Excess of Democracy

View Original

Visualizing the 2018 U.S. News law school rankings--the way they should be presented

The U.S. News & World Report ("USNWR") rankings have been released. Like most, I've long been a critic of much that USNWR does, from how it distorts law school admissions practices, to its deeply delayed reporting of relevant data. I've also critiqued how USNWR chooses to display information in its rankings, often displaying information it doesn't use in its ranking or failing to display (or even share) relevant data.

The ordinal ranking at the heart of the USNWR rankings is perhaps its greatest deceptions. It crunches its formula and spits out a score. That score is normalized to give the top-scoring school (Yale) a score of 100, and it scales the rest of the scores off that.

But the magazine then chooses to display rank order of each school--even if there are significant gaps between the scores. To highlight one such example this year, Berkeley has a score of 82, Cornell has a score of 81, and Texas has a score of 75--suggesting that Berkeley and Cornell are quite close, and Texas is somewhat farther behind those two (even if in overall elite company!). But the magazine displays this as Berkeley 12, Cornell 13, Texas 14--distorting the narrow gap between Berkeley and Cornell, and the much wider gap between Cornell and Texas. And even though the magazine displays the overall score, the ordinal ranking drowns out these scores. Indeed, as the rankings are ordinal, there is no space from one school to the next, suggesting that they are placed along an equal line.

This plays out elsewhere in the rankings, as law students agonize over small differences in ordinal ranking that belie fairly distinct clumpings of schools that suggest little difference--indeed, in many cases, differences likely only the result of rounding the raw score up or down to the next whole number.

Assuming one takes the USNWR formula seriously--which it doesn't even appear USNWR does, given its choice to rank--a better way would be to visualize the relative performance of each school based on the score, not assigning each school an ordinal rank. That provides better context about the relative position of schools to one another. And that can help illustrate sharp differences in the overall score, or groupings that illustrate a high degree of similarity between a number of schools.

Below is my attempt to visualize the rankings in that fashion. (Please note that this may look best on a desktop browser due to the size of the chart.)


See this content in the original post