New Data! Judging Fairly How Well States Are Doing At Educating Their Children
November 15, 2007, 05:05 AM
A+
|
a-
Print Friendly and PDF
One of the reasons that so much bilge is spewed about public school performance is that almost all objective data on educational achievement is overwhelmingly dominated by crude racial demographics: the more Non-Asian Minorities (NAMs), the lousier the test scores.

The facts are obvious, boring, and depressing, so people make up silliness to have something to say. For example, here are two smart guys trying to wink out in Morse code the real reason for differences in state test scores without getting Watsoned. As George Will coyly hinted in his obituary for Daniel Patrick Moynihan:

"The Senate's Sisyphus, Moynihan was forever pushing uphill a boulder of inconvenient data. A social scientist trained to distinguish correlation from causation, and a wit, Moynihan puckishly said that a crucial determinant of the quality of American schools is proximity to the Canadian border. The barb in his jest was this: High cognitive outputs correlate not with high per-pupil expenditures but with a high percentage of two-parent families. For that, there was the rough geographical correlation that caused Moynihan to suggest that states trying to improve their students' test scores should move closer to Canada.

S-u-r-e, Dan and George! It must be proximity to Canada!! Everybody knows that playing hockey makes people monogamous and smart!!!

One big problem with this is that we therefore don't have any fair ways to judge the performance of school administrations relative to what they have to work with. Take a look at SAT scores for public schools in Los Angeles County, the nation's largest county. People hear that at San Marino H.S. near Pasadena, 89% of these public school students get over 1000 on the SAT, while at Compton H.S. south of Los Angeles, only 1% crack the 1000 barrier, so they start assuming that San Marino's administration is doing a great job and Compton's a terrible job. Then they come up with brilliant plans like Compton should imitate San Marino and offer fewer remedial courses and more AP courses.

Is this correct? Maybe, but maybe not. After all, San Marino is 70% Asian, and not just any Asians—it's where Hong Kong zillionaires stash their families in case the Communist hammer comes down too hard on Hong Kong. Compton, in contrast, is all NAMs.

However, if you are willing to work hard enough with the data, you actually can start to remove the overwhelming influence of race and start to get at measuring the schools' value added.

Audacious Epigone takes the best crack I've seen yet at measuring how good a job public schools are doing in each state, by ranking the states on the change in the federal government's National Assessment of Educational Progress test scores for 4th graders in 2003 versus 8th graders in 2007. (The idea is to look at roughly the same children over four years of schooling. Of course, they aren't exactly the same, and in states undergoing rapid demographic change, such as North Carolina, this could affect the numbers.)

His list is interesting and he may be getting pretty close to measuring the actual performance of the public schools:

The five states that did best at improving test scores from 4th to 8th grade from 2003-2007 are Washington D.C. (even though the public school system there is ferociously criticized), Massachusetts (traditionally, a top performer), North Dakota, Montana, and Maryland.

The bottom five are (from bottom up) West Virginia, North Carolina, Connecticut, Michigan, and New Hampshire.

It's fun to see, for once, a ranking of school performance that's not obviously dominated by demographics. Off the top of my head, I can't see what the states at the top have in a common with each other, nor what the states at the bottom share. So, we may actually be finally getting at institutional effectiveness.

Still, demographic change could be biasing this list. So, I'd like to see this same list, but for each race. I suspect that North Carolina's poor performance is driven by heavy Hispanic immigration in this decade, but looking at just white performance could show I'm wrong.

Washington D.C.'s strong performance might be driven by the on-going driving out of African-Americans by immigrants and by white gentrifiers.

It's interesting to look at the two superstates with contrasting reputations for educational effectiveness: Texas, with its positive stereotype, does a little better than average, but still only ranks in 23rd place. California, with its negative stereotype (especially the LAUSD) does a little worse than average, but still only ranks in 36th place. So, the stereotypes are supported, but not dramatically so.

Here's the rest of Audacious Epigone's list.