There’s some very interesting public testimony by Luke Flynt, a Florida English teacher. At the end of the post, I’ve included the video–it’s short and very clear (maybe because Flynt is a professional teacher. Just saying). But his testimony about value-added testing is chilling. Basically, in value-added testing, a student is assigned a predicted score based on her previous performance along with demographic characteristics. If a student scores higher than predicted, the teacher is credited with ‘adding value’–increased the student’s score more than expected. If a student scores lower than predicted, the teacher is assumed to have ‘subtracted value’–decreased the student’s score more than expected*.
In the video, he describes how value-added testing has hurt him professionally, and, thanks to ‘incentives’, financially, even though many of his students do very well on the FCAT reading exam:
- According to the value-added model used, four of his students would have had to received scores that were literally impossible, or else the model would have stated that Flynt did worse than he should have with these students. The maximum score on the sixth grade FCAT is 283–meaning every question was answered correct–yet four of his students would have had to score several points higher according to the model. In other words, there was no possible way for Flynt to have taught these students adequately (in fact, one of these students did get a perfect score).
- Fifty of the 102 students used to determine Flynt’s value-added measure (‘VAM’) fell short of the model’s predictions. Of those students, 29 (58%) got at least 41/45 questions right on the exam. Five aced the test or got one question wrong. Remember these are the students Flynt supposedly did a poor job of teaching.
There’s simply no way for a teacher to perform well under this kind of regime. This is yet another case where the rhetoric of education ‘reform’ collides with the reality–and gets dragged fifty yards:
Whether it’s screwed up teacher evaluation metrics or assessing teachers based on students they haven’t taught (not making that up), it’s safe to say that the implementation of reform as it exists–not some magical non-existent reform movement existing only in pundits’ heads–is really screwed up. We give teachers–who are human just like the rest of us–a perverse set of incentives and then wonder why parents and students (not to mention many of those same teachers) are really angry. It’s simple: education reform rhetoric has nothing to do with reality. In many ways, it’s analogous to the ‘liberal hawks’ who foolishly believed that Bush et alia would fight wars and manage occupations the way the liberal hawks fantasized, when all of the Bush Administration’s actions to that point put the lie to that foolish, naive belief.
I would love to see curricular and pedagogical changes and improvements [as would many teachers], but the reform movement, with rare exceptions, isn’t about that. Reformers, who in many places, have been given wide latitude, and they have constructed a giant pile of idiotic malincentives and nonsensical policies.
What’s even more disconcerting is, like many reform policies, a lot of effort went into its design, and this was the end result.
*And if you think about it, value-added testing is, by definition, zero-sum: some teachers will have to do worse–that’s the underlying mathematics of the model.
Whom the gods would destroy …
Obviously the people who developed that VAM regression equation (aka random number generator) were leading statistians and methodologists.
Pingback: The meaning of the test cheating scandal in Atlanta | Phil Ebersole's Blog