On Sat, 7 Dec 2002, Doug Henwood wrote:
> Michael Pollak wrote:
>
> >In today's FT, Michael Prowse approvingly cites Richard Wilkinson's claim
> >that unequal societies are less healthy. He does say the claim is
> >"controversial," but also says that outside researchers have confirmed the
> >claim and bases his whole argument on its truth.
> >
> >My impression last time this got discussed onlist was that Wilkinson's
> >data and math didn't support his claim at all, and all reputable health
> >researchers, including many who wanted to believe it, were soon forced to
> >give up this argument up for dead. Has anything changed?
>
> Nope. As sociologist Christopher Jencks told me, you can make the
> relation between inequality and health stats appear and disappear,
> just by changing the specifications of your model. Not a robust
> statistical finding, in other words. I was sorry to see Prowse
> pushing this.
>
> Doug
>
This is a little facile. If you change the specification of any statistical model, it will affect the values of any existing regression or path coefficients. W's book reports various data sets that measure inequality and public health using many different indicators, in different time frames. They all support the claim that income and wealth disparity is negatively correlated with public health. Can you cherry pick data from specific time frames or specific nations that contradict the general trend? Sure. That doesn't make the general trend go away.
Granted, W's book is not a comprehensive literature review. Nevertheless, Jenck's criticisms (as I've heard them summarized) are not the slam dunk refutation that Doug assumes.
Miles