Saturday, May 22, 2010

Why Evaluation Systems must Change

Teacher evaluation systems around the country are in need of reform. But the most important reform that can be made is to link pay and performance. To be sure, evaluations as they exist today are tied too closely to over-worked and thus under-attentive administrators. This is why any new evaluation system must be on-going, robust, and involve multiple administrators and stake-holders. Parents, students, teacher-colleagues, subject specialists, and administrators must all play a role in the qualitative data gathering process. It is also clear that student performance data should be the wheels of any quantitative data gathering process. Considered jointly, these data points can successfully provide a window into the effectiveness of a teacher.

Currently, however, most evaluation systems do not do a very good job identifying struggling teachers, or in the very least, they do not effectively identify very many struggling teachers. As Jay Mathews notes in his Washington Post blog:

I asked some Washington area school districts, nationally regarded for their high performance and sophisticated management, what percentage of their teachers
were rated satisfactory in the last evaluation. I subtracted the number they gave me from 100 percent to calculate what portion of their teaching ranks could be considered ineffective. The results:

Fairfax County 0.9 percent, Montgomery County 5 percent, Loudoun County 1 percent, City of Falls Church 0.45 percent, Prince William County 1.7 percent. I could not find a school district in the area that admitted to having more than 5 percent unsatisfactory teachers. Most said the figure was closer to 1 percent. Experts tell me this is common throughout the land. I don't believe the numbers.

The evaluations are usually done by principals and subject specialists. I suppose they could be told to identify the bottom 10 percent, as some companies do, and offer them the severance funds, but as Hanushek points out their union representatives might object to that approach.

So the question remains, why are so few teachers identified as ineffective? Several years ago, MCPS asked itself this very question. It is unclear they had any definitive answers. I will spend some time now trying to answer this question.

In part, the answer is related to the fact that there is usually only one evaluator with the soul responsibility of determining whether a teacher should be identified as ineffective. But secondly, and most importantly, it is because there is virtually no incentive for an administrator to make this identification. And without performance based pay, this is unlikely to change in the near future.

In my next entry, I will attempt to show, using my admittedly limited knowledge of game theory, why so few teachers are identified as ineffective. I'll also offer a solution for improving the current state of teacher evaluations.


  1. I find that 5% claim suspect.

  2. Game theory, huh? I'll keep an eye out for that one! And thanks for linking to InterACT, where we're (hopefully) engaged in similar and useful exploration of these issues.

  3. Yoremo: I'll email Jay Mathews and see if he responds.

    David: Thanks for the note. I've learned alot already from reading about what your group is doing.