Correlation is not causality: income and health

MarginalRevolution:

This paper provides new evidence on the causal relationship between income and health by studying a randomized experiment in which 1,000 low-income adults in the United States received $1,000 per month for three years, with 2,000 control participants receiving $50 over that same period.

We can rule out even very small improvements in physical health and the effect that would be implied by the cross-sectional correlation between income and health lies well outside our confidence intervals.

In other words, in this randomized control trial (the gold standard in causal inference) higher income does not cause better health. The other explanation goes by the name of "selection bias" or "omitted variables bias," i.e., there is some other factor that causes both.  And this leads into the nature vs. nurture debate, covered so well in Steven Pinker's The Blank Slate.  (Earlier blog post on Pinker's book, Is Nature Stronger than Nurture.)