Social science versus common sense
Counter-intuitive results are usually counter-reality
Struggling to raise your kids right? Here’s a counter-intuitive scientific finding: really, what parents do doesn’t matter. Children come to resemble their parents, but that is all driven by their genes. Once you control for that, there are few systematic effects of upbringing.
That’s a big result and it has had a suitably big reception. Judith Rich Harris wrote an entire book on it in 1998 called The Nurture Assumption — a great example of clear, witty science writing, with a foreword by Steven Pinker. More recently Robert Plomin’s Blueprint pushed the idea:
In essence, the most important thing that parents give to their child is their genes. Many parents will find this hard to accept…. Parents differ in how much they guide their children in all aspects of development…. But in the population, these parenting differences don’t make much of a difference in their children’s outcomes….
And the argument has echoed through to more popular work.
Guess what, it’s wrong: parenting does matter. We know this because Ben Sacerdote studied Korean adoptees to the US. These were assigned to the first parents they were matched with, creating a nice natural experiment, and they share no more genes with their adoptive parents than with anyone else. From the paper:
Children assigned to the high education, small families are twice as likely to graduate from a college ranked by US News & World Report, have an additional .75 years of education, and are 16 percent more likely to complete four years of college
— a big difference. He also found that parenting affected adoptive children’s drinking behaviour, and the quality of college they attended. These basic results have been repeated in other studies, most recently by Jonathan Beauchamp.
Disheartened, you put down Fruit Ninja and listen to the faint cries behind the locked nursery door. Better pay the little buggers some attention after all. Still, you think, they are a blessing — until you recall the shocking finding that children make parents less happy:
… people tend to believe that parenthood is central to a meaningful and fulfilling life, and that the lives of childless people are emptier, less rewarding, and lonelier, than the lives of parents. Most cross-sectional and longitudinal evidence suggest, however, that people are better off without having children.
That’s not a one-off study: it’s a literature review, summarizing numerous studies. That children made parents unhappy was, if not a consensus, at least a very widespread finding, until around 2012.
Guess what again? It’s not that simple. It turns out a key variable is finances. If you can’t afford a child, it makes you less happy. Marital status, too:
Kids do not raise happiness for singles, the divorced, separated or widowed.
Perhaps for the obvious reason that raising a child on your own is hard. It also matters whether the children are your own or not. If you’re married and can pay the bills, raising your own children will probably make you happier, which I guess is what most people thought. Again, when we read the fine print, the big counter-intuitive result disappears in a flood of exceptions.
I am now prepared to bet on this as a general rule: when social science produces a result which challenges common sense, science is wrong and common sense is right.
This rule doesn’t just apply to one-off surprises. It still works for widely replicated, agreed-on results. Remember Kahneman on priming effects: “disbelief is not an option”. Not all those priming effects survived the replication crisis very well.
How can this happen? Indeed, one reason comes from the usual bestiary of the replication crisis: replicability, file-drawer effects, researcher degrees of freedom and all that malarkey, meaning that dodgy results survive longer than they should.
Another answer might be about hedgehogs versus foxes (the hedgehog, remember, knows one big thing, whereas the fox knows many things). “Family environment doesn’t matter” came from behavioural genetics, and specifically from classical twin studies, which use identical and non-identical twins to disentangle genes and environment. The evidence that it does matter comes from different designs — adoption studies with careful attention to randomization of the adoptees. Twin studies are great and have produced a ton of valuable knowledge, but they aren’t perfect. Too great a focus on a single tool may blind a subfield to results produced using different methods.
A subtler effect is when subfields choose the research areas that work for them, and then mistake their parish for the world. Perhaps the things behavioural geneticists chose to study tended to have large genetic effects. That would be natural: “genes don’t matter much” isn’t a very rewarding top line for a geneticist. And maybe surprisingly large genetic effects go along with surprisingly small family environment effects. Then, different researchers with different priorities chose different dependent variables and found different results. By the way, this is another reason it’s good to have economists mucking about in your field.
