Saturday, February 20, 2016

Correction

Image via Hora de Aventura.
Early in November, I was very startled by some research reported in a paper by Anne Case and Angus Deaton find that as mortality rates have generally been dropping all over the US there is one group where they have been going up rather steeply between 1998 and 2013, that being undereducated non-Hispanic white people (high school diploma or less) between 30 and 64, from causes dominated by suicide, liver cirrhosis, and drug overdose.

Some new work is putting a somewhat new reading on the numbers, I learn from Katie Surrence over at LGM, and I also find that I made an appalling error myself in my post back then, so I have to correct that.

My error was to say that it was white non-Hispanic men whose death rates were rising, which is not what that report said. I unconsciously made it up, I guess out of the belief that suicide and intoxicants are typically guy deaths, but the fact is there was no gender breakdown at all in the Case and Deaton study.

And in fact in the new treatment by Andrew Gelman and Jonathan Auerbach, adjusted for age groups and including gender breakdown, it turns out the people whose mortality rates are rising are non-Hispanic white women, ages 45 to 54, mostly in the South but to some extent also in the Midwest. (The rates for non-Hispanic white men in the South were rising pretty sharply from 2000 through 2008, but have declined since then.)


What it means I don't really know, but it does mean my Beavis and Butt-head scenario isn't exactly right, so let the record be revised accordingly.

No comments:

Post a Comment