rankings in Analyzing US News & World Report Rankings Using Symbolic Regression.
Seth has developed a four-factor model that predicts the U.S. News law school rankings with an R-Squared statistic of 97.7%. The four factors are:
- Peer reputation rating (Peer)
- Employed at nine months after graduation (Employed9)
- Student-faculty ratio (SFRatio)
- 75th percentile UGPA (UGPA75)
The formula is: 8.3124 + 0.04139Peer(1.1487Employed9-SFRatio)UGPA75
Example: Connecticut has a Peer rating of 2.9, an Employed9 rating of 96.4, a SFRatio of 11.8, and a UGPA75 of 3.63. The model predicts a score of 51, which is Connecticut's actual U.S. News score.
Seth also has developed a seven-factor model that predicts the U.S. News law school rankings with an R-Squared statistic of over 99%.
In our article, Ranking Law Schools: Using SSRN to Measure Scholarly Performance, 81 Ind. L.J. 83, 87 n.9 (2006), we discuss the claim by Tom Bell (Chapman) that he has reverse engineered the U.S. News rankings:
- The Puzzle of Penn Law School's Ranking (5/24/05)
- Gory Details, by Demand (5/24/05)
- Rank This Job (5/1/05)