The Behavioralist Goes to School: Leveraging Behavioral Economics to Improve Educational Performance
... when students were promised immediate rewards of $20 for improving effort on the current test relative to the last test, scores did improve...
In The Behavioralist Goes to School: Leveraging Behavioral Economics to Improve Educational Performance (NBER Working Paper No. 18165), co-authors Steven Levitt, John List, Susanne Neckermann, and Sally Sadoff note that "the crux of the education problem that we face with our urban youth" is that "effort is far removed from payout of rewards, making it difficult for students to connect them in a useful way" and that "the failure to recognize this connection potentially leads to dramatic under-investment" in education.
The authors report on experiments designed to determine whether financial and non-financial rewards affect student performance on standardized diagnostic tests in reading and math. They focus on tests taken about three times a year by students in elementary and high school. Their sample includes almost 7,000 students from over 30 schools in low-performing school districts in and around Chicago. Among other things, the experiments tested whether scores improved when students were promised immediate rewards of $20 for improving effort on the current test relative to the last test. Scores did improve, with mid-range improvement equal to 0.12 to 0.2 standard deviations, equivalent to 5 to 6 months' additional learning. Financial rewards of $10 and non-financial rewards of trophies yielded mixed results.
Both financial and non-financial rewards improved scores more robustly when they were framed as losses rather than as gains. If the reward was framed as a gain, students were told about it as they sat to take the test and were promised that they would receive it before they left if their scores improved -- or, in the test of delayed rewards, a month later. If framed as a loss, students instead received the actual reward before they took the test. They were told they could keep it if they improved their score, but that they would have to give it back if they did not. Delayed rewards did not improve student scores.
Younger students seem to respond more strongly to non-financial rewards. When framed as a loss, which took the form of trophies worth about $3, they increased the performance of younger students by 0.18-0.25 standard deviations. The authors conclude that non-financial incentives "may be a cost-effective alternative to monetary rewards", especially for those in grades 2 through 5. Older students only responded to financial incentives framed as a loss, with performance increasing 0.12 to 0.13 standard deviations.
Sub-group results suggest that the rewards had larger, and uniformly positive, effects on math tests; reading tests improved somewhat in response to the non-financial incentives. Boys responded positively to the $20 rewards, but there were no consistent effects among girls, leading the authors to conclude that girls may be more intrinsically motivated than boys. There was no evidence that providing rewards decreased effort on subsequent tests.
The authors conclude that "in the absence of immediate incentives, many students put forth low effort on the standardized tests that we study. These findings have important implications for policymakers because standardized assessment tests are often high-stakes for teachers and principals, but low-stakes for the individual students choosing to exert effort on the test."