Credit: Michael Pettigrew
Autism is in the news again this month, which can only mean one thing: Researchers have sparked another round in this blame game.
In the last 25 years, the incidence of autism—a developmental disorder characterized by impaired social and communication skills and repetitive behaviors—has exploded in the U.S..
Though the increase can be attributed partially to improvements in diagnosis, scientists have blamed the rise in autism on a myriad of factors, from food allergies to the now widely discredited idea of "cold" mothering. One of the best-known explanations, which linked mercury-based vaccines with autism, received enough publicity and support to send many new mothers scurrying away from the syringe.
Advertisement
But the recent dismissal of the vaccine hypothesis by the National Institutes of Health has left the door open for new theories, and Cornell University economist Michael Waldman provides one. In a recent working paper published on the Cornell website, Waldman and his colleagues proposed a relationship between autism and something far less insidious than a mercury-filled needle: television.
Though autism is now thought to have a large heritable component, scientists believe that environmental factors might help set the disorder in motion in people with a genetic predisposition. Inspired by research linking TV-viewing to Attention Deficit Hyperactivity Disorder, Waldman wondered whether television could be just such an environmental factor.
The researchers didn't have data about TV-watching across the lifetimes of autistic children, so they decided to look for indirect links between television and autism by correlating both with weather patterns. The Bureau of Labor Statistics' American Time Use Survey shows that television watching among young children is positively correlated with precipitation. When it rains or snows, kids watch television more than when it doesn't.
So, Waldman and his colleagues set out to determine whether precipitation was also related to autism rates, calculating the correlation between the two in various U.S. states and counties. They discovered a significant connection. During years when precipitation in a particular area occurred infrequently, so did autism diagnosis. When precipitation rates rose, so did rates of autism.
"Look at a map of the United States at what states have high autism rates," Waldman said. "Oregon, Minnesota, Wisconsin, Maine—rainy states. The data show that children spend more time indoors watching television in those places."
The trouble with this approach is that autism and TV are not necessarily correlated with each other just because autism and television are each correlated with precipitation. To account for this complexity, the authors examined another variable: cable television subscription. By showing that autism diagnoses are positively correlated with subscriptions to cable television—which offer children more opportunities to watch TV—they hoped to strengthen the case that autism truly is linked to the amount of time kids spend in front of the tube, not simply how much it rains in their communities.
Using this approach, the researchers concluded that as many as 38 percent of autism cases are attributable to early childhood television viewing.
But the results are controversial, and a host of criticisms have been made regarding the researchers' methods and conclusions.
"The authors have done some interesting work, but the nature of the problem makes it a really hard one to answer convincingly," said Steven Levitt, an economist at the University of Chicago and the author of Freakonomics."I applaud the authors for asking a daring hypothesis and gathering data to try to test it. My gut, though, tells me that this is a result which will not stand up to scrutiny."
To confirm that television viewing actually causes autism, researchers would have to design a controlled experiment, exposing groups of children to different amounts of television and then comparing how many cases of autism develop in each group. Because such research is not feasible or ethical when dealing with humans, scientists must rely on statistical correlations.