I'm tempted to repeat the oft said correlation is not causation. But really I have never thought that captures the problem I have with studies like this. Of course correlation does not necessarily imply causation. What I really find to be problematic is a small time series of data of a facet of a human behaviour across a population is probably going to correlate extremely well to many other time series across the sam…
1) Everything seems to work out perfectly, no matter how it is considered. For instance, if lead contamination in the ground is such a big deal, why does lead in gas track so closely to the crime rate? Wouldn't it be the case that it would track on the upside but not the downside?
2) He keeps on emphasizing lead lowers IQ. But the Flynn Effect suggests that IQ started going up in the 1930s and has only (maybe) started to level off very recently. It's the exact opposite of what you'd expect based on lead exposure from leaded gas.
3) Long term crime rates have been mostly going down for centuries: http://marginalrevolution.com/marginalrevolution/2011/06/lon...
Mind you, I'm fully in agreement that lead sucks and I sure as hell don't want my kid exposed to it. It's just that I'm far from convinced that the crime rates change mostly because of lead exposure. It seems far too tidy an explanation.