Are Crime Rates Causally Linked to Lead Exposure? Updated.

Steven Hayward on PowerLineBlog has a fascinating post about the dispute over crime rates and lead exposure (with a 23-year lag) in children.  The correlations are pretty striking, for sure – some folks see evidence that this is “causal correlation”, but many do not.  Hayward’s post has many links, including to Mother Jones, the Economist, Environmental Research, a back and forth between Kevin Drum and Jim Manzi, etc.  If this subject interests you at all, it’s a real resource.

Drum’s brief summation of his position is:

It’s the only hypothesis that persuasively explains both the rise of crime in the ’60s and ’70s and its fall beginning in the ’90s. Two other theories—the baby boom demographic bulge and the drug explosion of the ’60s—at least have the potential to explain both, but neither one fully fits the known data. Only gasoline lead, with its dramatic rise and fall following World War II, can explain the equally dramatic rise and fall in violent crime.

There are others who agree, but Manzi says that tweaking the way some of the variables are included in the statistical analysis (I’m lost, here – will get my sister to check it out for me) removes lead’s influence almost entirely.  On the other hand, Hayward points out that Drum’s article (perhaps inadvertently) hints at a “public choice” angle — the prison guards, construction companies, police unions drug warriors, political conservatives, etc. all have a vested interest in debunking any medical explanation for crime.

I don’t have an opinion, except that this would be REALLY good news, if it proves out.  And phooey on special interests that don’t like it — the buggy-whip factory owner was pretty P.O.ed about the Model T, as I remember.

UPDATE: My beloved baby sister, who sold her services as a statistics whiz to her fellow graduate students, and then worked for the state of KY doing similar work for a number of years, reports:

“Correlation NEVER “proves” causation (I knew this from my own stats classes), regardless of how strong the correlations are.  There are always alternative explanations out there.  For instance, this one:  http://www.freakonomics.com/2005/05/15/abortion-and-crime-who-should-you-believe/ (I remember reading this hypothesis, but was not aware that there was as much support for it as appears here.)

“It’s an interesting hypothesis and I certainly believe that lead exposure is bad for kids, but I would need to see some indication that it was actually the kids with the highest lead levels who were committing the assaults in order to say that lead exposure was linked to crime. (It would be VERY interesting to have these data in hand.) These analyses were not done at the level of the individual.”