What the hell, let’s get in trouble.

Go take a look at this article first, and then Diane Ravitch’s reaction to the article.

What, you didn’t read?  Okay, I’ll summarize.  Alabama thinks this is OK and Diane’s mad; the slightly ungrammatical intro is a quote:

• These are the percentages of third-graders expected to pass math in their subgroups for 2013 are:
• – 93.6 percent of Asian/Pacific Islander students.
• – 91.5 percent of white students.
• – 90.3 percent of American Indian students.
• – 89.4 percent of multiracial students.
• – 85.5 percent of Hispanic students.
• – 82.6 percent of students in poverty.
• – 79.6 percent of English language-learner students.
• – 79 percent of black students.
• – 61.7 percent of special needs students.

I can hear your brain, you know.  “What?” you’re thinking.  “That’s terrible!  How dare they set lower standards for everybody other than Asians and whites!  That’s racist!”

She’s hollering and yelling about it and so is everybody in her comments (most of ’em, anyway) and I barely had the strength to wade through a third of the comments on the original article.

Let that idea simmer a little bit; I’m sympathetic, believe me.  I’m gonna change the subject for a bit but I’ll get back around to this.

Let’s assume that, as we do at the moment, you believe that standardized tests of some kind are a good way to assess student learning.  For the moment, I’m not going to debate whether that’s actually true or not; let’s just mutually decide that we think they do.  It’s good enough.  If you care about standardized test scores, there’s a couple of different ways to pay attention to how schools/classrooms/teachers/kids/whatever do on them.  The first, which is the method No Child Left Behind followed for a decade or so, is the raw pass rate, and I think the pass rate is the number that most people are accustomed to wanting to look at.  You just calculate how many kids passed out of how many kids took the test, and you’re done.  One number, easy to compare to other numbers.  It’s great!

If you tell a school, or a district, or whatever, that they’re going to be judged solely on how many kids manage to pass Test X, here is what’s going to happen:  the “bubble kids” suddenly become the most important students in the building.  Billy, over there, has parents who made sure he was reading before he entered preschool.  His house has thousands of books in it, his parents both have graduate degrees, and he had the highest test score in your school last year.

You are not going to pay any attention to Billy.  Billy’s going to pass unless you blind him before he takes the test.

Shirley, on the other hand, is the product of a single-parent home and has an array of learning disabilities.  The only book in her house is the Bible, and no one in the house can read it.  Her mother has been unemployed for eight months and did not graduate high school.  Shirley currently lives with her mother, her aunt, and the six other children they have between them and their home situation is at best wildly unstable.  She struggles mightily with material four or five grade levels below her age.

Screw Shirley, too.  It doesn’t matter how much effort you put into her; she’s not gonna pass no matter what you do unless you take the test for her.

The kid you’re looking at is William.  William’s pretty bright, usually, but how well he does depends on his mood.  If he’s paying attention, and if he’s taken his meds, he can be a good student.  He struggles, and needs additional help, but he generally isn’t the type to just give up on you; he’s a hard worker if he thinks you like him.  He didn’t pass last  year, but he was close.

Your Williams are now the most important kids in your classroom.  Those kids– and they can be a large chunk of your room depending on what kind of school you teach at– are going to determine whether you pass or fail, or whether your school stays afloat or goes on probation.  You’re going to spend the lion’s share of your energy on the kids who have a chance to pass the test but aren’t guaranteed to pass the test.  The rest of them are what they are; one way or another, it’s a waste of energy.

Sad but true fact:  Billy is more likely (but not guaranteed) to be white or Asian, and Shirley is more likely (but again, not guaranteed) to be black or Hispanic.  William is a little more blended but he’s a bit more likely to be some shade of brown than otherwise.  The so-called “achievement gap” is so pervasive in American schools that I’m not going to waste the breath on talking about it beyond this sentence.  It exists.  It sucks that it’s real, but it is.

If you focus on pass-only as your measurement method, you’re going to get schools and teachers who focus solely on the middle rather than the top or the bottom.  Kids who can’t be moved from fail to pass, or who won’t move from pass to fail, are a poor use of effort.  You’ve got to get each and every one of those bubble kids because we have to show “improvement” on our grade’s numbers from last year, even though these aren’t the same kids we were measuring last year, they just happen to be the same age as those kids were.

