It has not come as a surprise to Dr. Juan Gilbert, an assistant professor in computer science at Auburn University, that some highly selective colleges and universities would experience significant cost increases to implement admissions systems to comply with last summer's U.S. Supreme Court decisions in the University of Michigan affirmative action cases. With the court preserving affirmative action but banning means, such as the point system that helped minority applicants in the Michigan undergraduate case, some schools have turned for the first time to an admissions process involving holistic and individual review of all applications.
Recently, the University of Michigan and Ohio State University reported that with revamped admissions processes they roughly spent an additional $1.8 million and $250,000 respectively, resulting in underrepresented minorities making up slightly lower shares of the fall 2004 freshman classes as achieved in previous years. The increase in admissions expenditures at the University of Michigan was 40 percent higher in 2004 than in 2003, according to the New York Times. The schools attributed the cost increases largely to the hiring of dozens of admissions counselors, added to ensure all applications underwent thorough review rather than merely scored by a point system.
"When I saw (last year's) ruling, I got the impression that institutions might say individual review is too hard," says Gilbert, who grew concerned that although most selective colleges and universities have pledged diversity in their student bodies some might eventually back away from their commitment over the expense or if they were legally challenged on it.
"It could cost them millions a year to do this. With that in mind, I set out to devise a process" that could help automate the process of individual review, explains Gilbert, who was named a Black Issues In Higher Education Emerging Scholar in 2002 (see Black Issues, Jan. 3, 2002).
Gilbert's concern led him to develop software using "cluster analysis" that could help admissions offices better organize the individual review of applications. If applied to undergraduate or graduate admissions, the concept of cluster analysis could group large numbers of applications into a predetermined number of clusters using a number of criteria. Membership in each individual applicant cluster would be based on grouping applications by their similarity to one another.
"This software uses well-known clustering algorithms from computer science and information retrieval to automatically compare thousands of applications to each other and place them into clusters or groups, based upon a holistic view of their similarity. The clusters represent diverse application pools with respect to a holistic view of each application," according to Gilbert. …