Many programs designed to increase college access try to reach students early. And with good reason: Decisions made starting in middle school can play a large role in determining students’ college options.
So can a program that doesn’t reach students until their senior year still make a difference in college enrollment?
It can, according to a new working paper published by the National Bureau of Economic Research. But the program the paper examines—which includes both mentoring and cash incentives—made a difference only for women.
The paper, “Late Interventions Matter Too: The Case of College Coaching New Hampshire,” describes the effects of a program designed to reach high-school seniors who are not sure of their college plans, who are intimidated by all they must do to apply, or whose default setting is not to go to college because none of the people closest to them have done so.
The researchers—Scott Carrell, an associate professor or economics at the University of California at Davis, and Bruce Sacerdote, a professor of economics at Dartmouth College—worked with a dozen high schools in New Hampshire. The schools identified students, about 1,150 over several years, who were on the fence about applying to college, and the researchers randomly assigned them to treatment and control groups.
Students in the treatment group were given a mentor—a current Dartmouth student—who would meet them each week until their application goals had been reached. The students were told they would qualify for $100 in cash if they completed college applications, requested transcripts and letters of recommendation, and began their applications for financial aid.
On the whole, students assigned to the treatment group were 5.4 percentage points more likely to enroll in college than were students in the control group. But when the students were broken down by gender, no effect was found for men. Women, in contrast, experienced a 15.1-percentage-point increase in college enrollment (from a base of about 50 percent).
As expected, the effect was quite a bit stronger for those women randomly assigned to the treatment group who actually participated in the program. In their case, college enrollment increased by 30 percentage points.
The researchers don’t know for sure why men and women responded differently to the program. Mr. Sacerdote said in an interview that he suspects it’s a combination of two factors. For one, men with no education beyond high school earn more in the labor market than do women who haven’t gone to college. Perhaps men were attracted to opportunities to earn a living right away. And second, it’s possible that the women came into the program with a stronger drive to follow through on their goals.
Because the program raised enrollment among students on the margin of college attendance, there was a risk that participants would be more likely to begin college and then drop out, Mr. Sacerdote said. It was important to check if the program was “setting kids up for failure,” he said. But there was no evidence it did: Students in the program had similar persistence rates to those in the control group.
The researchers also wondered whether they would find the same effects if they provided the cash incentive alone. In the last of several years of data collection, they offered the $100 reward to students in the control group. The researchers found that the money alone made no statistically significant difference in enrollment. Interviews with participants, too, indicated that money was not the most important factor in the program to those who went through it.
Financial incentives can change behavior when people already have a strategy in mind, Mr. Sacerdote said. That’s why home-buyer credits work. The fact that cash alone didn’t make students enroll suggests that they simply didn’t know how to approach the application process, rather than that they knew what to do but needed an incentive.
The researchers are continuing to test iterations of the program, Mr. Sacerdote said. “We clearly have an interest in programs that can be done on a larger scale,” he said. To that end they are tweaking the program to use administrative data instead of counselors to select the students and trying out less personal but less costly versions of mentoring.