We’re concerned about the effectiveness and reliability of requiring states to link data on student achievement to individual teachers for the purpose of teacher and principal evaluation. Teachers who work with disadvantaged students shouldn’t be “evaluated” based on whether their students hit a particular test target on a particular timeline. And we certainly shouldn’t base additional compensation on whether students meet particular testing targets on a particular day. We need to offer incentives so that our best teachers teach the students most in need of assistance, not incentives to teach students most likely to score highest on a standardized test. As with NCLB, good intentions can lead to unintended—and unacceptable—consequences.
An interesting sort of political science question is just how long the NEA can get away with vague and disingenuous comments like this. Nobody--nobody--wants to judge teachers based purely on the percentage of students who meet a given cut score on a test at the end of the year. That would be crazy. All reasonable conversations about the use of end-of-the-year student test score data for teacher evaluations begin with the assumption that we should (A) account for where students were at the beginning of the year and/or take into account other data about their academic histories, and (B) not rely exclusively on test-score data. Van Roekel knows this. Everybody knows this. The people pushing for the student data-teacher linkages are the same people who want to get more high-quality teachers in the classrooms of disadvantaged students. They're not idiots; of course they don't want to create an unfair evaluation system that would directly counter that goal. But conceding that makes the whole thing seem a lot more rational and then where would the NEA be?
Also, it's easy enough to raise the specter of measurement error by talking about "particular testing targets on a particular day." But that requires a level of seriousness and empiricism about how the numbers tend to play out. Measurement error is real and significant but also finite and measurable and thus subject to sensible decision-making and interpretation. If a given teacher's students all consistently fail to meet a particular target on a particular day, year after year, even though most of them were hitting targets in previous years--hey, that might mean something! But acknowledging that would take the conversation to a place the NEA clearly doesn't want to go.
7 comments:
I have to disagree that nobody wants to judge teachers based solely on test scores. In NYC, the mayor's emphasis is almost entirely on test scores, and administrators regularly browbeat everyone into passing as many people possible by virtually any means necessary.
I actually, by virtue of my own insanity, volunteer to teach ESL kids to pass the English Regents exam, a test that's totally inappropriate for them. No one really wants to do stuff like this, and I could certainly serve them better by teaching them how to speak and use English.
But they can't graduate unless they pass this test, someone has to do this job, so I do it. I could teach what I'm supposed to, ESL, and avoid standardized test scores entirely. I may indeed do that one day if enough pressure is put to bear on me. Who wants this hassle? No one, and most teachers, by virtue of teaching courses that don't include standardized testing, don't have it.
I'm sorry, but there is enormous pressure to pass as many people as possible where I am, in the biggest public school district in the country. It's been in the papers.
I've never seen principals evaluate teachers based solely on student test scores; however, I've not seen the concept of merit pay so implicitly tied to student test scores as it is by RTTI. While the rhetoric of evaluation remains reasonable and nuanced, is it possible that merit pay will be tied to a single-measure?
Actually, they are idiots. The only way to get a good teacher value-added measure is to test once at the beginning and once at the end of the year. Prior history doesn't cut it because, especially in high poverty schools, students can slip a lot over the summer. The tests must necessarily be quite extensive in order to correctly measure the child's beginning and end-of-year attainment. So you are essentially doubling the already-awful NCLB testing regime, with kids getting hit with one set of tests first thing in September. Parents in high performing suburban schools are already restive. Just watch how they react when you double down on the testing regime. If Obama and Duncan think they're taking flak for sending their kids to private school now, they ain't seen nothin yet.
Sorry, I forgot to add that you'll be running the doubletime testing for 3 years minimum before you've got sufficient data to calibrate the models and do anything "useful" with it. And how useful it will actually turn out to be is a big question mark. Then ya gotta think of the fallout when some doofus superintendant fires a beloved teacher on the basis of some computer error, cause you just know that is going to happen. And the high-powered academics from Stanford and Harvard scratching their heads on TV saying "We told them this was a happen, but you know how these K-12 folks are with their fads..." You think there's backlash against NCLB? Better buckle up now, cause it's gonna be a bumpy ride.
If you call that "willful misunderstadning," what you call it when other refuse to looking into matters like reliability and instructional sensitivity?
Making this kind of thing (e.g. value added measurements of teachers) work is a very technical matter, and some people steadfastly refuse to look at the technical obstacles. What should call them?
I hope we are smart enough to realize that this is really a debate on how to curb teachers pay and nothing more.. The poor performing inner city students are being used as a hatchet against the teachers unions in order to control salary.
When an enthusiastic follower told Adlai Stevenson that "every thinking American will vote for you," he replied, "Yes, but it takes 51% to get elected."
Only the most willful of the teacher-bashers will simply link test scores with evlauation, but there are plenty of them around. And Kevin enthusiatically supports some of them.
Yes, making VAM "work" is more complex but it is the same concept. Just as it as absurd to evaluate all teachers regardless of student background on a test score snapshot, it is equally absurd to evaluate diverse teachers on a snapshot of "growth."
Use growth models for the purposes that they were designed and we can get constructive data-informed collaboration. Encourage misuse of VAM for evaluation and you will get the unintended disaster described by Van Rockel and the other commenters.
I know we are asking noneducators like Kevin to look at the world through our eyes and see through some nuance, but we need to find a way so "refomers" understand that Kevin is offering a distinction without a difference.
Just as students come to schools with differing backgrounds and come to classes with differing levels of skills, they grow at different rates. How do you think the gaps got their in the first place? Had the poorest and most traumatized been increasing their knowledge for their entire lives at the same rate, the gaps wouldn't exist.
Kevin is just using some fancy words to get arouund his logic. He wants to end achievement gaps by fiat. He orders us teachers to end inequality in America, and if we don't do it single-handingly, then we are blameworthy and "reformers" should use whatever tools that are available to destroy us. As that happens, he'll replace the "status quo" with ...
Post a Comment