High-Tech Admissions

Predictive Modeling Software is Reshaping the Way Schools
Find and Communicate With Prospective Students

When Tuskegee University officials set a goal of growing enrollment from the current 3,000 students to 5,000 by the year 2006, they knew it would require an extraordinary approach. Especially given the fierce national competition for college-bound African American high school seniors.
They're hoping the latest wave in college recruiting software will give them just the edge they need.
Known as predictive modeling tools, the programs allow small schools like historically Black Tuskegee to receive a higher return on recruiting dollars by making more efficient use of prospective student lists. Using statistical analysis techniques borrowed from the direct marketing industry, the software determines the mix of prospective students most likely to enroll if admitted.
That can be particularly useful for schools looking to up racial and ethnic diversity or boost special population segments in the student body.
"We're looking to get a better profile on who are our best college-bound [high school] seniors," says Tuskegee's provost, Dr. William Lester.
Statistical modeling software, which has been on the market for about five years, is just one example of how information technology tools are transforming the college admissions process. The Internet is allowing schools to give prospective students access to college information chat rooms, interactive campus tours via Web sites and online college applications.
 "College admissions is being affected by innovations in the use of [information] technology," says Joyce E. Smith, executive director of the National Association for College Admission Counseling.
With college application sites popping up all over the Internet, communication between students and institutions is improving, and new software is easing the burden of processing inquiries and evaluating thousands of applications.
But Smith says the technology has raised new concerns as well. The Internet, for instance, has become a magnet for con artists looking to take advantage of students seeking information about schools and scholarships.
Congress last year, at the urging of U.S. Sen. Spencer Abraham, R-Mich., considered legislation to make so-called "scholarship scams" a federal crime with tougher punishment because of the rapid spread of such fraud.
Such fraud currently isn't covered under federal criminal law. The Federal Trade Commission has civil authority and can sue to recover consumers' money but can't impose prison penalties.
Despite these concerns, there is no question that positive applications of technology are on the rise.
"We're in the early stages of colleges utilizing the Internet in admissions," says Thomas Williams, president of the Noel-Levitz division of the USA Group, a student loan financing company. "In the next three to five years, you're going to see a complete shift to Web-based interaction between students and institutions."
Noel-Levitz is a consulting division of the USA Group and specializes in college admissions, enrollment management and retention services. It was among the first to offer predictive modeling software to its clients and is among a handful of firms that provide software, service and support to colleges and universities seeking to refine their recruiting campaigns. 
Searching for the
Most Likely to Enroll
Wabash College in Indiana, one of the few remaining men's college in the country, is turning to predictive modeling in its campus recruiting efforts because school officials believe that persuading young men to attend a male-only campus is one of the toughest sells in higher education.
"It's hard sometimes to convince a guy to come to an all-male college," says David Collins, the college's senior associate director of admissions.
In late September, his office submitted data on 9,000 prospective students to Bridge Technologies, a Maryland-based software company, to create a ranking of the students with an assigned value between zero and one or a percentage indicating the likelihood of a student's enrolling.
The data, which includes 10 variables, or 10 individual pieces of information about each student, is analyzed in comparison to profiles of the most recent freshman classes. The variables largely are demographic, student performance data and personal interest information, such as intended major.
 Software vendors typically do the number crunching for their clients and send the results to the institution. Colleges and universities then can concentrate their marketing efforts on the students determined most likely to enroll based on the statistical analysis.
At Wabash, the admissions staff will devote their resources marketing the school to the top 3,000 prospects. Collins says the analysis evaluates far more data relationships than admissions officers could while making guesses about two or three variables. Wabash enrolls around 270 freshmen a year.
"What we are trying to do is put our scarce resources towards people who are more likely to come here," Collins says.
For schools seeking to admit higher numbers of people in special populations, such as Black and Hispanic students, predictive modeling software can analyze and compare a special prospect list against the school's most recently enrolled freshmen.
Colleges frequently purchase lists that identify academically strong minorities from organizations such as The College Board that have identified minority students through standardized test results and other means.
Wabash currently wants to boost its enrollment of minorities and students from outside Indiana. Students of color represent roughly 15 percent of the school's population. Having a ranking of special prospect students allows a school to devote resources to recruiting them.
"We feel this software can help [us] do that," Collins says.
Noel-Levitz' Williams says that in judging predictive modeling results, schools only have to compare the enrollment rates of students coming from a prospect list generated by the new software against the enrollment rates from prospect lists generated by other means.
In 1998, Baylor University in Waco, Texas, another Noel-Levitz client, reported a 22 point jump in the percentage of students identified as likely enrollees that were heavily targeted by the school because of predictive modeling.