Charter Schools and Merit Pay

In a new post, educational historian Diane Ravitch says, among other things:

As I predicted on this blog, President Obama and Education Secretary Arne Duncan are now the spear carriers for the GOP's education policies of choice and accountability. An odd development, don’t you think? The Department of Education dangles nearly $5 billion before the states, but only if they agree to remove the caps on charter schools and any restrictions on using student test scores to evaluate teachers.

What is extraordinary about these regulations is that they have no credible basis in research. They just happen to be the programs and approaches favored by the people in power.

* * *

There is also no research that justifies the Obama administration’s belief that tying teacher evaluations to student scores will improve schools.

No research?

Take the charter school point first. In a study that Ravitch herself cites, the Center for Research on Education Outcomes at Stanford University found that "states that have limits on the number of charter schools permitted to operate, known as caps, realize significantly lower academic growth than states without caps, around .03 standard deviations."

To be sure, .03 standard deviations isn't huge. But it's something. And it's a "credible basis" for the Obama administration to give states a financial incentive to eliminate charter school caps. I am aware of no studies finding any benefit whatsoever from state laws restricting the number of charter schools that can open. Incidentally, Arkansas currently restricts the number of charter schools statewide to 24. There is no basis for this limit.

Second, take the merit pay issue. No research? Consider David N. Figlio and Lawrence W. Kenny, "Individual Teacher Incentives and Student Performance," Journal of Public Economics 91 no. 5-6 (2007): 901-14. Looking at national data from the National Education Longitudinal Study, they find that "test scores are higher in schools that offer individual financial incentives for good performance." To be sure, Figlio and Kenny concede that their cross-sectional study can't tell definitively whether it was better schools that adopted performance pay, rather than vice versa.

But here are a few studies that weren't cross-sectional:

1) Gary Ritter and Josh Barnett, "When Merit Pay is Worth Pursuing," Educational Leadership 66 no. 2 (2008). Ritter and Barnett studied a Little Rock merit pay program. After two years, "schools implementing the program achieved average gains of approximately seven percentile points for students in mathematics and reading. Scores of students in the pilot schools improved, whereas those of students in comparison schools decreased."

2) Adele Atkinson, Simon Burgess, Bronwyn Croxson, Paul Gregg, Carol Propper, Helen Slater and Deborah Wilson, "Evaluating the impact of performance-related pay for teachers in England", Labour Economics 16 no. 3 (June 2009): 251-261 (a working version is available here). Atkinson et al. use a sophisticated methodology to evaluate a merit pay scheme in Englnad, controlling for pupil effects, school effects, and teacher effects. They find that "the scheme did improve test scores and value added increased on average by about 40% of a grade per pupil."

3) Victor Lavy, "Performance Pay and Teachers' Effort, Productivity, and Grading Ethics," NBER Working Paper 10622. Lavy evaluates a merit pay program in Israel that gave cash bonuses to teachers whose students earned more "credits" on national graduation exams. He used two sophisticated methods: regression discontinuity design and propensity score matching. His results are substantively significant: As to one estimation, he notes that "the effect of treatment on credits earned in math is 0.256, a 18 percent improvement relative to the mean of the control schools (1.46). The effect of treatment on awarded credits in English is 0.361, a 17 percent improvement relative to the mean of the control schools (2.11)."

These aren't the only studies, of course, and incentive schemes sometimes don't show much benefit. Still, to claim that there is no evidence in their favor isn't accurate. Once again, the position that lacks evidence here is the position that Obama and Duncan are trying to combat, i.e., that it should be illegal to use test score data to assess a teacher's performance (as is the case in several states). These states might as well have passed a law stating that because so much of a patient's health depends on factors outside a doctor's control, it should therefore be illegal to consider whether a doctor's patients were killed by incompetence.

On the bright side, I applaud Diane Ravitch's announcement of the Partnership for 19th Century Skills.