A Low-Cost Way to Expand the Horizons of High-Achieving, Low-Income Students

March 29, 2013

For years, researchers have traced the phenomenon of "undermatching," in which high-achieving, low-income students rarely enroll in or even apply to the selective colleges their wealthier peers attend as a matter of course.

Now the researchers have found a way to broaden the horizons of the 35,000 students each year who finish high school with test scores in the top 10 percent and family incomes in the lowest quarter—for only about $6 apiece.

In a new paper, Caroline M. Hoxby and Sarah E. Turner explain the results of a randomized, controlled trial that tested whether sending such students more information changed their enrollment patterns. (Ms. Hoxby discussed some of their findings at the College Board Forum in October.)

Once the researchers uncovered how many students were high-achieving and low-income, they wanted to work on solutions to undermatching, said Ms. Hoxby, a professor of economics at Stanford University. And they wanted those solutions to be scalable to the tens of thousands of affected students. "Frankly, there are a lot of solutions that are good for 30 or 50 students," she said, "and they're already out there."

The two professors also wanted to test possible solutions before publicizing them, said Ms. Turner, a professor of economics at the University of Virginia. The experiment was designed to test whether some of the high-achieving, low-income students who undermatch would change their behavior if they had better information on colleges, and whether that information could be provided in a cost-effective way. In both cases the answer was yes.

The researchers designed and tested several different approaches, which they called "interventions." One focused on general college-search information, one considered net costs, and one consisted of application-fee waivers. They also tested a combined intervention.

Ms. Hoxby and Ms. Turner were able to partly customize the materials so the sets of colleges presented to test subjects included some that were close by. A total of 39,677 high-achieving, low-income students were assigned to one of the test groups or the control group. All of the interventions made a difference, and the combined intervention made the biggest difference.

In each case, materials were sent by postal mail, as previous research had shown that students and their parents were wary of online content.

'A Huge Decision'

The materials were sent by the Expanding College Opportunities project, a group the students had probably never heard of. The researchers used that approach because the low-profile project would avoid contaminating the control group. They believe more students would have paid attention to the materials if a well-known group like the College Board had sent them.

(The researchers may soon know if that's the case. They are working with the College Board to run this style of intervention next year, Ms. Hoxby said.)

The professors surveyed a sample of students in each intervention group over the phone to determine what share of them recalled receiving the materials. That step enabled the researchers to look at the results in two different ways: the effect the materials had on the whole group of students who received them, and the effect they had on the smaller group who remembered getting them.

The materials made a difference in both cases. For example, the whole group of students who received the full intervention applied to 19 percent more colleges than the control group did.

But the results among the recipients who remembered the materials are probably more reflective of how this kind of intervention would work if a well-known group took it on, the researchers said.

Students who received the combined intervention—and remembered it—submitted 48 percent more applications than did students in the control group. They applied to colleges with a 17-percent-higher graduation rate and an 86-point-higher median SAT score.

The results were similar in admissions and enrollment. Students who remembered the materials were admitted to 31 percent more colleges than students in the control group, and were 78 percent more likely to be admitted to a college where most students shared their level of preparation—what the researchers called a "peer institution."

Those students enrolled in colleges that were 46 percent more likely to be a peer institution, where graduation rates were 15 percent higher than for the control group.

"This is a huge decision for students, choosing which college you're going to go to," Ms. Hoxby said. The goal is not to sway high-achieving, low-income students to go to a particular kind of college, she said. It's to make sure they are as well informed as their more privileged peers. "To not make decisions well simply because you don't know what's out there," she said. "That's sad."