DON'T DO THINGS HALF WAY
JUDITH RICH HARRIS: Well, I'm prone to making statements like this one: How the parents rear the child has no long-term effects on the child's personality, intelligence, or mental health. I guess you could call that an extreme statement. But I prefer to think of myself as a defender of the null hypothesis.
The null hypothesis is the hypothesis that a putative "cause" has no effect, and it's supposed to be the starting point for scientific inquiry. For instance, when a new drug is being tested, the researchers are expected to start out with the hypothesis that the drug is no better than a placebo. If they find that the patients who received the drug are more likely to recover than the ones who got the placebo, then they can reject the null hypothesis at some level of confidence, some probability level.
This comes as a surprise to most people, but psychologists have still not managed to collect evidence of a sort that would enable them to reject the null hypothesis of zero parental influence. In the absence of such evidence, the only scientifically sound position is the one I've taken.
JB: But you can't prove that parents have zero influence, can you?
HARRIS: No. But why should I have to? Shouldn't the burden of proof be on the people who claim that parents do have an effect? Shouldn't the burden of proof be on the people who claim that their treatment the style of child-rearing they recommend works the way they say it will?
The problem is that the null hypothesis of zero parental influence hasn't been treated as the starting point, due to a pervasive pattern of thought I call "the nurture assumption." To say, "We will assume that parents do have an effect until you prove that they don't" is a demonstration of the nurture assumption in action. It's an a priori bias based on faith rather than evidence.
JB: But some people are saying that there is evidence plenty of evidence of parental influence, and that for some reason you have chosen to ignore it.
HARRIS: On the contrary, I've haven't ignored it at all. I've looked at it closely and found it to be worthless. The evidence they're talking about is hopelessly confounded and is contradicted by other evidence.
Let me give you an example. Many studies have shown that verbal, literate parents tend to have verbal, literate kids. Parents who talk to their kids a lot tend to have kids with an above-average vocabulary. Parents who read to their kids tend to have kids who become good readers. This evidence comes packaged with a moral: If you want your kid to get into Harvard, you'd better start drilling him 18 years before his application is due.
The trouble is, the evidence is ambiguous. It's clear that children resemble their biological parents; what isn't clear is why. Is it the environment the parents provided, or is it the genes they provided? Just knowing there's a correlation isn't enough we have to tease apart the effects of the genes from the effects of the home environment. One way to do it is by looking at adopted kids. And what we find is that the correlation disappears. The adopted child reared in a let's-read-a book-together home ends up no smarter, on the average, than the one reared in a don't-bother-me-I'm-watching-TV home. As far as Harvard is concerned, it doesn't make a dime's worth of difference whether the kid grew up listening to Mozart or Muzak.
In general, studies that provide a way of controlling for the effects of the genes by looking at twins, siblings, or adopted children show that the home environment has little or no effect on intelligence or personality. They show that the similarities between parents and their biological children, or between two biological siblings reared in the same home, are almost entirely a function of their shared genes. Eliminate the effect of the shared genes and you've eliminated all, or nearly all, of the similarity.
JB: You mean all that matters is having the right genes?
HARRIS: No, it's not just genes. The environment definitely matters too. In fact, for personality (which is what I'm mainly interested in), only about half the variation from one person to another can be attributed to the genes. More precisely, about half the reliable variance in measured personality characteristics the variance that remains after measurement error is subtracted can be attributed to differences in genes.