I will protect your pensions. Nothing about your pension is going to change when I am governor. - Chris Christie, "An Open Letter to the Teachers of NJ" October, 2009

Tuesday, August 7, 2012

The Incoherent World of Arne Duncan, Part IV

More of my breakdown of this revealing interview that Tom Moran of the Star-Ledger did with US Secretary of Education Arne Duncan. Collect each exciting issue!
  • Part I: Duncan's signature program, Race To The Top, has little evidence to back it up.
  • Part II: The real legacy of Duncan, Rhee, and Klein
  • Part III: Arne has a lot to learn about what makes teachers tick.
I must be a glutton for punishment, but here's some more to debunk:
Q. Let’s talk about tests. When scores were released in New York recently, we saw that some teachers were rated as great one year and awful the next. It dissolved a lot of confidence in test results.
A. You have to look at multiple measures. You look at value added over a course of years, not just one. You have multiple observations, too, from principals or peers. And you can ask the children, with student surveys.
So you need multiple measures because the teacher ratings based on tests are inaccurate?! Then why do you insist on using something that you don't deny is prone to big errors in the first place? 

Diane Ravitch calls test-based teacher evaluation "junk science," and she's right. Everyone who has any credibility in this field knows that value-added modeling (VAM) is going to produce large numbers of inaccurate teacher ratings.

And if Secretary Duncan would just listen to people who know a bit about statistics - like, say, Bruce Baker - he'd learn that even using VAM for just part of the evaluation can skew all of the high-stakes decision. But amazingly, Duncan doesn't seem to care. He thinks it's fine when good teachers are wrongly fired and bad teachers wrongly stay on the job:
Q. But doesn’t this whole movement to raise standards and impose accountability hinge on having good tests? I’m not sure we have that.
A. It’s never going to be perfect. We’re investing $350 million in the next generation of assessment, so it’s going to be a choppy couple of years until we get there. We always let the perfect be the enemy of the good in education, and we have to stop that.
Lord forbid we spend that $350 million on, oh, you know, schools. Duncan is so foolish he thinks throwing money at testing will somehow automatically make the tests more accurate. I'm sorry, Arne, but life doesn't work that way. As long as you have companies incentivized to produce and grade these tests on the cheap to maximize their profits, you're going to have tests with bad questions that are poorly graded and designed to meet pre-determined outcomes.

But let's suppose Duncan's dream does come true, and the tests magically do get better in a few years. Why, then, is he bribing states with Race To The Top money now to implement VAM-based measures as soon as possible? He admits we still have "a couple of choppy years" ahead of us. Shouldn't we be pilot testing all of these measures first to find out if they work before we implement them across the board? Shouldn't we make the decision based on evidence and research?

Baker has made the case before that basing teacher evaluations on testing is literally as accurate as rolling dice. I guess that's good enough for Arne; he doesn't think any real harm can come of it. Tragically, he is dead wrong.

It's worth reminding everyone that Arne Duncan thought it was just fine when the LA Times released its highly inaccurate teacher ratings to the public; he didn't see anything wrong with shaming teachers with error-ridden data. But when the release of ratings in New York City turned into a fiasco, he changed his mind. Too bad he changed his mind too late to help the teachers of Los Angeles and NYC who had their lives turned upside down.

Duncan's insouciance in the face of his misjudgment and vacillation is infuriating - especially when we remember the very real consequences for people whose only sin was to dedicate their lives to working with children. On the basis of that incident alone, I believe Duncan should have resigned or been fired; he clearly does not have to judgement necessary to lead this nation's school systems.

Believe it or not, there's even more of Duncan's foolishness to get to in this interview; stand by (if I can quell this rising nausea...).

2 comments:

Galton said...

Jazzman,
Hysterical, sad and true; allat the same time. I believe Duncan will cost Obamma many votes and perhaps even a swing state or two. he should ask Arnie to resign and appoint someone like Linda Darling- Hammond.

thanks for the great coverage...

Ken Houghton said...

"Lord forbid we spend that $350 million on, oh, you know, schools. Duncan is so foolish he thinks throwing money at testing will somehow automatically make the tests more accurate. I'm sorry, Arne, but life doesn't work that way. As long as you have companies incentivized to produce and grade these tests on the cheap to maximize their profits, you're going to have tests with bad questions that are poorly graded and designed to meet pre-determined outcomes."

It's actually worse than that. If you give me an additional $350MM, I'm going to assume I'm doing things correctly and do more of the same.

The tests will get worse as you allocate more money to them; the equilibrium they approach will not be the optimal one.