When researchers use percent of body fat data to assess obesity rather than body mass index (BMI), the huge gap in obesity rates between African-American and white women, for example, is cut in half, and white men are found to have a much higher risk of obesity than African-American men.

While the medical literature has long showed that percent of body fat is a more accurate measure of fatness than BMI, most social science researchers still use BMI because that is what is available in most social science-based data sets. But because BMI ignores the difference between fat and fat-free mass like bone and muscle, such studies, for example, overstate the obesity of African-Americans relative to whites. That’s because on average, African-Americans have more nonfat mass, says John Cawley and Richard Burkhauser, professors of policy analysis and management who have conducted an analysis comparing measures of obesity. Their report offers a conversion tool to enable researchers using social science data sets to calculate body fat percentages and other more accurate measures of obesity in data sets that only contain information on height and weight.

Using more accurate measures of obesity makes a difference. “We’ve already found indications of a correlation between high percentages of body fat and employment disability,” said Burkhauser. “And we wouldn’t have been able to determine that using BMI.”

The researchers’ working paper was released in June by the National Bureau of Economic Research.

Cornell University