Predictive Models Give Recruiting Edge | Colleges seek to optimize marketing
How does a student with a handful of college acceptance letters decide which school to attend? And what is the right amount of financial aid to offer?
Higher education officials in Central Massachusetts and beyond have been asking themselves those questions for decades.
But the method they use to determine the answers is a combination of data mining, social science, game theory and statistics known as predictive modeling.
Predicting The Unpredictable
Kristin Tichenor, vice president for enrollment and institutional strategy at Worcester Polytechnic Institute, summed up why colleges use mathematical models to hone recruiting and financial aid budgets. First, there is a lot of money in play. And second, “We’re dealing with 17-year-olds, who are capricious creatures at best,” she said.
Predictive modeling is a method prevalent across a wide swath of industries, from insurance to manufacturing to genetic research.
At Worcester Polytechnic Institute, the College of the Holy Cross and elsewhere, predictive modeling helps admissions staffs focus their recruiting efforts on desirable groups of students in the most cost-effective way.
The basic idea behind predictive analytics is using historical data to predict future outcomes, said Bogdan Vernescu, head of WPI’s mathematics department.
Vernescu said that a simple example of such modeling is how Amazon.com advertises certain products to people who have purchased a particular item in the past.
“If you are a frequent buyer of power tools, they will not advertise shoes,” Vernescu said.
Linda Cox Maguire, vice chairman of the Maguire Associates in Concord, helps colleges build their models for admissions and financial aid.
Maguire said that more and more colleges are realizing that predictive modeling is a necessary part of the business of running a school.
She said colleges want to optimize the dollars they spend on financial aid packages and on recruiting efforts.
“The competitive edge is having a much more precise understanding of a market’s reaction to an offer and not overspending to get that student to come,” Maguire said.
At the same time, a university’s goals and values come into play, and each model should be built around them, she said.
Safe Bets
At some schools, the work is done in-house, thanks to ever-advancing software that comes with training from the companies who make it.
Lynne Myers, director of the financial aid office at College of the Holy Cross, said that she and her staff use modeling software that allows them to predict how many of the financial aid offers the school makes each year will be accepted.
This is a vital function, Myers said, because financial aid offices typically offer more money than they have available, just like admissions sends acceptance letters to more students than they can actually enroll.
“When we’re predicting what we think our expenditures are going to be, it’s a safeguard to make sure it’s not a blank check out there,” Myers said. “This is something to hang our hat on.”
Holy Cross has been analyzing its data for years, she said. But where older software was designed to describe historical data, current software allows Holy Cross to better predict the future by modeling student behaviors and decisions based on nearly any characteristic imaginable.
Those characteristics could be family income level, gender, geography, test scores, area of study or other factors.
Myers said that she feels that Holy Cross can do much more with the tool in the coming years, including incorporating more modeling into admissions recruiting instead of just financial aid.
WPI is a school that has been using predictive modeling in admissions for years, Tichenor said.
The school uses the models to target desirable students more likely to enroll, to figure out what type of student succeeds at the school and to achieve enrollment goals, such as recruiting more women or recruiting from a diverse geographic area.
Having formulas for dealing with human nature is a welcome alternative to guessing, she said.
Just like financial aid, where colleges are trying to optimize each dollar, so too are there cost considerations in admissions.
Most colleges pay the College Board, which administers the SAT exams, for names of students and data about them, including test scores, Tichenor said.
Although the cost is less than a dollar per name, the school also spends money sending out informational brochures to prospective applicants in addition to other marketing efforts aimed at getting them enrolled.
Multiply those costs by 100,000 students each year and it starts to add up.
Models also must take the competition into account, said Vernescu.
“If you don’t understand what makes the student attractive to you and to the other players, then you end up offering the wrong financial package, and someone else will get the student,” he said.
Correction: An earlier version of this story incorrectly stated the name of the Concord-based firm that specializes in college enrollment management and market research. The company is called Maguire Associates.
WBJ Web Partners
Most Recent
Most Recent
0 Comments