View Single Post
Old 02-21-2013, 07:06 PM   #187
El Minion
Ring of Famer

Join Date: Apr 2005
Posts: 1,771

Originally Posted by Blart View Post
Scientists try to tease our causation from correlation, relying on what the data explains best, controlling for factors when they can, and coming up with the best explanation possible...

But you're right, it's not perfect, and there is no singular reason for crime. There are millions.
Parental involvement, education, community involvement, gun control, incarceration rates, drug control, mental health services, income inequality, access to birth control, etc.

However, despite the best efforts of Jim Manzi (expert statistician and conservative, who found that the above could explain up to 20 percent of the following, but failed to replicate those results based on further evidence) no combination of the known crime correlates could explain this:

Steven Levitt (of Freakonomics fame) pleased many liberals, including me, when he published data that showed a probable link between abortion and the drop in crime, which wasn't perfect but explained the phenomenon better than anything else at the time.

That's what I believed up until last month. Call me a flip flopper, but my beliefs change based on evidence.

Cities, states, and countries all over the world have seen a similar rise and decline. The similarity they all share? Not abortion services, not single motherhood, not black youths - just the rise and fall of leaded gasoline, which can explain up to 90% of the crime wave over the late 20th century.

The evidence goes further than the national, state, and city level - follow-up studies have found it on the individual level. Childhood blood lead levels are consistently associated with arrest rates for violent crimes.

All that leaded smog didn't completely go away, it's still in our soil - especially around inner-cities where traffic was worst. It's still in old houses, it's especially in old windows.

If you're not convinced, just wait. Western Europe used leaded gasoline longer, so we can expect crime to drop there in the next 20 years, with a drastic drop over the next 10. In Russia, the drop should happen a little later, since the Soviets used leaded gasoline longer. We should see the beginnings of a crime decrease in Latin America (they need it) around 2018, since they used leaded gasoline until 1995.

If crime decreases don't happen in the countries above, we can label the theory false or flawed.

Regardless, this is all much more interesting to me than Ted Nugent's opinions or txtebow's racial stereotypes.
Read this on Reddit, where Levitt participated in AMA, someone asked him about the lead hypothesis and he apparently addressed it back in 2007:

Lead and Crime

Steven D. Levitt
07/09/2007 | 10:04 am

Over the weekend, the Washington Post published an article suggesting that much of the decline in crime in the 1990s may have been due to the reduction of childhood lead exposure after the removal of lead from gasoline and house paint.

This is an intriguing hypothesis. There is evidence on an individual level that high exposure to lead is harmful to both IQ and the ability to delay gratification, two traits that could enhance the attractiveness of crime. There is also some suggestive time-series evidence of a relationship: the rise and fall in lead exposure at the national level match the rise and fall in crime. Still, although both Post reporter Shankar Vendantam and the cited economist, Rick Nevin (whom Iíd never heard of), appear quite convinced by the time-series data, I am not. When you have a variable like crime that goes up for a long time then goes down for a long time, it is easy to find other variables that share that pattern and appear to have a causal impact, even though the relationship is completely spurious.

About seven years ago, Michael Greenstone and I tried to look into this same issue using airborne lead measures at the local level, as well as other approaches. We ultimately gave up without finding anything. That largely soured me on the lead/crime link.

Recently, however, Jessica Wolpaw Reyes at Amherst has put together what appears to me to be the most persuasive evidence to date in favor of a relationship between lead and crime. Rather than looking at a national time-series, she tries to exploit differences in the rates at which lead was removed from gasoline across states. I havenít read her paper with the care that a referee would at an academic journal; but, at least superficially, what she is doing looks pretty sensible. She finds that lead has big effects (and, for what itís worth, she also confirms that, when controlling for lead, the link between abortion and crime is as strong or stronger as in our initial study, which did not control for lead.)

Roger Masters, a professor at Dartmouth, has also been doing interesting research on this subject, although I am also not very familiar with his work.

It will be very interesting to see how this research agenda plays out. If it can be shown here and in other areas that environmental factors have powerful and long-lasting impacts on human behavior, it may dramatically change the way we think about public policy.
El Minion is offline   Reply With Quote