Like my colleague Andy Rotherham and friend Matt Yglesias, I’m not particularly surprised that a just-released Vanderbilt* study finds that offering teachers bonuses of up to $15,000 for improving student test scores “simply did not do much of anything” (to quote Linda Perlstein quoting the researchers). And like Andy, Matt, and Rick Hess**, I don’t believe this study is particularly informative to the broader question of how to structure teacher compensation, albeit for slightly different reasons.
Andy, Matt, and Rick argue that the point of performance pay systems is not to make existing teachers work harder--as some crude characterizations suggest--but to change the overall composition of the profession by making it more performance-oriented and creating incentives for high-performers to enter and stay in teaching. I agree generally, but think my fellow bloggers give short shrift to the potential of well-designed incentive systems to help improve the performance of existing teachers, too.
To be clear, I don’t think that current teachers are lazy or that all that’s needed to produce better results it for them to work harder. But good incentive system should not only measure performance and provide rewards based on it--they should also provide individuals with the means to improve their performance in response to incentives. Systems that evaluate and reward teachers based solely on test scores lack this fundamental element.
But when we talk about teacher evaluation and incentives in the real, non-RCT, world these days, we’re not talking about measuring teachers solely based on test scores. Rather, we’re talking about evaluation systems that combine test scores with standardized observational evaluations to create a rounded picture of teachers’ performance and provide them with feedback to improve. That’s what the Obama administration has advocated and what states and districts are currently putting in place. The problem is that in most cases, we don’t know if those observational measures are actually measuring teacher behaviors that matter, or providing teachers the kind of feedback that would help teachers to get better.
But there is emerging evidence that at least some standardized, reliable, and validated observational measures of teacher behavior in the classroom can both predict teachers’ impacts on student learning and provide them with feedback that they can use to become better. And the body of evidence in this field will grow over time. That’s why I think we need to talk a lot less about incentivizing teachers based on test scores and a lot more about the valid and reliable observations of teacher classroom practice and performance.
*Go ‘Dores!
**Read Rick’s piece on this. It’s excellent--not just on the matter at hand, but also as a pretty merciless evisceration of the stupidity of how we often talk and think about research and education, particularly randomized controlled trials.