Harvard Study: Cash Incentives for Students Can Spur Achievement

Published May 10, 2010

A recent study by Harvard economist Roland Fryer Jr. suggests cash incentives could spur improvement in students’ grades, test scores, literacy rates, and even their behavior.

Fryer, who directs Harvard’s Education Innovation Laboratory (EdLab), spent more than $6.3 million since 2007 to “bribe” kids to improve their performance.

Fryer conducted experiments in four urban school settings: Dallas, Chicago, Washington, DC, and New York City. In all, 18,000 students in more than 250 urban schools participated in his trials. Payments varied in each location. So did the outcomes.

‘When You Pay … You Get Results’

Sara D’Alessandro, one of Fryer’s research assistants on the project, explained the motivation behind the study, released in April as a working paper by the National Bureau of Economic Research.

“There is a persistent achievement gap in America, and conventional wisdom has failed us,” D’Alessandro said. “In the last few decades, spending is up, class sizes are smaller, teachers’ education levels have increased, and still the gap persists. We have made progress, to be sure, but we have not yet closed the achievement gap.”

“We certainly never claimed that incentives would solve all educational problems,” D’Alessandro added. “But we were interested in testing their impact.”

And what did Fryer and his team find?

“We learned that well-designed incentives programs—those that pay students for specific actions—have better results than paying students for end results,” D’Alessandro explained. “When you pay students for actions—like attendance, good behavior, homework completion, reading books, and so forth—you get results.”

D’Alessandro noted, however, paying students for improved test scores did not have the same result. “They don’t necessarily know how to produce good test scores,” she said.

Boys, Minorities Benefit Most

Fryer’s EdLab team also found a few surprises in the data.

“We were surprised to learn that while most reforms [and] most education programs tend to benefit girls more than boys, we found the exact opposite with the incentives,” D’Alessandro said. “They worked more for boys than they did for girls, and they actually worked a lot more for minorities than they did for non-minorities.”

D’Alessandro noted, too, the incentives improved student behavior. “When you paid them to have good behavior and to come to class and to sit still, their test scores actually went up more than any other subgroup,” she said. “So, what’s great about this is that incentives work for students who are harder to reach.”

Data Supports Choice

Neal McCluskey, associate director of the Center for Educational Freedom at the Cato Institute, noted the Harvard research provides more data on the effects of paying students for performance but warned against drawing definitive conclusions about incentives. However, he said the study confirms the value of school choice.

McCluskey pointed to some possible methodological flaws. “It isn’t a truly randomized study because schools had to agree to participate in the first place, which opens up problems of hard-to-control-for motivations and behaviors at the school level,” he said.

McCluskey says he does see some value in the study, however. “We can see that different kids respond in different ways to different rewards for different behaviors, which illustrates the great difficulties of reaching definitive conclusions through social science,” he said.

“Some kids respond well to cash rewards for performance,” McCluskey added. “Others don’t. Decisions have to be made on the basis of individual children, with parents having school choice and schools having the autonomy to decide for themselves how they will motivate their charges.”

‘Nuanced Reward Systems’

Lisa Snell, director of education and child welfare at the Reason Foundation in Los Angeles, said Fryer’s work at EdLab has “high value.”

“It shows that serious scientific experimentation with gold-standard randomized trials can drive instructional practice and offer nuanced guidance on how education policy can be designed,” she said. “From this study we know that more nuanced reward systems that reward behavior that contributes to learning that are used in many schools such as Kipp Academies have value, while more direct payment schemes that pay kids for test scores may be less effective.”

Further Study Discussed

The Harvard project will likely continue, D’Alessandro says. The data from the first batch of experiments has given researchers plenty to work with for years to come, she said.

“It would be great to continue this research and see if the incentives could work in cities other than the ones we worked with,” D’Alessandro said. “I’d be interested in exploring how we can combine a student, parent, and teacher incentive program that would all be centered on the same outcome, get everyone on the same page to improve student achievement, and get a positive return on investment.”

Could Shape Policy

Snell says policymakers might look at incorporating the Harvard findings into their ongoing school reform efforts.

“The Fryer study clearly shows that it is more effective to incentivize behavior that contributes to learning rather than paying kids money for direct improvements in test scores,” Snell said. “So the cities where kids were incentivized to attend class, exhibit positive behavior, and read books had better outcomes.”

Programs that “incentivize positive behavior from students in combination with other solid instructional practices” should be the goal, Snell said.

Snell cautioned, however, that student incentives alone aren’t enough.

“Paying kids to do well in school will not help if other important elements are missing, such as high teacher quality or engaging a standards-based curriculum,” she said.

Sarah McIntosh ([email protected]) teaches constitutional law and American politics at Wichita State University in Kansas.

Internet Info

Roland G. Fryer, Jr.: “Financial Incentives and Student Achievement: Evidence from Randomized Trials” http://www.nber.org/papers/w15898