Two Myths About the SAT
Since the recent announcement that major changes are coming to the SAT, I've seen about 10,000 articles claiming that (A) the correlation between family income and SAT scores proves that the test is biased against poor kids and (B) the SAT is easy for the rich to "game" through test prep.
There are indeed problems with the SAT, including problems relating to parental income. Here is a persuasive argument by Charles Murray that the SAT should be scrapped entirely and replaced by subject tests like the SAT II. But we shouldn't be shocked by the simple fact that parental income is linked to children's academic achievement, and the effects of "gaming" the SAT have been wildly exaggerated.
The problem with treating the SAT-income correlation as an argument in itself is that richer kids, fairly or unfairly, actually do have higher academic capabilities. As I've written before, and frankly as anyone with eyes in his head ought to know, there are numerous ways that parents can pass advantages on to their kids -- genes, better schools, better neighborhoods, and so on. The simple correlation between income and scores tells us nothing at all about whether the SAT has an income bias, because it's exactly what we would expect from a legitimate measure of academic ability.
Income gaps are evident on basically every academic measure we have. Here is an informative paper bringing together all sorts of achievement-test results in regards to the 90/10 gap -- the gap between kids at the 90th and 10th percentiles of parental income.
It provides two important charts, with Y axes indicating the gap in standard deviations. Here's the gap in reading:
And here's the gap in math:
As you can see, across plenty of different tests, in recent years the 90/10 gap has usually been a full standard deviation or more. Measuring the gap with one test instead of another makes little difference.
[Update: Also, the ACT and SAT show very similar income trends, as I show here.]
And what about "gaming" the SAT? Well, here is a report finding that SAT prep typically raises math scores 10-20 points and reading scores 5-10 points. The standard deviation of each subtest is around 110 points, and the black/white gap is about the same. What's more, free test-prep services are targeted toward poor and minority kids -- surprisingly, black and Hispanic kids are actually more likely to use test prep, and they gain slightly more than whites from doing so (though Asians seem to gain unusually large amounts, as much as 70 points).
I'm glad to see a discussion about the merits of the SAT. But let's drop these two unhelpful talking points.
Robert VerBruggen is editor of RealClearPolicy. Twitter: @RAVerBruggen