Paying third-graders for better test scores

September 07, 2010

Efforts to improve education in the U.S. has included financial incentives for high-performing teachers and programs have targeted middle- and high-school students, but a recent study found success in giving money to kids as young as third grade who scored well on standardized tests.

Eric P. Bettinger

Associate Professor, Stanford University

In a paper published by the National Bureau of Economic Research titled “Paying to Learn: The Effect of Financial Incentives on Elementary School Test ScoresEric P. Bettinger of the Stanford School of Education looks at a program in the poor, Appalachian community of Coshocton, Ohio.

The pay-for-performance plan targeted third through sixth graders who took standardized tests in math, reading, writing, science, and social studies. The students could earn up to $100 — $20 per score of Advanced in each test. Students who scored proficient were awarded $15 per test. In order to make sure the proceeds went directly to the students, payment was made in “Coshocton Children’s Bucks,” which could only be redeemed by kids for children’s items. Participation in the program was randomized based on a lottery as specified by Robert Simpson, a local factory owner, who financed the effort.

The program showed generally positive results, with the biggest gains coming in math. Students who were eligible for the payments improved about 0.15 standard deviations, a statistically significant result. Though there were small improvements shown for other subject areas, the difference wasn’t statistically significant.

“Educational interventions often increase math scores with little to no impact on reading scores,” Bettinger writes. “Math particularly in the grades studied seems to be more elastic than other subjects… Math is less conceptual than reading in early grades. Students can memorize a series of facts in math that can adequately prepare them for most tests. By contrast, it is much more difficult for students to prepare for a specific reading text.”

The effects on the reading tests may not have been statistically significant, but Bettinger notes another benefit. Coshocton schools participate in a separate reading program that assigns points to books according to the difficulty, length, and importance of the book. Students in the incentive program had significantly higher point totals than the control group. So, while students’ test scores may not have increased as much, they were challenging themselves more in their reading.

That just leaves the bottom line. Was the program worth the expense? Over three years, Coshocton’s program cost about $52,000, roughly the same cost as employing one teacher in the system for one year and produced a clear benefit in math. “The Coshocton incentive program was a cost effective program which led to substantial math test score gains, especially for students at the bottom and top of the test score distribution,” Bettinger writes.