U Alabama Mines Data To Boost Enrollment, Retention

• By Linda L. Briggs
• 01/10/08

First-year enrollment at the University of Alabama is up 40 percent, at least in part because the school is using a sophisticated data analysis product to gauge which students to target in enrollment and retention campaigns.

The positive numbers--which also include an increase in student retention from 82 percent to 85 percent--are the result of a combination of efforts, according to Cali Davis, associate director of data analysis and specialized recruitment at UA. Strategic decisions by university top brass have played a part in the solid growth figures, Davis said. But having good data and using it wisely has helped with strategic decisions the administration has made.

The university is using predictive analytics from business intelligence company SAS. Although some of what UA is doing is no longer highly unusual in higher education--though Alabama was an early and creative user of data mining--the extent in which the university is using data about current and prospective students is noteworthy.

Davis previously used SAS for data modeling as a graduate student in the registrar's office, working in partnership with business school professor Mike Hardin in 2002 on a student retention project for his course in data mining. In that project, the two, working with Hardin's data mining students, used real-world data from the enrollment office to identify at-risk freshmen. Examining data points such as college entrance test scores, grade-point averages, and demographic features such as students who were single parents, the university began early intervention efforts that have helped prevent students from leaving the university.

The more advanced analytics began in 2005, Davis said, when "we saw the need for some type of model that would predict what makes our students likely to enroll." The university has an aggressive growth strategy in place--to increase enrollment from 20,000 to 28,000 over 10 years. And, to drive it, UA is employing extensive recruitment campaigns. As part of that effort, the university was seeing a significant increase in applicants, as well as students simply showing an interest in UA. "We needed to get a better feel for what our population of freshman students [looked like]," Davis explained.

For example, Alabama uses SAS to analyze data, collected in its Banner student information system from various sources, in order to predict which high school students from throughout the state are most likely to attend the university. By sharing those results with recruiters outside Alabama, additional prospective students can be targeted. Captured data includes college choice preference, financial aid and scholarship awards, college test scores, and residency status. Davis uses that data to segment students in a variety of ways, including by state or region, for targeted enrollment campaigns.

"If someone said, 'What does the typical University of Alabama student look like,' " Davis said, "we could throw out the average SAT score [and other conventional statistics], but we had no concept in terms of financial aid or scholarship information." They also didn't know whether students were sending their test scores to Alabama as a first choice, or simply as an afterthought--all valuable information in determining which students were the most likely prospects to attend the university, and thus should be targeted most heavily.

UA draws on a variety of data sources for its student enrollment program, Davis explained, including not just SAT and ACT scores, but also information each student supplies in submitting test scores. They also have data gleaned from student applications themselves, Davis said. And, more recently, "We've done a better job collecting data on the types of communication we've had with a student--e-mail, chat, letters, visits they've participated in." They haven't yet mined those bits of data, she said, but now have two years worth collected and are planning ways to use it.

Read More:

• University of Alabama
• SAS