[Michael Winerip pours some cold water on the impact of teachers.]
You knew there was a catch. Really sort of an educational Heisenberg Principle: if we measure teachers unobserved we can see the impact, but if we look? Then we get teach to the test.
Added comment: Testing does not an education make, that’s the work of teachers. Incentives do change behavior, but we can end up getting only the lip-service side of things, the superficial meeting of goals but not the long term accomplishments we seek. So the question, the “Heisenberg” aspect: can one conduct an evaluation that is both neutral and yet does not increase distortions? Or are distortions the necessary part of testing regime?