What kind of New York Times Article file name is that? A health-related one, it turns out. Let us set the scene, figuratively. A student walks into her advisor's office. The rustle of papers pauses. Her academic future is being crafted by the decisions they will make. Her professor has told her he has a research idea for her to pursue. A project for her to take responsibility for. What will be her contribution to human knowledge? What intellectual adventure will he send her on? What must she do?
She must go to supermarkets, decide if she thinks a child is ugly, and then see if they've been properly strapped to their parents' shopping carts.
Alright, so honestly I have no idea if that's how the project was conceived at the University of Alberta, but I gotta imagine that there was some interior drama when researchers were informed this was how they would spend their days. Back to the facts: According to Nicholas Bakalar, author of today's NYT article, "Ugly Children May Get Parental Short Shrift," Dr. W. Andrew Harell led a research team that went to 14 supermarkets, observing 400 child-parent interactions, rating each child's attractiveness on a 10-point scale. (That's right, children of Alberta, stumbling upon this blogpost in a few short years when you're old enough to search the internet--those childhood grocery store trips you remember so fondly? That guy with the clipboard and the funny mustache who was standing behind the lettuce? He was deciding if you were ugly or not. And you thought all Canadians were so nice.) Sorry. The facts, again, from the article:
When a woman was in charge, 4 percent of the homeliest children were strapped in compared with 13.3 percent of the most attractive children. The difference was even more acute when fathers led the shopping expedition - in those cases, none of the least attractive children were secured with seat belts, while 12.5 percent of the prettiest children were. Homely children were also more often out of sight of their parents, and they were more often allowed to wander more than 10 feet away.The study hasn't been published yet, so there's no quick way to get at the details of their methods and data. I'm not convinced by the statistics, but I'll leave that alone for now. The method as described seems a bit weak, though. Did the same researchers who rated the children aesthetically also rate the parents' safety precautions? That hardly seems like a double blind study. In such a perception-heavy field, that seems like it leaves a little too much discretion to the perception of the reseachers. Click the timestamp/permalink for more of my experimental critique. Here's what I would have done: Take a little video of each interaction, along with a good snapshot of each child. This is perfectly possible with today's little digital cameras, if not necessarily legal. Index them to each other, blur the faces in the videos, and then give one set (photographs) to one group of researchers to rate for ugliness, and another set (videos) to another group of researchers to rate for neglect or danger.
I felt a bit of experimental angst reading this data exactly because of the statistics. Since it is so very perceptive, it seems to me that you would want to duplicate this experiment many times in different places, with different people, before you ever said anything. Then you could at least look at the standard deviation of the results, etc. etc. As it is, you are basically stuck with the statistics of a survey. The researchers seem to have rather arbitrarily labeled their subjects as following into one of two catogies, A & B ( pretty and ugly) and then are basically interrogating those subjects (not literally, but figuratively) for a binary measure of neglect: yes or no. At best the relevant sample size is 400, and the confidence bars should go 1/SquareRoot(400), or 5%. But the 13.3% and 4% figures are each of the subgroups, ugly and pretty. Assuming, nicely, that there are 200 of each we are left with 7%. Which is just a bit to close to the actual difference (9.2%) to make me feel this is an earth-shattering conclusion. I'd also be more convinced if it wasn't plausible there was something special about the 9.2% difference that the researchers hadn't noticed. Or if the study was done of fraternal, same-gendered twins. But it's good enough for Maureen Dowd to get a clever column out of today.
This reminds me of the Akbar's wiseman Birbal. Akbar asked Birbal to find him the prettiest child in the capital. Birbal went and found what the courtiers thought was an extremely ugly child. They were standing around staring at the child, confounded by Birbal's choice, making the child very uncomfortable. The child started crying, and his mother ran out--after chastising the men for bothering her child, she took him on her lap, covering him with kisses and calling him her beautiful one. The Emperor got Birbal's point: to many a mother, their children are the prettiest. So far, my guess is still that Birbal is usually right.
Link from Slate's human nature by William Saletan.
