UPDATE: FREAKONOMICS‘ guru, Stephen Levitt, weighs in on the lead controversy HERE. And sentencing expert, Michael Connelly offers comments here.
********************************************************************
The award for the weirdest and, in some ways, the most intriguing story of the weekend goes to…. the Washington Post for it’s article titled: Research Links Lead Exposure, Criminal Activity.
It seems that, in the last few years, an economist named Rick Nevin has published a series of studies linking children’s exposure to lead with violent behavior later in their lives. Here’s how the article begins:
Rudy Giuliani never misses an opportunity to remind people about his track record in fighting crime as mayor of New York City from 1994 to 2001….
SNIP
Athough crime did fall dramatically in New York during Giuliani’s tenure, a broad range of scientific research has emerged in recent years to show that the mayor deserves only a fraction of the credit that he claims. The most compelling information has come from an economist in Fairfax who has argued in a series of little-noticed papers that the “New York miracle” was caused by local and federal efforts decades earlier to reduce lead poisoning.
Bill Bratton and Guilani have each labeled the study, “absurd.”
But before you too wave the notion away with a dismissive guffaw, it should be noted that Nevin’s work—conducted over a ten year period—is supported by an earlier study published in the 1996 issue of the Journal of the American Medical Association by a research team headed by Herbert Needleman, a professor of child psychiatry and pediatrics at the University of Pittsburgh., suggesting that teenage boys with higher than normal lead levels, were more likely to engage in bullying, vandalism, arson, shoplifting and the like, than their non-lead exposed counterparts. Needleman and his team presented an expanded study suggesting much the same thing at a 2000 joint conference of the American Academy of Pediatrics and Pediatric Academic Societies.
The idea of both the Nevin and the Needleman studies is that elevated blood-lead levels cause significant damage to the development of the prefrontal lobe of the brain, leading to a lowering of the IQ, poor impulse control and increased aggression once a kid hits adolescence.
Nevin got on to the whole lead issue after he was hired as an economic consultant by HUD to study the cost of removing lead paint from public housing. While Nevin was calculating the cost, he became curious about the correlation between lead exposure and violent crime.
The study that resulted, published in 2000 in the scientific journal Environmental Research, concluded that lead-exposure rates of children across the U.S. between 1941 and 1986 correlated precisely with national fluctuations in violent crime rates. As Nevin continued to analyze he became increasingly convinced that lead poisoning was having an effect that was too great to be dismissed as merely coincidental.
Of course, nobody’s suggesting that all of violent behavior can be swept away by, shall we say, getting the lead out. Yet, it’s another illustration of the fact that the best and most efficient way to find a “cure” for crime is to start with a good diagnosis—–something, say, a bit more nuanced and complex than “tougher sentences” and “build more prisons.”
True story. I spent this weekend with a friend who dynamites for a living. He was going to bring down an old bridge over a scenic river, but “the friends of the river” protested because the bridge had been painted with leaded paint, and they didn’t want lead getting into the water. So, to make them happy before the bridge was dropped, my buddy had to SANDBLAST the leaded paint off the bridge. Of course, all of the leaded paint blew into the air and fell into the river, but the greenies were happy…and dumb. Maybe lead got into their brains. BTW, when I was a kid I played with mercury and I came out normal…right?
This could be like the studies of children who are spanked are more likely to commit crimes or people who were sent to jail are more likely to commit crimes.
In any case, I am ALL for getting rid of pollution and toxic metals.
Celeste, one of the main problems with this study, and others like it is that is a correlation study. And while it may be (and probably is quite true) that lead may be a causitive factor, the fact that other children exposed to lead in paint etc. do not turn out to be criminals. The culture of the individual has a lot to do with it, someone raised in say South Philly may be more aggressive than say someone raised in Texarkana even if both were exposed to the same amount of lead. Also, I would point out that very close to 100% of all convicted murderers ages 58 to 40 were exposed to higher levels of radiation (due to above ground and atmospheric testing by the US, USSR and France from roughly 1948 to 1968). Now, so was everyone else who didn’t commit murder, but you see the similarity I trust. And of course, there are all those criminals were exposed to lead up till 1977 when it was banned in paint. Lead has been in paint for a long, long time. Question, has anyone done a study that shows a decline in criminality now that paint containing led is banned and those growing up NOT exposed because of the ban?
GM, can’t say I disagree. If it’s causative, it’s only one of many factors. (And speaking anecdotally, I’d put poverty combined with dysfunctional families and childhood physical abuse, wa-a-aaay ahead of lead paint in terms of causal factors. Yet, there are inevitably multiple causes for nearly every effect. Better policing helps.)
About looking at the decline in criminality as well as the rise, that was part of all Nevin’s studies. If you go to the WaPo article I linked in the post (annoyingly, the WaPo folks require registration), there’s a link to some charts that are kinda cool.
