What Does it Take to Convince Someone?Posted: January 27, 2013
A week ago, I came across this article in Mother Jones that I found completely convincing, not least because it appealed to many of my assumptions and intellectual interests. It contends that much of the spike in crime across the developed world in the 60′s was the result of lead pollution from the use of leaded gasoline in automobiles, and that the sudden drop in crime rates in the 90′s was due to the ban on leaded gasoline twenty years prior. Exposure to elevated levels of lead in childhood negatively affects the development of brain areas involved in decision-making and impulse inhibition, as well as permanently lowering IQ. So, the cohort born between the 40′s and 60′s was exposed to levels of lead in childhood that damaged brain development; these children grew into adults with a higher propensity for crime. As environmental standards eliminated leaded gasoline in the 70’s, people born after were exposed to less lead, and thus had less brain damage and were less likely to commit crime when they reached young adulthood.
The more interesting story to me, however, has been the article’s responses. Kevin Drum, the author, notes that criminologists were loath to consider the lead hypothesis. The scientists cited in the story discuss the dearth of interest their work has received and cite cross-disciplinary barriers (“…because criminologists are basically sociologists, they look for sociological explanations, not medical ones”). Drum suspects special interests and political ideology may explain part of the cold reception, and a romantic view about free-will accounts for another part. Yet for Drum’s article, there has been varying levels of support across the internet and across the political spectrum (see the Guardian and Reason.com).
One of the more critical responses comes from Jim Manzi at the National Review. He raises some methodological issues. One of the studies Drum cites as evidence establishes the causal link between leaded gasoline and crime by looking at when different states imposed bans and when crime rates dropped. It found that 20 years later, the states that first banned leaded gasoline also first experienced declines in crime rates. Manzi is skeptical that there are no systematic differences between the states that banned leaded gasoline earlier and those that banned later. He argues that any number of hidden variables may be correlated with banning leaded gasoline early and crimes rates dropping early (for instance, he gives the example of wealth: richer states may phase out first, but this relative richness also accounts for lower propensity for crime later on). Drum, for his part, does not believe Manzi’s possible hidden variables are plausible causes of the crime rates. Moreover, he points to the larger mass of studies, some of which were international, as evidence that there is something to the lead hypothesis.
I’m also not convinced of Manzi’s focus on systematic differences. Let’s say the states that banned leaded gasoline earlier were indeed wealthier or culturally different (e.g. a culture of honour). Why would the crime rate spike in the 60’s, then decline during the 90’s? If certain states were wealthier than others and this accounted for the different crime rate, you would expect a consistent difference between the states rather than a staggered onset of a decline in crime.
A fight over methodology, however, is not Manzi’s main bone of contention. Kevin Drum used his evidence of the link between lead exposure and crime as an argument for a government program of lead removal in inner cities and in old homes. This is arguably what Manzi finds most objectionable. As befits one of the most prominent conservative journals in America, he cites the spirit of Hayekian doubt over policy intervention. He does not believe econometric regression analyses, even a lot of them, would justify spending public money on an environmental remediation program. As evidence for his skepticism, he points out that randomized field trials for criminology interventions have yielded dismal results. He prefers to err on the side of laissez-faire policies for social and economic problems, and if interventions are used, to change incentives rather than try to change people. This is not purely ideological. He cites the greater effectiveness of unemployment benefit sanctions vis-a-vis job training at reducing unemployment, which has been borne out by at least one study.
So where does this leave policy professionals? Political beliefs are going to dictate the willingness to intervene in a given situation. For someone like Jim Manzi, the most convincing evidence source for a policy intervention are trials similar to the efficacy trials that drugs must go through. In other words, the evidence must reach the gold standard in science. The case of the lead-crime link, however, perfectly illustrates one of the difficulties this view poses. The benefits of lead cleanup will only appear in twenty years. Other interventions like early childhood education or exercise promotion show similar lags. In the meantime there could be a high opportunity cost, especially if there is a significant long-term cost to inaction. Kevin Drum, based on his quick estimation, believes that we would gain much more in the long run from lower crime than it would cost for cleaning up lead. This, of course, does not mean every intervention is worth trying out. Evidence must be marshaled to show why a policy intervention is likely to work, and why it is worth undertaking it rather than another policy or doing nothing at all. From the considerable body of evidence that Drum points to for his argument, I think lead cleanup is definitely one of those interventions.
Wilfrid Chan is 2014 Master of Public Policy Candidate at the School of Public Policy and Governance. He also has a B.Sc in Psychology and Biology from the University of Toronto. His interests include health policy and science policy.