It is difficult to imagine more of a hornets’ nest than the etiology of mental illness. So when Cornell University economist Michael Waldman poked it with a stick last fall, suggesting that television was to blame for the rise in the incidence of autism among young children in the US, he got plenty of stinging criticism, much of it from the medical community.
Last week The Wall Street Journal demonstrated why that newspaper remains the gold standard for journalistic explanation. In a story headlined “Mind and Matter,” WSJ reporter Mark Whitehouse laid out not only the controversy attending Waldman’s paper, but also described the nimbus of doubt that surrounds the current vogue among economists to discover “natural experiments” whose mysteries may be plumbed through the use of “instrumental variable” statistical techniques.
The rest of the page one headline asked, “Is an Economist Qualified to Solve the Puzzle of Autism? Professor’s Hypothesis: Rainy Days and TV May Trigger Condition.”
Autism, of course, is the condition mentioned by that name first in 1912 but authoritatively characterized only in 1943 that presents itself in some children, almost always before the age of three: they experience social and language difficulties, and engage in restricted and repetitive activity. Perhaps you’ll want to read the piece itself. A subscription to the online WSJ is required to read it here for free — otherwise you can buy it, get a copy from a friend or visit a library. Read the story carefully, and you’ll see that Whitehouse has laid out the story in sufficient detail to understand the controversy, and placed it within a broader context, as well.
For it is not just Waldman’s findings of the effect of television viewing on autism that are of interest, but those of Joshua Angrist on the effect of Vietnam service on draftees’ lifetime earnings; Caroline Hoxby on competition among schools and student performance; Steven Levitt on police “feet on the street” and crime. Venerable authorities appear throughout the story, like a Greek chorus, counseling caution: the University of Chicago’s James Heckman, MIT’s Jerry Hausman, Berkeley’s David Card.
In each case, to make an argument, the investigator had to find a proxy, an “instrument,” through whose variation could be gauged the suspected but otherwise unobservable relationship of cause and effect between primary variables.
In Waldman’s case, it was rain.
This business began in the autumn of 2003, when his two-year-old son was diagnosed as having a mild autistic disorder. This sometimes is enough to turn spouses on each other. Waldman, then 48, an applied theorist who had worked on wage and promotion practices, pricing strategies, copyright policies and business cycles, turned his energy outwards. He began an energetic study of the condition, and, in addition to pursuing the standard treatments, sharply reducing his son’s exposure to television. (A sister had been born the summer before.)
Within six months, his son ceased to present symptoms of autism. This in itself was unusual since, ordinarily, autism is expected to last a lifetime. Waldman formed the opinion that television had been the root of his son’s problem. He sought to interest medical researchers in his theory. Unable to do so, he embarked on research himself.
What he had in the beginning was his intuition. The incidence of autism diagnoses had been exploding (from one in 2500 children in the US in 1975 to one in 160 children today). The growth of television viewing by young children was thought to be growing dramatically, too, thanks to new technologies (cable, VCR and DVD). Perhaps the two were connected
But there are no Nielsen ratings to point to what little kids are watching, at least not yet. Nor is there any obvious way to launch a clinical trial of the no-television treatment, the way pharmaceutical researchers would investigate the properties of a new drug: identify a population of susceptible babies, randomly assign them to two groups, one with television privileges and one without, and see what happened. With rare exceptions, economists don’t work that way.
If television were a trigger for autism, Waldman reasoned, it should be more prevalent in communities where children watched more TV. So, unable to measure TV time directly, Waldman went looking for a proxy variable — an “instrument.” That’s where the weather came in. The more precipitation, the more time in front of the TV, at least for some kids — that much had been established by the Bureau of Labor Statistics American Time Survey. And precipitation was something that people did measure carefully.
Working with county-level data for three states in which the weather varied widely from place to place — California, Oregon and Washington — Waldman found that autism diagnoses correlated positively with precipitation. (What if it were too much time spent indoors that causes autism?) He tried a different technique, this time comparing California and Pennsylvania, finding that autism diagnoses correlated positively with cable television subscriptions, and venturing the unusually firm conclusion that the spread of cable had caused roughly 17 percent of the increase in autism in those two states in the years between 1972 and 1989. (What if cable TV news was covering the autism crisis?)
Along the way, Waldman enlisted in his project Cornell health economist Sean Nicholson and Nodir Adilov, a professor of economics at the Fort Wayne campus shared by Indiana and Purdue universities. Last October, the three presented a paper, provocatively titled “Does Television Cause Autism?,” at a workshop of the National Bureau of Economic Research. Cornell prepared a press release.
That it gained sudden prominence owed to a coincidence. Gregg Easterbrook, a political writer and science journalist, last September had written an article for Slate, called “In Search of the Cause of Autism: How about Television?” When he learned about Waldman’s paper a month later, he wrote a second story: “TV Really Might Cause Autism: a Slate Exclusive, Findings from a New Cornell Study.” The fat was in the fire.
In his Wall Street Journal article, Whitehouse noted the great swings in expert opinion about the causes of autism that have occurred over the decades. For twenty years or so after World War II, many psychologists believe that a lack of maternal affection produced the affliction in some children. Child psychologist Bruno Bettleheim, among others, popularized this “refrigerator mother” theory, also blaming absent fathers, making life miserable for uncountable numbers of parents — until clinical studies showed there was no support whatsoever for the hypothesis. Hereditary factors then took over as the preferred explanation, with a brief vogue for a link to measles vaccinations, which since has been dispelled.
“Some experts think that in reaction to the discredited theories the pendulum has swung too far away from the family,” wrote Whitehouse. He quoted Anna Baumgaertel, a developmental-behavioral pediatrician at Children’s Hospital in Philadelphia, who thinks TV “may lead to autistic behavior in susceptible children, because it interferes with the development of Ôlive’ auditory, visual and social experience.” (Some of her autistic patients have been heavy viewers of video and TV from a very early age, she says.) The American Academy of Pediatrics recommends against any television for children under two; Waldman thinks the age of three would be more appropriate until more is known.
The wide gap between Waldman’s findings and the skepticism expressed by many in the medical community can be accounted for in large measure by the difference in research styles, says Alan Krueger, a Princeton University economist who is a member of the board of reviewing editors of Science magazine. In medical science, no publicity is permitted until an article is accepted, he notes; authors can’t even put their papers up on websites until then. A very different ethic prevails in economics, because the journals are so slow. Papers may be publicized as soon as they are published as “working papers” by the Social Science Research Network or National Bureau of Economic Research. The lesson, says Krueger, at least in economics, is “buyer beware.”
Waldman ands his co-authors have been invited to submit their paper to a medical journal. There they’ll deal with the many objections that have been raised — including the conviction on the part of some researchers that there is no autism epidemic, that the dramatic rise in reported rates is the result of wider public awareness of the condition and better diagnostic tools rather than any significant change in its incidence. They’ll lay out the steps they think will be necessary to delineate the pathways by which they think television can produce the physiological changes that are characteristic of autistic children.
And, almost certainly, they’ll be forced into closer cooperation with pediatricians, brain scientists, cognitive psychologists, linguists, psycho-pharmacologists and other experts dealing with the riddle of autism, engaging in the “multilateral reciprocal scrutiny” that Thomas Schelling has advocated, in a slightly different connection.
Either that or they’ll go back to economics.