Wednesday, May 6, 2015

Not 'diagnostic' in the sense that it aids in diagnosis

I've been meaning to do a deep dive into how the language of the reform movement reflects its close connection with management consulting. If you've ever heard one of these consultants making a pitch to a high level executive, you may have noticed that while the words always have the connotation of precision and scientific rigor, the statements tend to be vague and inconsistent.

Diagnostic data needs to be specific and (to borrow a term from the business lexicon) actionable. It should also generally be multidimensional. On the individual student level, the tests being discussed here are not diagnostic data -- the results are neither timely nor specific enough to be actionable -- but Tisch suggests they are there to provide that level of information.("It is natural for parents to want to know how their kids are doing").

Check out this exchange. (emphasis added)

HAYES: OK. But there’s a whole — let’s sort of bracket the sociology of poverty for a moment… there’s lot of things I could contest about in what direct the causation of that link flows, right? But here’s the question to you, Miss Ravitch…I had someone who works in education who I respect compare testing opt out to people opting out of immunization, because basically it was like, look, yeah, your kid is probably not going to get measles and, like, if you think there’s some downside you can opt out, but then you’re just a free rider because the policy as a whole is a necessary means of figuring out where people are, assessing, right? You need this data.
If you start allowing people to opt out, you’ve just destroyed the entire dataset. Like, what are they going to do in West Seneca to judge anything year over year when one year they have data and the next they don’t have any data?
DIANE RAVITCH: It’s totally inappropriate to compare opting out of testing to opting out of immunization. One has a scientific basis, the other has none. The tests that kids take today have nothing to do with the tests that we took when we were kids. When we were kids, we took an hour test to see how we did in reading, an hour test to so how we did in math. Children today in third grade are taking eight hours of testing. They’re spending more time taking tests than people taking the bar exam.
Now, when we talk about the results of the test, they come back four to six months later. The kids already have a different teacher. And all they get is a score and a ranking. The teachers can’t see the item analysis. They can’t see what the kids got wrong. They can’t — they’re getting no instructional gain, no possibility of improvement for the kids, because there’s no value to the test. They have no diagnostic value.
If you go to a doctor and you say, ‘I have a pain,’ and the doctor says, ‘I’ll get back to you in six months,’ and he gets back to you and tells you how you compare to everyone else in the state, but he doesn’t have any medicine for you.
HAYES: Respond to that.
TISCH: Well, I would say that the tests are really a diagnostic tool that is used to inform instruction and curriculum development throughout the state. New York State spends $54 billion a year on educating 3.2 million schoolchildren. For $54 billion a year I think New Yorkers deserve a snapshot of how our kids are doing, how our schools are doing, how our systems are doing. There is a really important data point…
HAYES: Wait. … I just want to point out something. That was interestingly nonresponsive to what she said, right? She’s saying this does not work as diagnostic tool for the child or for the teacher, you’re saying this is a diagnostic tool for the taxpayer who is funding the system to see if the system is working, right? Those are distinct.
TISCH: No, let me finish because we’re talking about what happens when parents opt out and what the system can then report back to parents and to the state. The point of the matter is, you know, two weeks ago I was with my grandson at a pediatric visit. There was a new mother sitting next to me and she was comparing growth charts for her 4-month-old son. She wanted to know how he was doing on a continuum.
It is natural for parents to want to know how their kids are doing. And as for the diagnostic nature of these tests and the amount of
information that is gleaned from them, school districts report to us all the time that they design curriculum around the results of these tests.
I agree with Diane. There is no such thing as a perfect test, absolutely not. But the ability to glean information from these tests and
use them in very direct ways to inform instruction and curriculum in classrooms is actually really important.

If these tests aren't diagnostic, what are they? Mike the Mad Biologist has a suggestion:

What Ravitch touches on, but I wish had made more clearly, is that these tests are not about assessing individual students. The rhetoric Tisch uses is disingenuous, as the tests can not–as a matter of education policy and contractual obligations with test providers–to tell individual students (and their teachers and parents) where they need to improve.
The tests exist solely to grade teachers. These are not educational tools, as Ravitch notes, but managerial ones. They are used to hire and fire teachers. That is why the NY Department of Education is panicked by the opt-out movement. It’s not the potential inability to assess state-wide or even school level student performance (certainly for the former, there are enough students in the state of New York taking the exams for the statistics to work).
No, it’s the possibility that the state won’t be able to evaluate individual teachers with the exams. I’ve discussed many times before how sample size issues make teacher evaluations incredibly imprecise and are inappropriate in hiring and firing decisions. Imagine if a significant number of teachers can’t be evaluated because too few of their students decide to take the tests (there aren’t a whole lot of strong conclusions that can be reached if only eight students per class take the tests). It certainly would give grounds for teachers to challenge the conclusions drawn from the tests.
What Tisch doesn’t want to say out loud, what she politically can’t say out loud, is that she, along with many other reformers, believe if only we could fire the bad teachers–and she believes there are a lot of them–then our educational problems would vanish. But many reformers, having realized the majority of parents* don’t believe this, understand they can’t explicitly make that claim. So they lie about why we supposedly need annual high-stakes* testing.

1 comment:

  1. It's not just that eight kids are not enough to make good decisions, it's that they are likely to be a biased sample of kids in a class as well.

    When the tests don't matter to the students than they can jerk around which means any statistics gathered are pretty much useless.

    But the biggest thing about the tests is that students aren't randomly assigned to teachers - any good school will try to match kids with kids who work well together and with a suitable teacher. It's not fair to judge a male teacher badly because he's been given a bunch of boisterous boys that noone else can manage or the empathetic female teacher who has been given the anxious girls who find tests very scary.