Effects of Teacher Evaluation and Incentives on Student Achievement: Evidence from DCPS
Thursday, November 12, 2015
Flamingo (Hyatt Regency Miami)
*Names in bold indicate Presenter
In recent years, many states and districts have introduced policies aimed at improving teacher job performance through increased use of accountability and incentive measures. These policy changes have been strongly encouraged by federal education policy during the Obama administration through the Race to the Top competitive grant program and the terms of No Child Left Behind waivers. While we have extensive evidence that high-quality teachers meaningfully impact students’ educational and economic outcomes (e.g., Chetty, Friedman and Rockoff, 2014), we know much less about teachers’ ability to respond to rigorous evaluation and incentive systems. The evidence to date has been mixed: evaluations of short-lived financial incentive programs that make awards based on improvements to student achievement have generally found null effects (Fryer, 2013; Glazerman & Seifullah, 2012; Springer et al., 2010), while the results of studies on observation-based evaluation programs have been more encouraging (Sartain & Steinberg, in press; Taylor & Tyler, 2012). The District of Columbia Public Schools (DCPS) was the first major urban district to introduce a dramatically redesigned teacher evaluation and compensation system, IMPACT, in the fall of 2009. IMPACT attaches uniquely high powered incentives, including financial bonuses and permanent salary increases for high-performing teachers and the threat of performance-based dismissal for low performing teachers, to a rigorous evaluation comprised of multiple measures of job performance. Dee and Wyckoff’s previous work in DCPS finds that the incentives embedded in IMPACT improved the job performance of low-performing teachers just rated Minimally Effective by 27 percent of a teacher-level standard deviation in the next year. They also find that the financial incentives associated with continued highly effective performance further improves performance by 24 percent of a teacher-level standard deviation.
We extend this work to examine the effects of IMPACT’s strongest incentive for low-performing teachers on student achievement, and examine how these effects differ by school context and individual teacher traits such as experience. We leverage IMPACT’s unique program design, which creates sharp incentive contrasts based on teachers’ continuous overall score in the prior year, to estimate these effects using a regression discontinuity (RD) design. In preliminary analyses, we find that teachers at the Minimally Effective performance threshold are able to improve student achievement in the next year in Math after the second and third years of program implementation. We see some evidence of improved student achievement in reading following the second year of the program, but this result is less robust to alternative specifications than the effect in math. These preliminary findings provide new evidence that highly incentivized teachers at the low end of the performance distribution are able to respond to incentives and change not only their overall job performance, but also their impact on student learning outcomes in the next year.