The raw death rates for the group (which appeared in the Case-Deaton paper) are in red, and the age-adjusted death rates (weighting each year of age equally) are in black.
So . . . the age-adjusted mortality in this group increased by 5% from 1999 to 2005 and has held steady thereafter. But if you look at the raw data you’d be misled into thinking there was a steady increase. That’s the aggregation bias I’ve been talking about here and here.
For some reason it’s not so easy to get the numbers before 1999. But, following Deaton’s tip, I grabbed the 1999-2013 data and made some plots. All are renormalized to be relative to 1999.
Based on my earlier analysis, I’m guessing that age-adjusted mortality in this group dropped pretty dramatically from 1989 to 1999. Hence the title of this post.
The natural next step is to break this one up by men and women, and by ethnic group. And someone should do this. But not me. I got a job, and this ain’t it.
P.S. In the original version of this post I referred to “non-Hispanic white men.” I don’t know why I wrote that. All these graphs are for non-Hispanic whites, both sexes. As noted above, it would be easy enough to do separate calculations for men and women, but I didn’t do that.