Why Parenting May Not Matter and Why Most Social Science Research is Probably Wrong
Parenting effects usually play some role in our conception of why some people behave differently than others
I want you to consider the possibility that your parents did not shape you as a person. Despite how it feels, your mother and father (or whoever raised you) likely imprinted almost nothing on your personality that has persisted into adulthood. Pause for a minute and let that heresy wash across your synapses. It flies in the face of common sense, does it not? In fact, it’s the type of claim that is unwise to make unless you have some compelling evidence to back it up. Even then it will elicit the ire of many. Psychologists especially get touchy about this subject. I do have evidence, though, and by the time we’ve strolled through the menagerie of reasons to doubt parenting effects, I think another point will also become evident: the problems with parenting research are just a symptom of a larger malady plaguing the social and health sciences. A malady that needs to be dealt with.
In terms of compelling evidence, let’s start with a study published recently in the prestigious journal Nature Genetics.1 Tinca Polderman and colleagues just completed the Herculean task of reviewing nearly all twin studies published by behavior geneticists over the past 50 years. For some background, behavior genetics is the field devoted to studying human differences, and let’s be honest, whether you are a scientist or not you are interested in why people are different from one another. Besides being inherently fascinating, the reality of those differences impacts your life daily. The knowledge that some people are more trustworthy, honest, violent, impulsive, and aggressive than others is essential to navigating life. It’s simply not a good personal policy to assume that everyone you stumble upon in life has your best interest at heart.
It doesn’t matter if you’re a behavioral scientist or a plumber; we’re all theorists about these differences. People speculate about human variability in their free time constantly (think about how often you’ve wondered why your boss is such a huge…source of inspiration). Parenting effects usually play some role in our conception of why some people behave differently than others. Behavior genetics, luckily, provides us with meaningful insight regarding the sources of human differences in the population (unfortunately I can’t say anything about your boss specifically). So what about the results of that massive review of twin research? Genetic factors were consistently relevant, differentiating humans on a range of health and psychological outcomes (in technical parlance, human differences are heritable). The environment, not surprisingly, was also clearly and convincingly implicated, but interestingly it wasn’t the “environment” you might have anticipated.
Before progressing, I should note that behavioral geneticists make a finer grain distinction than most about the environment, subdividing it into shared and non-shared components.1,2,3,4 Not much is really complicated about this. The shared environment makes children raised together similar to each other.3 The term encompasses the typical parenting effects that we normally envision when we think about environmental variables. Non-shared influences capture the unique experiences of siblings raised in the same home; they make siblings different from one another. Another way of thinking about non-shared environments is that they represent the parts of your life story that are unique from the rest of your family. Importantly, this also includes all of the randomness and pure happenstance that life tends to hurl in our direction from time to time. Returning to the review of twin research, the shared environment just didn’t matter all that much (that’s on average, of course, for some traits it mattered more than others). The non-shared environment mattered consistently.
The pattern of findings mentioned above is nothing new.1,2,3,4,5 The importance of genetics and the non-shared environment (and the relatively minor importance of the shared environment) was already so entrenched in behavior genetics that years before the Polderman study was published it had been enshrined as a set of “laws.”2 The BG laws, though, are based largely (but certainly not completely) on twin studies, the meta-analysis by Polderman et al. was comprised of twin studies, and if you pay attention to this sort of thing you’ve probably heard some nasty things about twin studies lately.3 You’ve read that twin studies contain an insidious flaw that causes them to underestimate shared environmental effects (making it seem like parents matter less than they do). The assumptions of twin research, however, have been meticulously studied. The methods of twin researchers have been around for decades and have been challenged, critiqued, refined, adjusted, and (perhaps most importantly) cross validated with other techniques that rely on different assumptions entirely.3,4 They work, and they work with impressive precision.
Based on the results of classical twin studies, it just doesn’t appear that parenting—whether mom and dad are permissive or not, read to their kid or not, or whatever else—impacts development as much as we might like to think. Regarding the cross-validation that I mentioned, studies examining identical twins separated at birth and reared apart have repeatedly revealed (in shocking ways) the same thing: these individuals are remarkably similar when in fact they should be utterly different (they have completely different environments, but the same genes).3 Alternatively, non-biologically related adopted children (who have no genetic commonalities) raised together are utterly dissimilar to each other—despite in many cases having decades of exposure to the same parents and home environments.3
David C. Geary
One logical explanation for this is a lack of parenting influence for psychological development. Judith Rich Harris made this point forcefully in her book The Nurture Assumption (an absolute must read). 6 As Harris notes, parents are not to blame for their children’s neuroses (beyond the genes they contribute to the manufacturing of that child), nor can they take much credit for their successful psychological adjustment. To put a finer point on what Harris argued, children do not transport the effects of parenting (whatever they might be) outside the home. The socialization of children certainly matters (remember, neither personality nor temperament is 100 percent heritable), but it is not the parents who are the primary “socializers”, that honor goes to the child’s peer group (a fascinating topic, but one that merits its own separate discussion).
