Yoga Girl

Rich or Poor, Fit or Fat — Beware Generalizations

A new study in the American Journal of Preventive Medicine brings out some dangerous generalizations about people who are rich or poor, fit or fat. It starts with generalizations by the investigators that their data can’t really support. Then The Atlantic amplifies it with a provocative headline that straddles a line between condescending and offensive:

Rich People Exercise, Poor People Take Diet Pills

The study analyzes self-reported data on trying to lose weight and the strategies people say they use to lose it. The data comes from the National Health and Nutrition Examination Survey (NHANES). Through a multivariable logistic regression, they determine that people with high incomes (≥$75,000/yr) are more likely to use recommended strategies for weight loss than people with low incomes (<$20,000/yr).

Two problems get in the way of the sweeping generalizations the study’s authors try to make:

  1. Misclassification. The criteria these authors use for classifying weight loss strategies as “not recommended” are a bit murky and, in any event, flawed. For example, they count exercising to lose weight as a recommended strategy. But, says Diana Thomas, “Research has shown that with heavy doses of exercise, people eat more and with low doses, it’s not enough to lose weight.” Thomas is Director of the Center for Quantitative Obesity Research at Montclair State University. While exercise is recommended for good health and preventing weight gain, it’s not an effective way to lose weight.
    But two evidence-based weight loss methods — liquid meal replacements and prescription obesity drugs — are counted in this analysis as “not recommended.” Obesity medicine physicians would disagree.
  2. Self Reports. Basing a quantitative analysis on self reports begs for trouble. Self reports introduce well-documented bias that can nullify the findings. People significantly misreport weight and weight-related behaviors. The misreporting is variable and has been shown to vary with characteristics of the respondent, such as income. It’s entirely possible (perhaps likely) that higher income respondents are more likely to misreport that they use more “recommended” methods. “Why yes, I work out all the time.”

All of this fits neatly with a bias that obesity is a disease of poor, ignorant people who make really bad choices. That’s where The Atlantic comes in with their headline and condescending essay.

The over-arching problem with these generalizations is that they’re used to explain a false assumption that obesity is a disease of social and economic disparities. Roland Sturm, a senior economist at Rand, recently published a detailed analysis showing this to be false.

Rising rates of obesity are uniformly affecting diverse segments of population grouped by race, ethnicity, income, education, or geography. Though different segments started with different rates, Sturm presents a compelling analysis that shows all segments are moving up in parallel.


“All generalizations are dangerous, even this one.” — Alexandre Dumas

Click here to read the study in AJPM and here to read the story in The Atlantic. For Sturm’s analysis, click here.

Yoga Girl, photograph © Prashant ZI / flickr

Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.