Oct 9, 2014

More bad data from U.S. News


Rollins College

The 2015 U.S. News Best Colleges guide has hardly been out for two weeks and already a couple of bad information to the data bank used to compile rankings and provide a steady source of income for the magazine.
colleges are being forced to confess to providing

In his blog, “Morse Code:  Inside the College Rankings,” Robert Morse  announced yesterday that  Lindenwood University in Missouri and Rollins College in Florida, both provided inaccurate information during the data collection phase of the annual U.S. News project.

It’s hardly unusual.  Over the past several years, Claremont McKenna College, Bucknell University, Emory University, George Washington University, York College of Pennsylvania, Dominican University of California, the University of Mary Hardin-Baylor, and Flagler College have also been in the same unenviable position of having to admit to providing inaccurate data to U.S. News and the other publications behind the Common Data Set.

This time, the mistakes seem innocent enough.  

In the case of Lindenwood University, it looks like an additional digit was mistakenly added to the number of alumni donors in 2012-13, increasing the correct figure of 2,411 to 12,411 as originally reported.

But because the error had an effect on the Best Colleges rankings methodology, the misreported data resulted in U.S. News moving Lindenwood to the “unranked” category for the rankings promoted on usnews.com.  It’s too late for the magazine, however, and the numerical rank in the Regional Universities (Midwest) category will stand for anyone purchasing a hard copy.

Rollins College advised U.S. News that it submitted incorrect counts for the number of students admitted for fall 2013.  This information is used to compute the school’s acceptance rate.

Instead of admitting 2,233 students, Rollins actually gave offers of admission to 2,783 students resulting in an acceptance rate of 58.8 percent as compared with the incorrect rate of 47.2 percent. Because the error had no impact on rankings, Rollins was not punished even with what appears to be a significant change in perceived selectivity—over 11 percentage points.

And whose fault is this?  At first blush, the blame appears to rest solely with the two colleges.

But really?  When U.S. News fails to provide even the most basic training and technical assistance to colleges completing survey forms for the Common Data Set and refuses to make even random checks on what information is being provided, it seems blame should be shared.  

Mistakes happen.  But in the case of such persistent indifference to accuracy, the publications compiling and selling the numbers bear some blame.

Until the big three organizations behind the Common Data Set take some responsibility for what they sell to the public and make even the most basic effort at quality assurance, this will continue—U.S. News will keep unranking institutions and college-based staff will face dismissal in the wake of unintentional scandals.

In fact, U.S. News seems to anticipate there’s more to come.  Morse adds at the end of his column, “U.S. News will continue to handle each case of data misreporting on an individual basis.”

No comments:

Post a Comment