It seems to be a fundamental assumption of modern economics that while correlation by itself doesn’t equal causation, correlation plus “controls” does. Thus we see this from Thomas Friedman in the New York Times:
Parents more focused on their children’s education can also make a huge difference in a student’s achievement.
How do we know? . . . Andreas Schleicher, who oversees the [PISA] exams for the O.E.C.D., was encouraged by the O.E.C.D. countries to look beyond the classrooms. . . . The PISA team went to the parents of 5,000 students and interviewed them “about how they raised their kids and then compared that with the test results,” [Schleicher said.] . . . Two weeks ago, the PISA team published the three main findings of its study:
“Fifteen-year-old students whose parents often read books with them during their first year of primary school show markedly higher scores in PISA 2009 than students whose parents read with them infrequently or not at all. The performance advantage among students whose parents read to them in their early school years is evident regardless of the family’s socioeconomic background. Parents’ engagement with their 15-year-olds is strongly associated with better performance in PISA.”
Schleicher explained to me that “just asking your child how was their school day and showing genuine interest in the learning that they are doing can have the same impact as hours of private tutoring. It is something every parent can do, no matter what their education level or social background.”
Say that you take 100 poor people who read to their kids and 100 similarly poor people who don’t, and find that the former group’s kids score better on tests. Hopefully, you’d at least consider the obvious explanation: While both groups of parents are comparably poor, the former parents are probably more intelligent, conscientious, and bookish on balance, and may have passed those traits on to their children in various ways — genes, overall home environment, etc. Instead, the PISA researcher and a widely read NYT columnist simply assume that parents’ reading to the kids caused the higher test scores all by itself. Hey, they “controlled” for a variable, and probably used some fancy regression analysis to do so, so they proved causation, right? The researcher makes the whole thing even more ridiculous by claiming explicitly that asking your kid about their school day also causes higher test scores.
It’s certainly possible that these parental behaviors contribute to higher test scores in some way — when I have kids, I’m certainly going to read to them and hound them about their homework — but this research does not even come close to proving it, and the stronger evidence we have isn’t encouraging. This piece from Bryan Caplan offers a good overview of adoption research, which is the closest we can come to actually separating the effects of genes and environment. Kids tend to be much more like their biological parents than like their adoptive ones, and adoptive parents seem to have no effect on some traits at all.