Enough With the Middle-Age Whites, Already

In a Washington Post column over the weekend, Fareed Zakaria wrote this:

In a well-known paper, economists Angus Deaton and Anne Case found that over the past 15 years, one group — middle-age whites in the United States — constitutes an alarming trend. They are dying in increasing numbers.

I suppose I'm tilting at windmills here, but this misconstrual of the Case/Deaton paper really bugs me. It's not "middle aged" whites who are dying of drug and alcohol poisonings, suicide, and chronic liver disease in vastly greater numbers. The chart on the right is taken directly from data in the Case/Deaton paper, and it shows two groups with outsize increases in mortality: 30-34 year olds and 50-59 year olds. So there are three age cohorts here. Roughly speaking, one cohort was born in the early 50s and shows big problems; another was born in the 60s and shows only moderate problems; and the third was born in the early 70s and once again shows big problems.

These cohorts might change if you examine the data using different age buckets, different diseases, and a different timeframe. Who knows? Regardless, if you're going to put forward an explanation about why this is happening, it better account for all three age groups. You can't just pretend the data points only to "middle-age" whites and then spin your theories from that.

Oh, and the paper is by Case and Deaton, not Deaton and Case. Either way, though, I sure wish one of them would step in to correct this.