January 25, 2017

Lead, Race, and Crime

In this article, I will review the literature linking lead exposure to criminality and then discuss what role lead plays in the American Black-White crime gap. Data on lead exposure can be divided into two categories: data showing how lead levels relate to regional crime, and data showing how lead exposure relates to individual criminality. First, let’s look at regional data. 

Regional Data

Stretesky and Lynch (2001) looked at the association between air lead levels and homicide rates in 3,111 US counties. Controls included poverty, % black, educational attainment, and whether or not a county was in the south. Going from the county with the least lead to the one with the most was associated with an increase in homicide by a factor of 14 without controls and 4.1 with controls. However, the 95% confidence interval of this estimate stretched from 1.09 to 16.615, suggesting that a very wide range of effect sizes are plausible given the data. 

Stretesky and Lynch (2004) looked at how air lead levels predicted crime across 2,772 US counties. Their controls included a measure of general air pollution, % black, poverty, income inequality, whether or not a county was in the south, the unemployment rate, the divorce rate, and the average crime rate of the 5 nearest counties. Lead exposure had a significant relationship with violent crime with an effect size of .164 and property crime with an effect size of .319. It was also found that lead explained more variation of crime among poorer counties.

Feigenbaum and Muller (2016) used data on 545 cities and found that the installation of lead pipes in the 1880’s predicted homicide rates in the years 1921 to 1936. The use of lead water pipes was common at this because it was more durable and more malleable than iron pipes. This resulted in lead being in city water which in turn caused people to have much higher blood lead levels (BLL) than they do today.

The paper found that a city using lead pipes and its distance from a lead refinery were both positively related to its homicide rate even after controlling for home ownership rates, % black, sex and age demographics, latitude, longitude, and population density. % Black was also a significant predictor even with lead and the other covariates held constant.

Nevin (2007) found a very strong association between the blood lead level of preschool children and subsequent crime rate trends 19 years later when those preschoolers would have become young adults. This association was found over several decades in the USA, Britain, Canada, France, Australia, Finland, Italy, West Germany, and New Zealand. The range of years looked at varied by nation, but in some countries it ranged all the way from the 1930s to the 1990s.

Nevin also reported on a cross-sectional analysis which looked at what variables predicted the murder rate of 124 US cities between 1985 and 1994. Nevin found that the percent of the population with lead poisoning (LP) was a significant predictor. LP explained 14% of the variance in murder rates. However, the percent of a city which was Black was found to be a stronger predictor and when it was controlled for LP no longer had a significant association with a city‘s murder rate.

In totality, the regional data clearly shows an impressive cross-sectional and longitudinal relationship between lead exposure and crime which persists even after controlling for a large number of potentially confounding variables.

 Individual Level Data

Now let’s look at studies on crime at the individual, rather than reigonal, level. Stretesky and Lynch (2004) described a study published in book format thusly:

“Deborah Denno’s (1990) longitudinal study of the link between lead and crime among African American males, which controlled for many other potentially influential independent variables (including environmental and social factors, such as parents’ income and occupation),found that early childhood lead poisoning was one of the most important predictors of three types of delinquent outcomes: (1) disciplinary problems from ages 13–14; (2) juvenile delinquency from ages 7–17; and (3) the number of adult offenses from ages 18–22.”

Marcus et al. (2010) meta-analyzed the relationship between lead exposure among children and adolescents and conduct problems across 19 studies with an aggregated 8,561 participants and found a statistically significant correlation of .19.

Thus, there is a clear relationship between lead exposure and anti-social behavior at both the regional and individual level. However, given the increasing rarity of lead exposure, lead exposure may not play a very large role in explaining why it is, in modern society, that some people are more anti-social/criminal than others.

 Racial Differences in Lead Exposure

In America, African Americans have higher BLLs than do Whites and Hispanics during childhood. White et al. (2015) reviewed 5 national samples taken between 1998 and 2004 and found that African American children had a mean BLL about 1.4 ug/dl higher than non-Hispanic White American children did.

More recent data from the CDC shows that in 1999-2004 Black children had a mean BLL which was 1 ug/dl higher than White children and that this BLL gap shrank to 0.9 ug/dl in samples taken between 2003 and 2006, and to 0.5 ug/dl in samples taken between 2007 and 2010.

Interestingly, both studies also looked at White/Hispanic children differences in BLL and both found that there were no consistent differences between these groups.

Black/White differences in BLL during childhood are real, but slight. Their size can be exaggerated by looking at what proportion of each race has a BLL which exceeds some arbitrary threshold. Consider, for instance, this chart from Drum (2016):


That these differences seem larger is not surprising. Small mean differences lead to large differences at the tails of distributions.  However, there is no justification for the thresholds used here. In fact, these thresholds are so arbitrary this one chart uses three different standards. This reflects changes in the CDC guidelines about safe levels of lead. For decades, the CDC has been lowering this “safe threshold” as better research has come in with larger samples and more precise measures which have allowed researchers to document the damage that lead does even at very low levels.

Furthermore, this slight racial gap in blood lead levels diminishes to virtually nothing by adulthood. CDC numbers [taken from Tsoi et al. (2016) and Pricket et al.(1994)] show that BLL levels no longer differ significantly by race among the total population.

 BLL by Race.png

Given this, even though lead impacts crime, the fact that the races barely differ in terms of lead exposure as children and not at all as adults suggests that lead probably plays little to no role in the American Black-White IQ gap. This conclusion is consistent with Feigenbaum and Muller (2016) and Nevin (2007) who found that the proportion of an area which was Black continued to predict its crime rate even after its degree of lead exposure was controlled for. On the other hand, lead exposure is more common in the developing world and more play a role in the high crime rates of nations in Africa.

Facebook Comments