9.25.2009

Annals of questionable science

I see, via HuffPo, that there is a new study being publicized that says spanking your kids will make them dumber. From America's premier source of hard science news, aka the Today show, I give you this:
Spanking can get kids to behave in a hurry, but new research suggests it can do more harm than good to their noggins. The study, involving hundreds of U.S. children, showed the more a child was spanked, the lower his or her IQ compared with others.

[snip]

Straus and his colleague Mallie Paschall of the Pacific Institute for Research and Evaluation in Maryland studied nationally representative samples of two age groups: 806 children ages 2 to 4, and 704 ages 5 to 9. The researchers tested the kids' IQs initially and then four years later.

Both groups of kids got smarter after four years. But the 2- to 4-year-olds who were spanked scored 5 points lower on the IQ test than those not spanked. For children ages 5 to 9, the spanked ones scored on average 2.8 points lower than their unspanked counterparts.

[Massive preemptive disclaimer -- just because I think this study is weak and flawed does not mean that this post should be interpreted to mean "up with spanking." Like the vast majority of my colleagues in pediatrics, I do not support corporal punishment.]

First of all, I am innately suspicious of any scientific finding that is first reported in the popular media. The media is, like most people, more interested in stories than science, and doesn't provide the kind of scrutiny that peer reviewed journals would. Further, it appears that the first time this study will be subject to critique from the scientific establishment, it will be at a conference instead of through the more rigorous review provided when such studies are submitted for publication in reputable journals. I don't know much about the International Conference on Violence, Abuse and Trauma, but I wonder if they will be inclined to accept the study's findings with minimal scrutiny because they comport with the biases of those who would attend such a conference in the first place.

Further, the design of the study seems weak to me. They followed a cohort of children over time, and compared the kids who were spanked to those who weren't. Now, obviously, you can't do a "better" study by randomizing kids to a "spanking" group and a control group. ("We're going to hit a bunch of kids in the name of science" would probably fail to impress the Internal Review Boards of most centers for higher learning.) But this study (and, again, it hasn't been submitted for peer review as far as I can tell, so this is just a guess) leaves open the possibility of confounding factors. In other words, even controlling for socioeconomic status, is there something about the kinds of families in which children are likely to be spanked that limits intellectual development, and not the spanking itself? I have doubts that this study sifts out the noise.

Now, again, if this gets parents to hit their kids less, then it's all good, right? From a certain perspective, sure. But I don't like shoddy science, particularly when it gets hyped, even if I like the findings. If we fail to adequately criticize science just because we like what we hear, we will lose the ability to conduct scientific inquiry with objectivity. In other words, at all.

4 comments:

  1. It makes no sense to me, I would like to see the hard science, that a hit to the butt causes neurons to die, or some biochemical reaction to take place. My take is that parents who are pre-disposed to hit (but for whatever reason didn't) with all of the environmental factors remaining the same, the IQ's would have the same effect. This seems like too much junk science, and the only way to even come close to proving it, have a set of identical twins with one being spanked and the other not, is far more horrendous to family dynamics then treating them equally.

    I have 3 boys. I very rarely spank them, but I have and the times I have it has been very effective, they simply have not repeated the offense that led to it, and even if the few cases I have if it leads to 3 point drop, well in some cases even that is worth it (I am thinking of the time one of my boys ran out in the street while I was screaming no) It wasn't punishment, it was reinforcement, and it worked.

    charo

    ReplyDelete
  2. Dear Lord. Can someone please sit these people down and explain the difference between a cause and a correlation? Really. Not that hard.

    Charo, meant to reply to you on my post a million years ago and agree that I did not mean to imply there was a one-size-fits-all parenting.

    ReplyDelete
  3. Failure to understand the difference between correlation and causation is one of the hallmarks of Bad Science. It is such a useful marker that I would suggest NOT explaining it to the purveyors of Bad Science. The cat comes pre-belled, why tell it to take the bell off?

    ReplyDelete
  4. Note: just because there are significant limitations in a study doesn't mean that one not ought to publish the results (just that one ought to be very clear about the limitations in the study itself).

    I don't know how much of this is bad study design, and how much of it is just crappy science reporting. Generally, when I investigate these sorts of things, I find the culprit is crappy science reporting, not crappy science. Nobody should be tootin' up a conference paper in the general media as being *anything* resembling a significant finding, unless it's a truly remarkable result. The only way to find out if it's a truly remarkable result is to sit in on the conference paper session and watch the audience dissect the results, and then follow the conference attendees around and see which results are the topics of conversation at the bar after the keynote dinner speech.

    ReplyDelete