Now, the astute critic will respond with their own research in hand, papers centering on the deleterious impact of child abuse and severe neglect. There is a wealth of evidence linking child abuse with all sorts of developmental delays, and Harris fully acknowledges this. Mercifully, child abuse is not pervasive in the population, meaning that most kids don’t experience it and it is unlikely that it explains large swaths of why some kids are more extroverted or intelligent than others.6 That said, consider an analogy shared with me by the psychologist Steven Pinker: dropping your iPhone from six floors up is guaranteed to ruin it—iPhones don’t bounce. The impending destruction awaiting your phone as it plummets toward the Earth is assured, and the fact that you played no part in designing or building your phone will not atone for your slippery fingers. The same analogy applies to parenting, in some respects. It is possible for parents to wreck something that they did not construct (i.e., their child’s healthy development, language growth, cognitive ability, etc.) if their parenting style is harsh enough. Hopefully it is evident that this type of “parenting” is not the topic at hand.6
So why mount a frontal assault on parenting? I love my parents deeply, so it has nothing to do with some latent Freudian bitterness traceable back to the first few years of my life. The lack of parenting effects happens to represent an effective avenue for making a larger point about most social science research. It cannot be fully trusted. Brian Nosek’s incredible work on reproducibility in psychology (along with a cadre of collaborators) makes this point in one very powerful respect.7 However, I want to look at the issue from a slightly different vantage point.
The vast majority of research in the social sciences involves non-experimental observational research. What this means is that researchers collect data on individuals. The data collection might start early, perhaps shortly after birth.3 The researchers will usually want to know something about the parenting strategies used with the child (it’s usually the entire point of the study), and they will probably collect data on a host of other topics. All of the information is compiled into a large database that researchers can then analyze. With age, researchers can ask children directly about their behavior, while also collecting personality inventories and any other bit of information that they deem important. In some cases, researchers have collected data for so many years that they can use the personality traits of children to predict their behaviors decades later. This is fantastic but the problem is that you just can’t fully interpret what it means. It doesn’t matter if you collect data from a kid every year from the day they are born to the day they die, if you only gather data from one child per family you cannot pull out the genetic effects that we know are there (I would be remiss not to mention that newer techniques permit estimating genetic influences in the absence of twin subjects, using genetic data from very large samples of unrelated individuals).3 It really doesn’t matter if you show that maternal withdrawal experienced at age 3 predicts poor psychological adjustment when the child is 50, it would be impossible to make a concrete interpretation about the finding.
You must remember that parents share genes with their children and that overlap must be accounted for in research design. As psychologists pointed out years ago5, because parents pass along two things to their kids: genes and an environment, it shocks virtually no one that the two would be correlated. It is not surprising, based on shared genetics, that children resemble their parents, not only in appearance, but also in temperament, behavior, intellect, athletic prowess, etc. The environments that parents construct for their children when they are young, moreover, tend to mirror their natural inclinations (bright parents provide enriched environments). So if you’re wondering whether parents might selectively foster certain preexisting skill sets (i.e., buying an instrument for a child interested in music) the answer is, sure.6 In that case, parents might also shape things further by deciding on the type of instrument (guitar over drums, etc.).6 However, when you introduce controls for that genetic overlap in studies probing the impact of parenting on some outcome more generally, the effects that we often see can vanish.
I’ll give you a concrete example to mull over. Children who are spanked (not abused, but spanked) often experience a host of other problems in life, including psychological maladjustment and behavioral problems.8 In a study led by my colleague J.C. Barnes, we probed this issue in more detail and found some evidence suggesting that spanking increased the occurrence of overt bad behavior in children.8 We could have stopped there. Yet, we went one step further and attempted to inspect the genetic influences that were rampant across the measures included in our study. What we found was that much of the association between the two variables (spanking and behavior) was attributable to genetic effects that they had in common. The correlation between spanking and behavior appeared to reflect the presence of shared genetic influences cutting across both traits.
What about studies not directly devoted to examining parenting effects? Let’s say you’re interested in whether one trait in an individual predicts some other trait in that individual. For instance, let’s assume you think that novelty seeking predicts a higher likelihood of experimenting with drugs, or that eating fatty foods increases your body mass index (BMI). Simple enough: measure novelty seeking and measure drug experimentation; or eating habits and BMI. Then see if the two correlate. Let’s say that they do. Fantastic, write it up, publish it, get famous. But there’s a problem, what if there are genetic effects on novelty seeking and drug experimentation (or eating habits and BMI)? Even more troublesome, what if some of the same genes that predict novelty seeking also predict drug use, and the same genes that predict eating habits, also predict BMI? This means that the traits are correlated at the genetic level, just like in our study that I described above.
