Lead and Crime: Is Correlation Also Causation?

membership ticker

Autocratic forces are ascendant.

Truth-telling independent media is one remaining bulwark against an autocrat’s unrestrained exercise of power. At a time when billionaire owners of corporate media are making accommodations to political leaders, our nonprofit newsroom cannot be bought, bent, or broken. Depend on it, and please stand with us.

Autocratic forces are ascendant in America. But truth-telling independent media is one remaining bulwark against the unrestrained exercise of power. At a time when billionaire owners of corporate media are making accommodations to political leaders, our nonprofit newsroom cannot be bought or broken. Please stand with us.

Fight disinformation: Sign up for the free Mother Jones Daily newsletter and follow the news that matters.

A big chunk of the evidence linking gasoline lead emissions to violent crime rates is based on statistical analysis. This naturally leads to the criticism that correlation is not causation, so we don’t really know if lead caused crime rates to go up and down. If you’re curious to learn more about this, Rick Nevin has just posted a short paper titled “Lead and Crime: Why this correlation does mean causation.” Here’s a small excerpt that addresses just the statistical evidence:

The key statistical issue that needs to be addressed by the correlation-never-means-causation crowd is whether they honestly believe that:

The observed association between lead used in paint and USA murder rates from 1901 to 1960 with a time lag close to the peak age of homicide offending was a coincidence;

The association between USA gasoline lead and violent crime from 1964-1998 with a similar time lag was another coincidence;

The “experimental evidence” from violent crime since 1998 (including a 45% drop in the juvenile violent crime arrest rate from 1998-2011) tracking earlier trends in lead exposure is a coincidence;

Analysis of crime in nine nations shows the same consistent relationship between lead exposure and crime trends through 2002, with statistical best-fit lags that reflect the peak age of offending for each crime category, by coincidence;

This consistent relationship within every nation studied happens to explain otherwise bewildering changes over time in USA and Canada crime rates relative to Britain, France, and Australia, by coincidence;

Experimental evidence from international crime trends since 2002 tracking earlier trends in lead exposure in every nation is also a coincidence.

There’s much more at the link, all based on a 9-part test proposed in 1965 by Austin Bradford-Hill for distinguishing mere correlation from true causation. Nevin examines a broad range of evidence, including statistical studies, longitudinal studies, medical studies, imaging studies, and more. It’s worth a look if you’re still skeptical about the lead-crime connection.