The Obama Administration's Lone Star Mistake
The Obama administration made what was supposed to be a pre-emptive strike on Texas Governor Rick Perry, but which turned out to be, in baseball terms, a “swing … and a miss.” The administration unleashed Education Secretary Arne Duncan to attack Texas’s record on education, with Duncan saying he feels “very, very badly for the children.” When pressed to explain precisely what Texas has done wrong on education, Duncan came up a bit short on specifics. The Education Secretary’s arguments have generated a lot of useful discussion across the web, but I thought I would throw some rudimentary data analysis into the picture.
To start, if you want to get an accurate picture of the quality of a school system, you ideally want to measure inputs and outputs: what does a school system generate given what it starts with? One input is money, which we’ll look at. But another, more important, input is the kids entering the system. Put simply, Texas schools start with many kids from poor or single parent homes, or for whom English isn’t a first language. You wouldn’t, after all, expect schools in inner city Chicago—those once overseen by Secretary Duncan—to outperform schools in suburban Lake Forest. Controlling by race isn’t a perfect way to handle this, but to ignore race is to say that schools bear full responsibility for the performance of kids who come from extremely challenging backgrounds, which is silly.
If you look at Texas’s simple average test scores in reading and math for fourth and eighth grade students, they’re about average—not great, but not terrible. Why Duncan would pick on a state that’s in the middle of the pack rather than a true laggard is anybody’s guess. But even then, the comparison is bogus simply because Texas schools serve a population with several challenges, in particular many low-income and Spanish speaking children.
To account for this, I tabulated average test scores by race for each state using data from the National Center for Education Statistics, then weighted the scores using Census data on the racial composition of each state. I then calculated Texas’s standing in the distribution. A percentile figure indicates the percent of states whose scores Texas beats; for instance, Texas’s weighted average test score in fourth grade reading is in the 62nd percentile, meaning that Texas students score better than 62 percent of students nationwide. (For those of you who didn’t attend Texas schools, that means that Texas did pretty well.)
This exercise is really nothing more than what the genius blogger Iowahawk wrote up a few months ago. (Is Iowahawk a genius because he understands weighted averages? No, he’s a genius because he writes stuff like this.) What the data shows is that, once you take the racial composition of the states into account, Texas does a lot better than Duncan gave it credit for. In eighth grade math Texas is at the 88th percentile, meaning it beats almost nine out of ten states. Overall, Texas is at the 71st percentile, which isn’t shabby at all.
Texas looks particularly good when you consider how much it spends on education. The next chart shows the weighted average fourth and eighth grade reading and math scores against state spending per pupil. The idea is to get some feel for the “bank for the buck” generated by the education systems of different states. The diagonal regression line shows the average relationship between spending and test scores—for each extra thousand dollars of per pupil spending, average test scores rise by around 1.1 points. The chart and regression exclude two huge outliers, Hawaii and Washington, D.C., which have very high per pupil spending but generate low test scores even after adjusting for the racial composition of their populations. (Hawaii may fall into this group because of its large population of Pacific Islanders, who are lumped together with generally high-performing Asian students. D.C. is, well, D.C.)
The key to this chart is that states to the upper left of the regression line seem to generate a better bang for the buck—they produce higher scores at lower costs than the average state. Texas, shown in red, is in this group. Based on its level of spending, Texas’s average test score “should” be around 246, the 36th percentile, but in reality it’s 253, the 71st percentile. And while Texas’s test scores are in the 71st percentile nationwide, its per pupil spending is in the 26th percentile—meaning that it performs better than roughly three-quarters of states while spending less than roughly three-quarters of states. Put another way, Texas’s education system generates scores that the average state would spend an extra $6,350 per pupil to produce.
Finally, let’s look at high school graduation rates. After all, Duncan said that “Far too few of [Texas] high school graduates are actually prepared to go on to college,” for which I presume that graduating high school is one important prerequisite. Across all racial groups, Texas graduates a greater percentage of its high school freshmen than the U.S. average. In one case, Asians, Texas’s graduation rate is well above average—99 percent versus 91 percent nationally. And in the case of Native Americans, Texas’s graduation rate is massively higher—85 percent versus only 61 percent nationwide.
My takeaway isn’t that Texas is unlucky to have Rick Perry overseeing their schools. Rather, it’s that Texas is unlucky to be situated next to Mexico rather than Canada. If Texas had the population of, say, Maine, its average test scores and graduation rates would be considered outstanding. (In fact, Texas’s weighted average test scores are well above those of Maine.) Given the huge challenges of educating children who come from single-parent homes, or who aren’t fluent in English, or who simply grew up poor, Texas isn’t doing that bad a job. In fact, Texas seems to outperform in most areas, and to do it at a significantly lower cost per pupil than many other states.
Andrew G. Biggs is a resident scholar at the American Enterprise Institute.
No comments:
Post a Comment