There must be counter-intuitive results that do survive contact with reality. Of course, physics is full of them: our “folk physics” is just wrong about a bunch of stuff. Are there places where “folk sociology” is equally off? Macroeconomics might have some. It’s not obvious that we can make ourselves poorer by saving too much: our intuitions about individual behaviour don’t transfer well to the collective level.
Conversely there are surely some counter-intuitive results that haven’t yet been refuted, but will be. Here’s one I’d put money on: the bad effects of corporal punishment. This has been claimed for so long that it probably isn’t counter-intuitive any more to most Westerners. But most people in the world used corporal punishment on children for most of history. A simple evolutionary argument suggests that behaviours that seriously harm one’s own offspring won’t become widespread. And the quality of the “evidence” on this topic is almost unbelievably bad. It’s a very difficult research question, beset by problems of reverse causality, conceptual boundaries, selection bias and what have you. Yet for years, child development researchers basically ran correlations between corporal punishment and bad behaviour, and effectively assumed causality ran the opposite way to what common sense suggest.1
I mean, don’t beat your kid because I said so. Just take the scientists with a pinch of salt.
More generally, some humility is in order. Humans know a lot about human society. Only a minuscule proportion of that knowledge is scientific. As Donald Davidson would tell you2, belief is veridical: if people think stuff, on average, it’s probably true. If social scientific knowledge vanished from the world overnight, we might have some difficulty setting interest rates. If informal knowledge vanished, we’d struggle to do the shopping (“what’s a shop?”) The best social science builds on what’s obvious to get a tiny bit further than we could otherwise have got.
Coincidentally, all my examples have involved parenting. There’s a long history of “experts” giving self-confident advice to parents, especially mothers:
whatever procedures were held to be proper science at the time, were given inordinate weight against poor old defenceless folk knowledge.3
In the 1930s, the vogue was to avoid too much affection:
Never hug and kiss them, never let them sit on your lap. If you must, kiss them once on the forehead when they say good night. Shake hands with them in the morning.4
This from John B. Watson, an early behaviourist, both of whose sons later attempted suicide, one successfully.5 And women followed this advice or felt guilty if they didn’t!
I give him too much affection. I yearn over him…. But I am learning to suppress my emotions.6
Beware of the one-sided expert!
If you liked this content, then I would love you to do three things:
Subscribe to this newsletter. It’s free, posts are occasional, and subscribers make me happy.
Share Wyclif’s Dust on social media. This newsletter is a new venture, so by telling your friends and/or followers, you’ll be doing me a huge favour.
Read about the book I’m writing. It’s called Wyclif’s Dust, too. You can download a sample chapter.
Davidson, D. (1986) “A Coherence Theory of Truth” in Pore, ed., Epistemology: an anthology pp. 124-133.
Kessen, W. (1979) “The American child and other cultural inventions”, American Psychologist, 34(10): 815–820.
Watson (1928) Psychological Care of Infant and Child, quoted in Hays (1996) The Cultural Contradictions of Motherhood.
Duke, C., Fried, S., Pliley, W., & Walker, D. (1989). “Contributions to the history of psychology: LIX. Rosalie Rayner Watson: The mother of a behaviorist's sons.” Psychological Reports, 65(1): 163-169.
Settlement mother, quoted in Grant, J. (1998). Raising baby by the book: The education of American mothers. Yale University Press.
I'll take you up on that wager. Not that I think (mild) corporal punishment is necessarily bad, just that all punishment is misdirected. Punishment suppresses behaviour. Reinforcement strengthens behaviour. When you're suppressing, you're often likely to suppress the wrong behaviour. The classic example is if you punish a child for not doing their homework, they'll try to avoid coming to class. They might even develop more severe behavioural problems so you stop minding the homework. Reinforcement is targetted at a specific behaviour, and is therefore more effective.