Helping the Poor in Education: The Power of a Simple Nudge

By SUSAN DYNARSKI, January 17, 2015

There are enormous inequalities in education in the United States. A child born into a poor family has only a 9 percent chance of getting a college degree, but the odds are 54 percent for a child in a high-income family. These gaps open early, with poor children less prepared than their kindergarten classmates.

How can we close these gaps? Contentious, ambitious reforms of the education system crowd the headlines: the Common Core, the elimination of teacher tenure, charter schools. The debate is heated and sometimes impolite (a recent book about education is called “The Teacher Wars”).

Yet as these debates rage, researchers have been quietly finding small, effective ways to improve education. They have identified behavioral “nudges” that prod students and their families to take small steps that can make big differences in learning. These measures are cheap, so schools or nonprofits could use them immediately.

Let’s start with college. At every step of the way, low-income students are more likely to stumble on the path to higher education. Even the summer after high school is a perilous time, with 20 percent of those who plan to attend college not actually enrolling — a phenomenon known as “summer melt.” Bureaucratic barriers, like the labyrinthine process of applying for financial aid, explain some of the drop-off.

While they were graduate students at Harvard, two young professors designed and tested a program to help students stick to their college plans. Benjamin L. Castleman, now at the University of Virginia, and Lindsay C. Page, at the University of Pittsburgh, set up a system of automatic, personalized text messages that reminded high school students about their college deadlines. The texts included links to required forms and live counselors.

The result? Students who received the texts were more likely to enroll in college: 70 percent, compared with 63 percent of those who did not get them. Seven percentage points is a big increase in this field, similar to the gains produced by scholarships that cost thousands of dollars. Yet this program cost only $7 per student.

The same researchers also tested a texting program to keep students from dropping out of college. The problem is important because the graduation rate of low-income college students is dismally low; two-thirds leave without a degree. Community college students received texts reminding them to complete their re-enrollment forms, particularly aid applications. Among freshmen who received the texts, 68 percent went on to complete their sophomore year, compared with 54 percent of those who got no nudges. This, too, is a big impact — especially for a program that cost only $5 per student.

We know these programs worked because they were evaluated, like all the innovations cited in this column, using a randomized, controlled trial. Randomized trials, once rare in education research, are increasingly common. The What Works Clearinghouse, which reviews and rates the quality of education research, lists 242 randomized trials.

Students were randomly assigned to receive texts or not receive them. Because the two groups were randomly defined, they were basically indistinguishable at the start of the study. They diverged as the texts altered the behavior of those who got them compared with those who did not.

Text messaging won’t help everyone get through college, and cheap interventions won’t solve every problem. But they solve some problems for some students, freeing up time and financial resources for those who need other kinds of help.

Some students need personal counseling to help them balance the demands of school, family and work. Unfortunately, counselors are stretched thin, often carrying caseloads of thousands of students.

Two researchers at Stanford University, Eric P. Bettinger and Rachel Baker, analyzed an innovative counseling program in which a professional academic coach calls at-risk students to talk about time management and study skills. The coach might help a student plan how much time to spend on each class in the days approaching finals, for example. The results are impressive, with coached students more likely to stay in college and graduate. This program is more expensive than texting — $500 per student, per semester — but the effects persist for years after the coaching has ended.

Can nudges help younger children? Susanna Loeb and Benjamin N. York, both also at Stanford, developed a literacy program for preschool children in San Francisco. They sent parents texts describing simple activities that develop literacy skills, such as pointing out words that rhyme or start with the same sound. The parents receiving the texts spent more time with their children on these activities and their children were more likely to know the alphabet and the sounds of letters. It cost just a few dollars per family.

Researchers at the University of Chicago and University of Toronto are also working on methods to develop literacy. Ariel Kalil, Susan E. Mayer and Philip Oreopoulos sent families texts with tips about how to read with their preschoolers. The result was that parents spent substantially more time reading with their children.

Researchers are also testing the effect of giving parents more information about their children’s efforts in school. A school in Los Angeles, in collaboration with Peter Bergman of Columbia University, sent personalized text messages to parents of middle and high school students. The texts told parents when their children did not hand in homework assignments, listing page numbers and specific problems for students to complete. The parents and students responded: Completed homework went up 25 percent and grades and test scores rose. Other forms of communication between the school and parents improved, too, with parents twice as likely to reach out to their children’s teachers.

These light nudges can’t solve every problem, by a long shot. But at a low cost, they can help many students.

Why aren’t schools, districts and states rushing to set up these measures? Maybe because the programs have no natural constituency. They are not labor- or capital-intensive, so they don’t create lots of jobs or lucrative contracts. They don’t create a big, expensive initiative that a politician can point to in a stump speech. They just do their job, effectively and cheaply.