Two years ago, 10-year-old Kimberlee Ann Palmer of Portland, Oregon, frightened and depressed by her father’s suicide, picked up her brother’s .22-caliber rifle and shot herself in the head. Kimberlee was one of hundreds of American children between the ages of 5 and 14 who kill themselves each year with firearms-a statistic that prompted Mathew Miller of the Harvard Injury Control Research Center to investigate how much the availability of guns contributes to child suicides. His results have led to a heated debate on statistical methods.
Miller reports that in the five states where people own the highest number of firearms, far more suicides are committed by kids between the ages of 5 and 14 than in the five states with the lowest gun ownership. The fact that childhood suicides from causes other than guns are roughly similar in both groups of states, says Miller, further indicates that gun availability accounts for the difference.
But Miller’s dramatic findings are vigorously contested by Paul Blackman, research coordinator at the National Rifle Association. Chiefly, Blackman challenges the method by which the study estimates gun ownership.
Since many states do not require people to register their guns, the number of Americans who actually own firearms cannot be precisely gauged. Some states provide estimates on gun ownership based on telephone surveys and, where possible, Miller and his colleagues used them. But they also used what statisticians call proxies-that is, they analyzed indirect measures of gun ownership. One, called Cook’s index (named after Duke University economist Philip Cook) extrapolates the number of firearms in a state by averaging the percentage of suicides and homicides that are committed with firearms. A second extrapolates solely from the percentage of suicides that were committed with a firearm. “These proxies aren’t reliable,” says the NRA’s Blackman. Cook’s index, he says, is so inaccurate that “not even Cook accepts it anymore.” And the telephone surveys were conducted on people over 18-an unreliable measure, he argues, of how many guns are available to kids. What really needs to be measured “is the percentage of households with guns.”
Miller concedes the point, but stands by his proxies. He acknowledges the shortcomings of Cook’s index but says he used it only as a point of comparison with the other proxy, which, he claims, has been validated by other studies. Moreover, both proxies, and the surveys, he says, yielded essentially the same result. And that, he says, is the strength of his study.
But Blackman remains unimpressed. “Even if the data were reliable,” he argues, “the whole study, in a sense, is trivial. It looked at what amounts to only 1.5 percent of gun-related deaths in America.”
Miller couldn’t disagree more. “We’re talking about the deaths of almost 7,000 kids over the decade we studied,” he says. “That is not trivial.”
Children aged 5-14: 23 million
Gun suicides: 153
Non-gun suicides: 69
Children aged 5-14: 22 million
Gun suicides: 22
Non-gun suicides: 82
* High-gun states: Louisiana, Alabama, Mississippi, Arkansas, and West Virginia
** Low-gun states: Hawaii, Massachusetts, Rhode Island, New Jersey, and Delaware