Michael Friedman tackles the topic of whether testing produces better teachers:
It seems obvious - impose qualification tests on teachers and you will improve teacher quality, thereby improving teaching. Unfortunately, however, according to an NBER report last year, it doesn't work.
Does Teacher Testing Raise Teacher Quality? Evidence from State Certification Requirements by Joshua Angrist and Jonathan Guryan. [Note - I don't have a subscription, so I'm going off the summary and Friedman's comments.]
Angrist and Guryan look at teacher quality and compensation in states and districts that require and do not require certification testing.
Obviously (although not examined by this study), certification testing eliminates some of the very worst teachers. Logically, by reducing the supply of possible teachers, certification testing increases salaries for teachers. The effect for teachers with BAs by an average of 3.3%. Disturbingly, certification testing, by increasing the costs of being a teacher, disproportionately discourages the very best teachers who have other options besides teaching.
Angrist and Guryan believe this type of impact is "ambiguous," and I agree. I find it hard to believe that certification tests would be more likely to discourage the best, rather than the worst, teachers from going into the profession, but I suppose it all depends on how one defines "best" and "worst."
In the Angrist and Guryan paper, teacher quality was estimated by average SAT score of the undergraduate institution from which each teacher graduated. I find this a very odd estimate of teacher quality, unless one is willing to assume two things. One, that a teacher's competence in the classroom is very strongly correlated with the average SAT score of their institution, and two, that the SAT scores of students in education programs are comparable to the SAT scores of their instituations as a whole. But students who enter education programs tend to have the lowest SAT scores of any students; only those planning to major in Public Affairs or Home Economics have lower averages. Thus, the second assumption isn't valid. It might still be the case that teacher quality is correlated with institutional SAT mean, but I've never seen any research to support this assumption.
When using this estimate of teacher quality, what Angrist and Guryan apparently found is that state-mandated certification testing didn't necessarily identify the teachers who came from the schools with the highest SAT means. Again, this doesn't really mean anything unless we believe in this definition of teacher quality, and I'm skeptical about that.
Friedman has this to say:
Angrist and Guryan recognize that this result is not very helpful. One possibility is to set a floor on teachers' SAT scores. I can hear Kimberly Swygert reacting already... The SAT score has never been validated to confirm whether teachers with higher SAT scores teach students better.
True, but my reaction to this idea isn't as negative as Friedman assumes. Yes, the SAT has not been validated as a measure of overall teacher quality, but it's certainly a measure of academic aptitude. It would be expected (if as yet unverified) that potential teachers with higher SAT scores would have a stronger grasp of academic concepts and core subject matter. While the grasp of the subject matter is not the only element that goes into creating a competent teacher, it would not be unreasonable to perform a study that calculates a different measure of teacher compentency, and then correlates it with SAT scores. I'd be very surprised if those teachers who had SAT scores below a certain level were found to be competent by any measure, no matter where they got their degrees or how much they loved dealing with kids.
Another possibility is to increase the rewards for the best teachers by offering higher salaries in their first few years to teachers who do particularly well on certification tests. That could reduce the incentive for the best teaching candidates to find other occupations.
True, but I still believe that it's the worst, not the best, teachers who are the most likely to be scared off by the idea of a certfication test. And by "worst" here I mean both those teachers who performed very poorly on the SAT, and those who graduated from ideology-heavy, anti-testing, "progressive" education programs in which fostering self-esteem is more important than fostering academic progress.
Whatever the answers are, however, this study is very disconcerting to people like me who instinctively look to standards based initiatives to improve the quality of schools and students. I have no problem if testing raises teachers' salaries if it also improves the quality of teachers. If it just raises salaries then what is the point?
Again, I didn't reach the same conclusion, because I disagree with this study's estimate of teacher quality. I believe that standards-based certification tests can and should be used, and that teachers should be required to demonstrate skills in the subject area they teach, in addition to their classroom-management and "theory of education" knowledge.
Posted by kswygert at March 8, 2004 09:13 AM