This, obviously (I hope) makes no goddamn sense at all.

Here’s a better way:  pay little or no attention to pass rates, and instead focus on improvement.  Okay, it’s nice that Billy passed.  Did he do better this year than he did last year?  By how much?  If the answer is “yes,” you’re doing a good job.  If the answer is “no,” there’s a problem.  If you have a bunch of Billys and they all did worse than they did last year, you have a real problem.

Shirley?  We’re not so worried about whether Shirley passes right now.  Shirley entered sixth grade reading at a first-grade level.  If you moved Shirley up to third grade in a single school year, you did a good job.  Same as William– his score was 40 points higher than he did last year, but the pass cutscore moved up by 45 points.  He still didn’t pass, but he did better than he did last year.  He showed improvement.  This is a good thing.

It is also hideously complicated, and that hideous complication is why we tend to focus on pass rates instead, even though focusing on pass rates is stupid.  Pass rates aren’t complicated and they aren’t hard to explain.  Indiana, in particular, uses the improvement method, technically called a “growth model.”  Every kid that takes the ISTEP is compared to every other kid who got the same score they got last year, and then they’re ranked as either a High Growth, Average Growth, or Low Growth kid.  The problem is that sometimes you’ll get some random-ass score that not many people got and a “high growth” score is a point.  Or– and I saw this happen– some kids will lose points and still be High Growth, and other kids will gain immense amounts of points but because everyone who got a 512 last year ate their Wheaties before the test this year, that will somehow count as Low or Average growth.

It’s still better.  Indiana’s model has stupidities embedded in it– those kids I talk about in the last paragraph are not hypothetical– but it’s still better than relying on pass rates, because every kid’s scores count.  I don’t want to just focus on my bubble kids.  I want to focus on everybody, because Shirley having a bad year hurts me just as much as William or Billy.  When we get down to the nitty-gritty details of who’s High Growth and who’s Low Growth I might find some places to quibble, but as an educator I can’t afford to prioritize one group of students over another.

And that’s a good thing.  We want that.

---

Back to NCLB for a minute.  One of the bigger pains in the ass of NCLB was the way it disaggregated groups of kids into subgroups.  Not only did your school have to pass a certain percentage of its kids to stay in good standing, but a certain percentage of your black kids and a certain percentage of your ELL kids and a certain percentage of your special ed kids and a certain percentage of your girls and a certain percentage of your blah blah blah blah blah all had to meet Adequate Yearly Progress goals.  If just one of your subgroups– and these damn things could be on the level of a single family in a smaller school– was out of compliance, the whole school was.  God help you if you had a diverse student body.  Somebody was bound to mess up; you were screwed.  It made being in a basically segregated school a good thing.  If everybody was black (and my school in Chicago was) then you didn’t have to worry about eight different racial subpopulations screwing your numbers up because the Garcias or the Nguyens just got here last year and their kids don’t speak English well enough to pass the test yet.  Or because your school has a reputation for having a great special ed department, so you have lots of special ed students because parents fight to get their kids in your school– which means when that large group of special ed students don’t make AYP, your whole building is labeled “failing” because of the very thing that made your building successful.

It sucked, mightily.

What we’re seeing, up there, in that looks-really-racist chart, is a combination of a growth method and NCLB’s disaggregated student populations.  They’re acknowledging that the achievement gap exists.  We cannot simply state that we want 80% of everybody to pass (to pick a number at random) because it’s unrealistic for all of our groups of kids.  Instead, what we want to see is for everyone to make improvement.  Well, to improve, our Asian kids have to hit a 93.6% pass rate.  Our black kids aren’t passing at the rates that the Asian kids are.  We still want improvement from them, but for that group, a 79% pass rate represents improvement.  We’re still trying to bring up all of our kids; we’re just realistic about how much we might be able to bring them up from year to year.

Of course, when you present it that way, without– at the moment– 1764 words that no one will read of explanation first, you look racist as hell.  And it was a terrible mistake for the state of Alabama to release these numbers like this.  But it was a political mistake, not a pedagogical one.

Tomorrow– because this is already too long– I’m going to talk about what it actually means to have “high standards” and “high expectations” and how it works with this type of model.

Unless, of course, I come up with something more interesting.