Tuesday, August 17, 2010

LA Times Article on Evaluation of Teacher Performance

This LA Times article on comparing teacher performance is a must read for anyone interested in education issues. Essentially, they used statistical analyses of how a teacher influences the performance of his/her students relative to teachers in the same school which should control for student background and other variables. A lot of the comments following the article take the LA Times to task for teacher bashing. Do you think the article is unfair? If so why?

A few take home messages:
A) The district never performed a similar statistical analysis to show their teachers how they're doing at teaching math and English basics.

B) It is almost impossible to fire teachers that are demonstrably ineffective. That doesn't mean they don't care or aren't personable. In fact, experience, education or personality type don't necessarily tell you which teachers are actually effective at teaching basic skills.

C) Apparently the teachers union is contemplating boycotting the LA Times for this article. What?!? This is some of the better journalism I've seen lately. Asking tough questions of powerful monopoly institutions and demanding performance is what our journalists SHOULD be doing.

D) I don't think anyone would advocate using student performance as the ONLY factor in a teacher's evaluation. The problem is, it currently is not used in the LA school district--or most I would imagine. Why shouldn't student performance be part of the equation?

Here's a rebuttal claiming that the value-added statistical analysis is flawed for individual teachers because student assignments aren't random. It think she's probably wrong because the analysis still factors in a student's individual performance in prior and subsequent years so you can spot a year where a student did worse than their normal performance. If this pattern repeats for most of the students in the class then you should be able to draw some robust conclusions and this should control for some slight heterogeneity in classroom populations.

An FAQ on the methods used in the analysis.

The technical nitty gritty (I did not read this).


  1. I think it's a very good article and study with sound methods, but I can see why it got some people upset. No one likes having an objective lens pointed at them when it is going to show them in a less than favorable light. Singling out some of the poorly performing teachers by name in the article is surely going to cause some administrators headaches as parents demand their kids be moved out of those classes. Administrators tend to back up their teachers in most cases, but this will give the parents some ammunition that they will have a hard time refuting. However, having this kind of information available can only be of benefit to the schools and students overtime.

  2. Right. I feel for the teacher who was presented as less effective--but he appreciated knowing how he was doing so he could consider ways to improve. I think most teachers would want this information--but probably wouldn't want it to be broad public knowledge.

  3. Cool article. I'm amazed that the school district collected these data and then did nothing with them (only the most pretentious people insist on recognizing that 'data' is plural). It's really expensive.

    Anyway, it'd be great if student performance was actually used in teacher evaluation. It shouldn't be the only part of an evaluation, and the technical study suggested as much (I noticed an article by my intermediate micro professor from BYU was cited in the paper, by the way), but if experience and training don't mean anything, student performance has to be included.

  4. This article points out a lot of the reasons I'm a fan of charter schools, both for my own employment and my kids' education.
    1. the principal is also the boss. She can raise (or not raise) salaries as she sees fit to keep the right people on the bus.
    2. no unions or tenure. If someone isn't pulling their weight, they're more easily dismissed.

    What the article doesn't point out is how hard it is to be an effective teacher like the ones they mention. Knowing how many teachers we need in this country, and how much they get paid, how can you expect them all to be like that Miguel Aguilar?

  5. Great question Brandon. We know some teachers are more effective than others, now what?

    1. If a teacher is consistently underperforming they should be warned and given some professional development opportunities, maybe twice. If after a third year they still underperform then they need to be let go. I imagine each district or school can decide how they define underperform. I would imagine it would have to be some pretty low, relative peformance benchmark.

    2. How to get more teachers. Well, I'm a fan of merit pay, vouchers, and charter schools. Anything that can create incentives for performance will attract people who want to perform.

    3. They can't all be like Miguel Aguilar, but the study points out skills the successful teachers appear to share. Teachers should be taught those skills and expected to meet some minimum benchmark. As a teacher, I know if I set expectations then the students who want to succeed will meet them--and as long as the expectations are clear and reasonably attainable most students will meet them. I don't think most of us are any different.

  6. One thing that Regan brought up about this Value Added measurement is that you can't expect test scores on standardized tests to increase indefinitely. They're standardized, which means that scores are re-centered periodically, so that continual improvement of average scores is impossible.

    I assume that the people who use test scores to do these kinds of academic studies understand this. I'm not so sure that people who administer No Child Left Behind do. I saw Bush's secretary of education on Jeopardy and she was aawwwwwful.