Revisiting my prejudices
Gepost door RBL op 03/03/2010 om 22:27
Toegevoegd onder: Uncategorized
So you will be surprised to discover that, when it comes to the “problem” of my former alma mater and its supposed reincarnation, I am like a dog with a pig’s ear.
I just can’t let that shit go until it is gnawed to ribbons.
Having now mastered a fancy new statistical technique, I turned my attention to a pair of related, and funny, little problems:
a.) API scores are going up, up, up, like a boner a Brittany concert, while
b.) SAT scores are flat, flat, flat, just like a white boy’s ass.
I have to do something to liven up a discussion of statistics, don’t I?
Anywho, so there’s that conundrum: academic performance index scores are rising, across the board (and even, for whites and Asians, converging to some degree — meaning the “spread” of variation across districts and/or schools is narrowing), while SAT scores are flat enough to raise an eyebrow (there is usually some noise, after all, even with only three years’ data).
How is this possible?
Being a good sociologist, I offer two competing hypotheses:
a.) API scores are utter horsehit.
b.) API scores mean something.
If (a), then their rise portends precisely nothing, and we should find no relationship between the factors associated with said rise and that which we know to be a reliable indicator of, well, something anyway, namely the SAT.
If (b) , then their rise portends a more complicated story than that which I have heretofore been prepared to admit.
So I ran a regression, predicting API scores from what scanty data are available from the CA Dept. of Ed website, which is to say, parental education, percent non-white, and location. I included only those metro areas of Northern California that exceed 500,000 people: Bakersfield, Fresno, Oakland, Sacramento, Stockton, San Francisco, San Jose, Santa Rosa, and Vallejo. Because, really, who gives a shit about Chico, anyway? I ran a couple different versions, just to test out my prejudices, and basically found what I expected.
The story is not particularly complicated, and I don’t actually need to revist my prejudices. Except in one major regard, one that puts me in a very, very old school conservative camp.
Which I do not like one bit.
API is a product of (a) parental education, (b) ethnic heterogeneity (which is a good thing, mind you), and otherwise not really a matter of where you live. Though some schools still do better than others. Such as, just to name a few: Lowell, Oakland, Oakland Tech, and Ballard (the “McClatchy” of Fresno). I also found that, even when you control for other factors, API scores keep rising year after year.
Then I ran the regressions, but this time with the dependent variables being SAT score, whether Reading, Math, or Writing. The results were basically the same, but with some important footnotes. Parential educational attainment is, of course, the main predictor. And some metros are doing better than others — especially those in the valley: Bakersfield, Fresno, Sacramento, and Stockton (compared to Santa Rosa, the reference category).
But here’s the catch: reading continues to go up, year after year, once you control for other things, while math and writing do not.
Think about that, and think about this: parental educational attainment in California is going down. Everywhere.
The average number of parents with a BA or higher declines in most school districts by something like 0.5% per year. That may not sound like much, but think about how that builds. Over, say, a decade. For the metro areas included in this little equation, that means that the proportion has declined by something like 6%.
That’s real. That means parents reading to their kids. It means parents who know how to help their kids with their algebra. It means parents who know how to fill out a FAFSA. It means role models.
And that decline in parental education is, as far as I can tell, almost entirely a product of immigration.
And yet, and yet. API scores keep going up, and SAT scores say the same.
Maybe, just maybe, what NCLB hath wrought involves clinging to the edge a slope made ever-more slippery by the decline in parental educational attainment. Or maybe it’s utter cussing equine fo’shizzle, and has precisely zippo to do with the 3-card monte game that’s been sold to parents who send their kids to charter schools, etc.
Or maybe all it’s really accomplishing is the assurance that at least our children can fecking read, even if they can’t make 2+2 equal four, and even if they can’t write a comprehensible sentence.
You want my advice for the coming quarter century? Know your math, and know how to write. ‘Cause evidently you’ll be ahead of just about everybody else who’s entering the job market at the moment.
5 Reactie’s op “Revisiting my prejudices”
Geef een reactie:
Toegelaten Xhtml: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>
04 Mar 2010 om 9:49
Probably the demographic data for people taking the SAT has not changed as much as the same data for the population taking the tests in the API. So that while the SAT scores are flat, it may be that if all the students took the test the scores might actually go down to reflect the large immigrant population. The API, on the other hand, can rise because the lowest performers are doing a little better at a time, while the higher scorers (those more likely to take the SAT) are remaining flatter.
04 Mar 2010 om 9:50
Maybe. I haven’t look at this data.
04 Mar 2010 om 12:45
When I looked at API data–for a blog post I’m still intending to write–what stood out to me–and this was hardly shocking–is that, relative to one another, performance by individual schools was very roughly the same year-to-year, with variations that were likely in the noise. (The exception were first to second year scores of new charter schools, and one could expect that it would take a while for these to settle down.) The conclusion I drew was that some years the API test is easier than others, so year-to-year variation of a particular school isn’t really that valuable a measure of a school’s performance.
04 Mar 2010 om 12:51
One thing which would also be interesting–although it sounds like the data aren’t available–would be how the effect of parental educational attainment varies across ethnic groups. My experience in high school suggests that there are a significant number of Asian immigrants, whose parental educational attainment ranged from “raised in a culture without a written language” to “damn near nothing” who nevertheless did well academically, at least in math. But there is probably a huge selection bias in my sample of such examples.
11 Mar 2010 om 19:57
The SAT data are not broken down by ethnicity, much less by parental educational attainment, so it’s hard to know. Since the proportion of students taking the SAT has not really changed — but then again, we only have 3 years as opposed to 11 years of data — my guess is that the demographics are changing more slowly than the population covered by the STAR tests(though surely still some, given what’s happening to the school-age population more generally).
All of which is to say, there may be a reason why middle-class muckety-mucks like myself are not much impressed by NCLB: it doesn’t affect us, and it was never intended to.
RE: THM’s first point. Don’t look at year-to-year changes — look at the time series. Everyone is going up. I can send you the data, if you’d like.
RE: THM’s second point. The data that speak to this aren’t contained in the STAR reports, but there is PLENTY of research on this question. And, to radically simplify what is a pretty complicated picture: (a) there is a wide variation in the parental SES of Asian immigrants, especially, (b) there is less variation, overall, in the parental SES of immigrants from Mexico and Central America, (c) the relationship between SES and child outcomes is pretty strong, regardless of ethnic group (d) there appears to be a measurable positive effect associated with immigration, regardless of the country of origin (an effect, by the way, which disappears for most groups in the second generation), and (e) there was a almost certainly selection bias in your sample. Test this last with a simple question: what did the parents of the people you knew _do_ for a living? If you know, you know more than I: the few cases I can recall qualified as as petty-bourgeois, inasmuch as they owned/ran small business such as restaurants. That their children then did well in school is not surprising whatsoever. Test yourself with a second question as a follow-up: although it is true that there were kids from Lao/Mien/Hmong communities at our school, I didn’t actually know any — which is to say, they weren’t taking AP or otherwise advanced classes.