There is slow but steady progress towards using student outcomes in evaluating teachers. For example, see this recent editorial in the Milwaukee Journal Sentinel. On the whole this strikes me as a hopeful trend, but I also see it as in danger of going off the track.
The danger I see is too many of the people pushing this seem to operate out of a Theory X framework. The solution is to reward good teachers, in their view, and punish bad teachers. I certainly have no problem with identifying the small minority (I would estimate around 5%) who shouldn’t be teaching and moving them on to some other career. When I was on the Milwaukee school board teachers had the right to appeal to us before they were dismissed, and some were very, very bad.
But, aside from this small minority, improvement through a scheme of rewards and punishment is unlikely to lead to significant results. Hoping that all teachers will be above average is a pipe dream, if for no other reason than the need to do massive hiring in the next few years. Instead of winnowing out the pool, we need to give average people the tools they need to succeed.
One model is the typical application of statistical process control in industry. Rather than a boss telling him whether or not he is doing a good job, the worker is given a set of tools that allow him to measure his output and the ability to make changes when the output is not measuring up. The problem, in most cases is not lack of motivation, but a lack of information on how well students are learning combined with a lack of effective strategies for increasing student learning.