Tuesday, May 31, 2011

An Eloquent Argument Against Rating Teachers

Over a Bridging Differences there's a good article on the dangers inherent in rating teachers by student test scores. This paragraph was striking:
I recently met with the principal of an elementary school in Brooklyn who hates the city's teacher data reports. She told me about an excellent teacher of a gifted class whose students started the year at the very top, near perfect. At the end of the year, their scores had dropped 5/100 of a percentage point. Given the margin of error on all tests, this is a meaningless difference, but the "drop" in the scores caused the teacher to be ranked in the bottom sixth percentile of all teachers.  
What is being done to teachers and schools right now is so wrong!


  1. It seems more like an argument against doing a bad job rating teachers than against rating teachers.

    Though I guess maybe the idea is that if you try to, then you'll sometimes do a bad job, and it isn't worth that?

  2. Sorry, my shortened title may not be quite right here. The argument is against using student test scores to rate teachers.

  3. Nice article, thanks for the information.

  4. Even restricting attention to using student test scores to rate teachers, it seems like more of an argument against doing so badly. I think the point here is that the students were already so close to a "perfect" score on the test that you can't really expect improvement, and a tiny decline shouldn't be surprising due to random noise, even from a good teacher.

    I think any reasonable scheme for using test scores to rate teachers would acknowledge cases where the scores do not provide information. (I guess the scheme being used isn't a reasonable one.)

    So as I tried to allude to above: Maybe the danger is that it's too hard to come up with a scheme that's reasonably, or that no one will put in the necessary effort?

    Sorry if this is coming across as pedantic. It seems like a relevant distinction to me, but I know that sometimes distinctions that seem relevant to me don't to anyone else.

  5. Not pendantic. Did you read the whole post? I think I remember it talking about other problems.

    One that was raised either there or in something else I read recently was the possibility of a student doing badly on purpose to sabotage the teacher. Scary! Much of schooling is oppressive, and students often resent teachers. It's such a hard job to do well.

    But to me, the bigger problem is that the most important parts aren't easily measurable. Dang, who just wrote about that?

    You aren't going to measure the success of my student who got a D, but said he scared himself, opening the math book after the semester was over. I think many of my students are taking charge of their learning in ways they haven't before. It may not show up in their grades right away, but it's a huge change for the better. Probably not measurable...

    Am I making sense?

  6. And thanks for the pushback, it makes me try harder to explain myself.

  7. Yeah, the rest of the article makes this sound like a seriously bad idea. I was just narrowly focusing on that one argument (another weakness of mine, probably).

    ... the most important parts aren't easily measurable.

    Yeah. And then there's the incentives you create for teachers who do have to respond to the tests. I know.

    Am I making sense?

    Definitely! :)


Math Blog Directory