The brilliant Charles Murray is finally wrong about something.
Iconic blogger Steve Sailer recently blogged about the following comments by Murray:
The results are always the same: The richer the parents, the higher the children’s SAT scores. This has led some to view the SAT as merely another weapon in the inequality wars, and to suggest that SAT should actually stand for “Student Affluence Test.”
It’s a bum rap. All high-quality academic tests look as if they’re affluence tests. It’s inevitable. Parental IQ is correlated with children’s IQ everywhere. In all advanced societies, income is correlated with IQ. Scores on academic achievement tests are always correlated with the test-takers’ IQ. Those three correlations guarantee that every standardized academic-achievement test shows higher average test scores as parental income increases.
What Murray is saying, obviously, is that rich kids do well on the SAT because they have high IQ genes and the rich environment they were raised in has little effect. I’d hate to throw Charles Murray under the bus, because he along with Steve Sailer really paved the way for people like me to talk openly about behavioral genetics, but in my scientific judgement, he’s wrong.
On the combined verbal and math section of the new SAT, kids from homes earning over $200 K a year average 1157 while kids from homes earning less than 20 K a year average 895. On a scale where the U.S. mean is set at 100 (SD = 15) these scores are equivalent to IQs of 118 and 100 respectively (note only the most academic third of America tends to take the SAT so scores are high). That’s a difference of 1.2 standard deviations in kids coming from homes that are 3.07 standard deviations apart in normalized income. 1.2/3.07 = 0.39, suggesting the correlation between SAT scores and the income of your parents is about 0.4.
But even 0.4 is an underestimate, because among rich kids, taking the SAT is very common, but among poor kids with little hope of affording college, it’s largely the best and brightest who take the SAT, so if everyone took the SAT, the correlation between SAT scores and your parents’ income would be well above 0.4.
But 0.4 is widely cited as the correlation between IQ and one’s own income (see Jensen, 1998). One would expect IQ to correlate much less with your parents’ income, because a teenager’s IQ correlates no more than 0.6 with the mid-parents’ IQ; thus the correlation should be 0.4(0.6) = 0.24. Instead, if the SAT is used instead of an official IQ test, it’s almost double that.
Clearly, the SAT is culturally biased in favor of kids from rich homes (and probably educated homes) and biased against kids from poor (and probably uneducated homes). And this is not the first time we’ve seen this effect. In the Minnesota Transracial Adoption Study, black kids adopted into white professional homes scored the equivalent of IQ 95 on measures of scholastic aptitude and achievement, but scored an IQ of only 84 (about the same as black Americans raised by their biological parents) on the WAIS (an official IQ test).
What this suggests is that the SAT probably underestimates the ability of kids from low SES homes, and overestimates the ability of kids from high SES homes. I’ve even noticed this with celebrities. Bill Cosby, Howard Stern, and Rosie O’Donnell have all come from low SES homes and have all claimed to have scored poorly on the SAT, but all strike me as highly intelligent. By contrast George W. Bush did well on the SAT and comes from an extremely high SES background, but does not strike me as intelligent. The SAT would probably better reflect IQ if scores were corrected for SES, but one must be careful not to over-correct, because over half the test’s correlation with such measures is probably genetic.