Brian Boutwell
My colleagues and I have tried to quantify just how irksome genetic correlations might be for studies that cannot account for them.9 What we found is that in some cases a genetic correlation can render a phenotypic correlation (the correlation between two outwardly observable traits) nonexistent. For some studies it might “look like” personality trait A correlates with behavior B, but it could simply be that the same genes influence both traits, thus explaining why the two outcomes are correlated to begin with. So, why should you care? Could this just be overly technical, nerdish handwringing? Not even close. This matters because most of the social science research that rockets into the headlines, grabbing your attention when you fire up the web, is likely wrong.
Whether it’s a study purporting to link some aspect of parenting to child development, or a study intended to link some new diet fad to weight loss, the results are unclear if they did not control for genetics. Lest someone put words into my mouth later, this does not mean that every correlation reported by social scientists is the result of correlated genetic influences. The point, however, is that we have spent decades churning out correlations and we have no idea whether the findings were polluted by unmeasured genetic factors. That’s frightening, especially since public policies have been built on some of these potentially illusory correlations. The standard way of doing business in the social sciences ignores genetic influences, and has for years. Be careful which findings you cling to. Most social science research can only reveal associations; which is important, no doubt, but I presume you want to know something about causality also (i.e., if you eat bacon everyday what’s the chance that it’ll cause you to get cancer; that sort of thing). To even begin approximating causality (assuming you cannot do an experiment, which you can’t with most social science research), you must account for all confounding factors—genes included.
Let’s return then to the overarching theme of our discussion, parenting. Is it possible that parents really do shape children in deep and meaningful ways? Sure it is. In line with the phrase often trotted out by my ilk: “it’s an empirical question.” The trouble is that most research on parenting will not help you in the slightest because it doesn’t control for genetic factors. What we do know (largely from twin studies) is that beyond the genes they contribute, parents are not responsible for autism (or schizophrenia, or ADHD, etc.), and they likely bear zero responsibility for injecting general intelligence or a personality into the heads of their children. So, why the dogmatic adherence to the idea that parents are the “puppet masters” in our lives? The are many reasons, some of which are explicitly religious (the whole “spare the rod spoil the child” bit) and some are more secular, rooted in dubious research, but we should nevertheless let them all go.
Natural selection has wired into us a sense of attachment for our offspring. There is no need to graft on beliefs about “the power of parenting” in order to justify our instinct that being a good parent is important. Consider this: what if parenting really doesn’t matter? Then what? The evidence for pervasive parenting effects, after all, looks like a foundation of sand likely to slide out from under us at any second. If your moral constitution requires that you exert god-like control over your kid’s psychological development in order to treat them with the dignity afforded any other human being, then perhaps it is time to recalibrate your moral compass; does it actually point north or just spin like a washing machine (see Pinker’s work for this same point made more eloquently10)? If you want happy children, and you desire a relationship with them that lasts beyond when they’re old enough to fly the nest, then be good to your kids.10 Just know that it probably will have little effect on the person they will grow into. I think it’s fitting to let Judith Rich Harris6 have the last word. Here is a short poem from The Nurture Assumption:
How sharper than a serpent’s tooth
To hear your child make such a fuss.
It isn’t fair—it’s not the truth—
He’s fucked up, yes, but not by us.
References
1. Polderman, T. J., Benyamin, B., de Leeuw, C. A., Sullivan, P. F., van Bochoven, A., Visscher, P. M., & Posthuma, D. (2015). Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nature genetics, 47, 702–70.
2. Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9(5), 160-164.
3. Barnes, J. C., Wright, J. P., Boutwell, B. B., Schwartz, J. A., Connolly, E. J., Nedelec, J. L., & Beaver, K. M. (2014). Demonstrating the validity of twin research in criminology. Criminology, 52(4), 588-626.
4. Wright, J. P., Barnes, J. C., Boutwell, B. B., Schwartz, J. A., Connolly, E. J., Nedelec, J. L., & Beaver, K. M. (2015). Mathematical proof is not minutiae and irreducible complexity is not a theory: a final response to Burt and Simons and a call to criminologists. Criminology, 53(1), 113-120.
5. Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of genotype→ environment effects. Child development, 424-435.
6. Harris, J.R. (1998). The Nurture Assumption: Why Children Turn out The Way They Do. London: Bloomsbury.
7. Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716.
8. Barnes, J. C., Boutwell, B. B., Beaver, K. M., & Gibson, C. L. (2013). Analyzing the origins of childhood externalizing behavioral problems. Developmental psychology, 49(12), 2272.
9. Barnes, J. C., Boutwell, B. B., Beaver, K. M., Gibson, C. L., & Wright, J. P. (2014). On the consequences of ignoring genetic influences in criminological research. Journal of Criminal Justice, 42(6), 471-482.
10. Pinker, S. (2002). The Blank Slate: The Modern Denial of Human Nature. New York: Viking.
