Michael asked what the problem was with the original Wilkinson study
(showing that mortality is related to income inequality among developed
countries.) I don't know that it is problem, but I think it was Lynch who
added more developed countries to the list and found the relation was less
clear than suggested by Wilkinson's original results (but it still exists).
I attended a recent talk by Lynch where he made a convincing-sounding
argument about how eating pasta and drinking red wine in Southern Europe
are another part of why their populations are so much healthier than in
Northern Europe (and the worst case of all, the United States).
A recent article by Lynch (with Ross et al.) in the British Medical
Journal, on the relation between income inequality and mortality in
Canadian and US cities, is available at
http://www.bmj.com/cgi/content/full/320/7239/898. They estimate that if the
income share received by the poorer half of all working age households was
increased by 1%, then mortality would hypothetically decline by 21 deaths
per 100,000. How many other policy interventions would match this rate?
In an earlier paper Lynch showed that mortality in US states is more
closely related to income inequality than to income. US cities with greater
income inequality have significantly higher mortality independent of their
average income.
In case this is where Doug's uncle comes in, the BMJ paper also cites
evidence against the argument that the observed relation between health and
income inequality is an artifact of the non-linear relation between income
and mortality at the individual level.
One question I have about these results is how income and income inequality
are treated as 'independent' variables (independent of each other). Is it
really true this can be 'fixed' with fancy adjustments to the regression
equation and results?
Bill Burgess
