Demographic Doctoring
Journalist and researcher Colin Woodard has followed in the footsteps of historian David Hackett-Fischer in finding new and novel ways to break the United States down into geographical and demographic segments with defined common traits. In his magisterial 1989 book Albion’s Seed, Hackett-Fischer examined four American folkways by tracing their roots back to the parts of Great Britain each region’s forebears came from, and taking a closer look at the similarities between their British source populations and the American progeny.
Colin Woodard has undertaken a similar challenge, but widened the scope of source countries and cultures for the constituent American identities. His book on the topic was American Nations, published in 2011. He has been refining his hypothesized eleven cultures ever since, and most recently wrote up an analysis of the different cultures and their divergent life expectancies in Politico magazine, as Josh has sent in via email.
As Woodard says in his article:
The results show enormous gaps between the regions that don’t go away when you parse by race, income, education, urbanization or access to quality medical care. They amount to a rebuke to generations of elected officials in the Deep South, Greater Appalachia and New France — most of whom have been Republican in recent decades — who have resisted investing tax dollars in public goods and health programs.
That would seem to be an informative finding—if your aim was to inform others that your political preferences, for instance of a Democratic-leaning tendency, are not only right but born out by statistics. But if you approach the analysis from a different political perspective, the associations may not be as persuasive.
For one, the numbers for average life expectancy are very broad, encompassing large swaths of geography. It raises the question of how the life expectancies would further break down within those geographic areas. There’s a good chance there are population centers that bend the numbers in distinct ways by being outliers.
For another, the essay takes these very broad and potentially overgeneralized numbers and draws some surprisingly specific conclusions from them relevant to today’s political fights.
We repeated the experiment using counties that fell in the worst quartile for clinical care and saw the gap grow even wider, with Greater Appalachian (74.6 [years of average life expectancy]) and Deep Southern (74.7) life expectancy in those communities lagging Yankeedom by about 3 years and New Netherland by about five and a half. That there are fewer counties where most people can afford and access top-notch clinical care in these southern regions than the northern and Pacific coast ones isn’t really a surprise: laissez-faire political leaders tend to create systems that have looser health insurance regulations, leaner Medicaid programs and fewer public and non-profit hospitals. That those that do manage to have decent services nonetheless underperform suggests reversing these gaps won’t be easy.
We could all live together happily—if your preferred health policies weren’t shortening our lives!
I guess my gut reaction to this sort of motivated reasoning is contrary motivated reasoning. Still, to the extent there’s much useful information in epidemiological survey data, making it fit into oddly specific policy preferences should make the researchers a bit more circumspect. It’s possible there’s helpful information buried somewhere in the data, but you won’t find it by limiting your search to things aligning with your political preferences.
In today's headlines, just a few minutes ago:
Fortune: Entrepreneurs not wanted: Research suggests hiring managers discriminate against small business founders
NYT: Entrepreneurs not wanted: Research suggests hiring managers discriminate against small business founders
Is it just me? Why read these? Aren't they likely just common sense dressed up to look important? In the Fortune article, they even refer to research to say what, I think, is common knowledge in the hiring world and certainly among former self-employed folks.
That function works about 90% of the time. The other 10% usually gets me in trouble.