I don’t know that the media always reports the results of research well. (And, for the record, I agree that correlation is not causation.) I was taken aback at the assertion, “Sixty-five to ninety percent or more of the substantial variation in violent crime in all these countries was explained by lead.” That assertion suggests a couple of things to me. (1) A claim of X% of the variation tends to reflect a Goodness of Fit test (R-squared, or adjusted R-squared), (2) which is generally associated with a regression analysis – often Ordinary Least Squares, (3) and, one generally reads that X% of the variation is explained by the model – not a single variable within the model. So, I went looking to see if I could find more detail about the research methodology. Google Scholar kicked up a HUD report, but it was apparently written for a non-technical reader. References to “the model” were frequent, as well as a discussion of validating the model, but the model itself wasn’t revealed, nor was the precise methodology revealed – or, at least, in a form that was recognizable to me. The best I could come up with is the abstract below. The notion of “best-fit lags (highest R2 and t-value for blood lead)” would be the key to understanding the claim. That notion isn’t immediately familiar to me, so I toss it to the back of my head, and someday I’ll stumble across an explanation and think, “Oh, yeah.” I do wonder what else might have been in the model. And, wonder what blood lead might proxy for, that isn’t in the model, but might be closely correlated with blood lead. Regardless, it’s an extraordinary finding.
For anyone who wants to pursue it more closely…
http://tinyurl.com/2y9xew
Understanding international crime trends: The legacy of preschool lead exposure
Rick Nevin,
National Center for Healthy Housing, USA
Received 12 August 2006; revised 20 February 2007; accepted 23 February 2007. Available online 23 April 2007.
Abstract
This study shows a very strong association between preschool blood lead and subsequent crime rate trends over several decades in the USA, Britain, Canada, France, Australia, Finland, Italy, West Germany, and New Zealand. The relationship is characterized by best-fit lags (highest R2 and t-value for blood lead) consistent with neurobehavioral damage in the first year of life and the peak age of offending for index crime, burglary, and violent crime. The impact of blood lead is also evident in age-specific arrest and incarceration trends. Regression analysis of average 1985–1994 murder rates across USA cities suggests that murder could be especially associated with more severe cases of childhood lead poisoning.
Keywords: Lead poisoning; Crime; IQ; Behavior; Violence
Environmental Research
Volume 104, Issue 3, July 2007, Pages 315-336
Well, as it would turn out, I was able to get online access to that journal article through my university library’s proxy server – briefly. Asked a bit much of my phone modem internet connection, it choked on all of the graphs when I tried to download it as a pdf 🙁 Should have been content with the online version. Oh, well.
FWIW. Based on what I saw it was a single independent variable model – blood lead – against the dependent variable ‘crime index.’ In some cases (9 nations included) unemployment was included as an independent variable, and while statistically significant, didn’t substantially change the R-squared value when it was excluded. The N’s for some countries seemed small, but you couldn’t argue with p-values so close to zero they might as well be zero. I assume the lag had to do with when blood levels were drawn as compared to the year in which the crime indices were drawn. Had the ‘flavor’ of a time series analysis, but I didn’t hang on to the connection long enough to verify that’s what it was.
Thanks for the update Celeste. This one is kind of interesting to me. Levitt confirms my thought that Nevin’s was a time series analysis. And, takes a small swat at a one-note-song model. ‘Course his own abortion inference was a one-note-song, too. Reyes’ approach is as close as a social scientist can come to using a human being as a guinea pig in this age of the IRB – the ‘natural experiment.’ Something changes in the environment (typically, a change in government policy) that allows you to compare results for an independent variable’s state ‘before the policy change’ and ‘after the policy change.’ It’s as close as economics can come to a the gold standard, double blind study. As for Levitt never having heard of Nevin – yeah, well, whatever. I expect there a number of economists toiling in world that Levitt’s never heard of. In truth, the gimme-that-olde-time-discipline of economics would look askance at any study that doesn’t ground itself in theory as a matter of first principles. You and GM are well on your way to developing a grounded theory. The two of you ought to write it up and propose that Levitt, Nevin, and Reyes’ look at confirming or disconfirming it.
Sorry. I am having an absolute love affair with this post [insert sheepish, gulity grin here].
Hoisted from the NY Times The Opinionator: http://tinyurl.com/rn3uh (sorry, it is behind a paywall)
Listener, your “sheepish grin” is well deserved. Good info and thanks for posting it.
Great link, Listener. Hysterical—in a scary kinda way.
BTW, here’s the link to the whole Bradford Plumer post that the NYT guy is quoting (and no paywall!)
http://plumer.blogspot.com/
Thanks for the link, Celeste. Yeah, almost funny. Almost. You and GM are to be commended for not simply telling me to SHADDUP ALREADY. Expect you both are far too gracious for that.
Moi Gracious? Surely not, after all, I’m a knuclkedraggingneanderthalic conservative (who forgot to use his space bar) 😉
Chuckle.