How should we measure teacher performance?

Charlie Rose asked a panel of award winning teachers how to measure teacher performance, adding the question of whether it should be based on student tests. Let me free associate on the question.

I think one of the definitions of "a professional" is someone who does a difficult intellectual task with little supervision, and thus is qualified by professional training and licensing to do so. I think teachers are in that sense professionals, and that it is very hard to measure their performance.

There are objective measures of professional performance, but peer review tends to play and important role in some fields.

The basic law of measurement is not to measure anything if you are not going to use the information obtained. It costs money to measure something, and if you don't have any way to use the information, that money is wasted.

A corollary is that you should not measure anything that you can not measure sufficiently accurately to be useful.

Another corollary is that, since testing is often destructive, you should not measure anything that you are likely to make worse by the measurement and/or the use of the results.

It is in theory better to measure the impact of the teacher than to measure "output variables" such as facts or analytic techniques mastered by students or student and parent satisfaction with the teacher. It is in theory better to measure those outputs rather than "input variables" such as the teachers training and use of supplies and aids.

On the other hand, if you reward people for a certain kind of performance, then you are likely to encourage more of that kind of performance. If you measure the facts learned by students and use that as the only basis for evaluation of teachers, don't be surprised if you find teachers teaching more to increase the ability of students to respond to factual questions. If you want student creativity and analytic skills to be encouraged, then evaluate teachers accordingly.

Figuring out what you want schools to do is not easy. Indeed, different stakeholders have different(implicit) desires for and expectations of teacher performance. It is not clear that what will most please parents will most please future employers or the students themselves.

When different stakeholders have different objectives, it is sometimes better not to provide full information on how well all of them are being fulfilled. Sometimes you can better get consensus on means if you are not too specific on ends and their achievement.

There is the story of the system of county fairs in the United States that gave prizes to farmers every year. The corn prizes were for the most beautiful ears of corn. After a decade or two it was discovered that farmers were growing more and more beautiful ears of corn, but that yields were going down. Since cattle gain more weight if they eat more corn, not if they eat more beautiful corn, the system was profoundly dysfunctional. Measuring the wrong thing and using the information effectively to maximize what you measure may be profoundly dysfunctional.

Have you noticed that student evaluations of faculty performance sponsored by universities focus on indicators most related to the teacher's ability to attract fee's paying students to his/her classes?

Bottom line: measure teacher performance very carefully. Schools have been working for a long time with not very good measures of teacher performance. Unintended consequences of measurement processes are not uncommon. You would not want to make schooling worse by measuring performance more.