Lead and violence: all the evidence

Kevin Drum offers a response to a recent meta-study on the link between lead and violence, blogged by me here.

I’ll take this moment to explain why the lead-violence connection never has sat that well with me.

Let’s say we are trying to explain why 2022 America is richer than the Stone Age.  We could cite “incentives, policy, and culture,” noting that any accumulated stock of wealth also came from these (and possibly other) factors.  You might disagree about which policies, or which cultural features of modernity, and so on, but the answer to the question pretty clearly lies in that direction.

Now let us say we are trying to explain why America today is richer than Albania today.  You would do just fine to start with “incentives, policy, and culture.”  You could add in some additional factors, such as superior natural resources, but you would be on the same track as with the Stone Age comparison.  You would not have to summon up an entirely new theory.

Why is Nashville richer than Chattanooga?  Again, start with “incentives, policy, and culture,” noting you might need again supplementary factors.

Broadly the same theory is applying to all of these different comparisons.  Across time, across space, across countries, and across cities.  There is something about this broad unity that is methodologically satisfying, and it helps confirm our view that we are on the right track in our inquiries.

Now consider the lead-crime connection.  Insofar as you elevate the connection as very strong, you are tossing out the chance of achieving that kind of unity.

Why was violent crime so often more frequent in earlier periods of human history?  It wasn’t lead, at least not for most periods, perhaps not for any of the much earlier periods.

Why was there more peace in Ethiopia five years ago than in the last few years?  Again, whatever the reasons it wasn’t a change in lead exposure.

Why is the murder rate in Haiti today much higher than during the Duvaliers?  Again, no one thinks the answer has much to do with changes in lead exposure.  Mainly it is because political order has collapsed, and the country is ruled by gangs rather than by an autocratic tyranny.

How about the violence rate in the very peaceful parts of Africa compared to the very violent parts?  Again, lead is rarely if ever going to be the answer to that one.

So we know in the true, overall model big changes in violence can happen without lead exposure being the driving force.  Very big changes.  In fact those big changes in violence rates, without lead being a major factor, happen all the time.

And many of those big changes are mysterious in their causes.  It really isn’t so simple to explain why different parts of Africa have different murder rates, often by very significant amounts.  You can hack away at the problem (e.g, Kenya and Tanzania have very different histories), but there is no simple “go to” theory.  Furthermore, since both violence and peace often feed upon themselves, in a “broken windows” increasing returns sort of way, the initial causes behind big differences in violence outcomes might sometimes be fairly slight and hard to find.

That to my mind makes “the true model” somewhat biased against lead being a major factor in changes in violence rates.  In the broader scheme of things, lead exposure seems to be a supplementary factor rather than a major factor.  It doesn’t rule out lead as a major factor, either logically or statistically, if you wish to explain why U.S. violence fell from the 1960s to today.  But the true model has a lot of non-lead, major shifts in violence, often unexplained or hard to explain.

Addendum: I am also surprised by Kevin’s comment that there isn’t likely to be much publication bias in lead-violence studies.  I take publication bias to be a default assumption, namely the desire to show a positive result to get published.  That hardly seems unlikely to me at all.  And in this particular case there is even a particular political reason to wish to pin a lot of the blame on lead exposure.  Correctly or not, people on the Left are much more likely to elevate lead exposure as a cause of social problems.

And to repeat myself, just to be perfectly clear, it strikes me as unlikely that the effect of lead exposure on violence in zero is the last seventy years of the United States.


Comments for this post are closed