Economists continue to debate the determinants of crime trends. Donohue and Levitt sparked an exciting debate about the role of legalized abortion. An alternative theory is the role of listening to heavy metal music. A third theory points to pregnant women’s smoking habits and the consequences of their heavy smoking for later life outcomes for their children. “Researchers led by Angela Paradis of the Harvard School of Public Health reviewed the health and criminal records of 4,000 American adults between 33 and 40 years old. The men and women were part of a long-term health study in Rhode Island designed to track the long-term effects on children of conditions during pregnancy and around birth. Information was collected about the smoking habits of the mothers, who were enrolled in the study between 1959 and 1966. Children whose mothers smoked at least 20 cigarettes per day while expectant were 30 percent likelier to end up with a criminal record, and were also likelier to be repeat offenders.”
This finding is similar in spirit to Jessica Reyes’ work linking early life lead exposure to later crime rates. She argues that the rise of unleaded gasoline in vehicles starting in the early 1970s played a key role in explaining why crime dropped starting in the early 1990s (once the kids were adults).
In both pieces of research, early life exposure to pollution is posited to be the causal mechanism that increases the probability of committing crimes as an adult. Jim Heckman’s work on human capital and skill formation would further strengthen the logic here as the children of smokers tend to fall behind (low birth weight for these kids see this) and never catch up.
I find such “early life events” papers to be quite interesting. Doug Almond has been one of the leaders in this field investigating later life outcomes of such events as exposure to Chernobyl radiation and living through the Influenza epidemic of 1918. Here are his papers. Economists are useful people and most of us are lovable people (once you get to know us).
“early life exposure to pollution is posited to be the causal mechanism that increases the probability of committing crimes as an adult.”
I hope that they were careful not to commit the fallacy of post hoc ergo propter hoc. Women who smoke while pregnant display a lack of concern for their unborn children, and perhaps they are more likely to retain that lack of concern for their children after birth, which in turn might increase the likelihood of their children turning out to be criminals.
In addition, in the United States, a substantial percentage of criminals are criminals merely because they are addicted to illegal drugs. Therefore, if their mothers were addicted to cigarettes, and if a tendency to addiction is inheritable, then that may contribute to their criminality.
“I hope that they were careful not to commit the fallacy of post hoc ergo propter hoc.”
This is a good point.
You should email Angela Paradis and Steven Levitt and ask them if they have ever heard of this issue before…
If they ignore you, keep emailing until they reply.
Then post their responses in this forum.
Please.
The first hypothesis has to be that these mothers pass on bad genes to their kids, and that it is these genes that are responsible for later criminality, rather than the smoking itself. Cigarette addiction is very strongly correlated with many other harmful traits, including alcohol and drug addiction, and general lack of self-control. Smoking during pregnancy on its own may plausibly lead to low birth weight and increased likelihood of birth defects, but not a massive increase in the likelihood of becoming a criminal.
A free society excludes the possibility of controlled experiments in human behavior, seems to me. In a free society, human behavioral experiments will always have the problem of systematic bias due to sample self-selection. We have natural experiments, with the problem illustrated here. I don’t see that it’s really a problem so far as recommendations goes: take care of yourself, don’t hurt children, avoid impulsive people who accept large future costs to themselves and others for transient immediate benefit to themselves.
Not that I’m complaining, here, but does “Economics” nowadays include anything statistical which deals with behavior (human or otherwise)?
The relevant time period of 1959 through 1966 would, hopefully, reduce the risk identified by the first comment, in that the health effects of smoking were not widely advertised — that might be why they used that time period.
On the other hand, you definitely would have to be sure you were controlling for education and income and perhaps other factors (mental illness?) even in this time period, because smoking among women was not the norm and other factors associated with a woman’s decision to smoke to begin with could explain antisocial behavior — IOW, they weren’t careless about the health of their babies so much as they were antisocial in some relevant way.
According to this article — http://www.nytimes.com/1986/11/21/us/report-asks-states-to-set-smoking-age-at-18.html — 34 percent of American women smoked in the mid-1960s.
Matthew
Thought-provoking post. One of the things that makes me wary of attributing the great 1990s drop in crime rates to prior environmental hazards is the disproportionate share of the national rate drop that was accounted for by the greater new york city area. Murder is a good example because unlike many other crimes it’s hard for agencies to fudge the numbers. From 1993 to 1996, the annual number of murders in the country dropped by about 4,900 (off a base of about 24,500) — huge drop. The drop in the annual number of murders in New York city alone over that period was about 1,000 (1,927 to 986). I used to know all these sorts of numbers well and I don’t any more, but the murder rate data is one example of much the national crime drop was disproportionately an NYC phenomenon. Unless there is some reason to think the environmental risk or change in risk was heavily concentrated there, its hard to explain the data with that hypothesis.
It’s easy to think of reasons why environmental risk is concentrated in big cities, although not so much NYC. Lead poisoning of children comes mostly from peeling paint and (in the old days) auto exhaust. There is more housing stock and auto density in a big city than in a suburb.
Ebenezer: No argument about cities in general, but not all cities had dropping crime in the mid-1990s, and none approached the drop in NYC - there is something unique in NYC that remains to explained (the change in policing strategy perhaps).
Henry, thanks for the link. In the interests of full and fair disclosure, the percentage of men smoking far exceeded the percentage of women.
“The study, mandated under the Comprehensive Smoking Education Act passed by Congress in 1984, found that 30 percent of Americans smoked and that the rates were now only slightly higher among men than among women. In the mid-1960′s. when the Surgeon General of the United States first warned of the health hazards of smoking, 45 percent of Americans smoked, and male smokers significantly outnumbered female smokers.”
As of 1986:
* Thirty-three percent of American men smoke, as against 53 percent in the mid-1960′s.
* Twenty-eight percent of American women smoke, as against 34 percent in the mid-1960′s.
Note that the fall off in smoking is disproportionately among males, so that by the 80s, at least, the rate of smoking between the sexes was almost at parity.
“The first hypothesis has to be that these mothers pass on bad genes to their kids, and that it is these genes that are responsible for later criminality, rather than the smoking itself. Cigarette addiction is very strongly correlated with many other harmful traits, including alcohol and drug addiction, and general lack of self-control. Smoking during pregnancy on its own may plausibly lead to low birth weight and increased likelihood of birth defects, but not a massive increase in the likelihood of becoming a criminal.”
Guy, the traits you point to are also highly dependent on environment. Substance abuse and self-control, as well as a multitude of other dysfunctional behaviors, usually have roots in development and learning. To the degree that we see these behaviors in much higher rates in certain populations, we can see the effects of learning going on that is determined by many social factors - poor parenting being crucial.
They say they ruled out the standard confounders, but you have to wonder. Given the stigma against women smoking in that era, the kids would also have been marked as different by peers’ parents and authorities, with obvious issues for development and socialization.
Eli and no doubt others accept the conventional social science wisdom that “it’s all about how you were raised.” Actually, it’s not. Judith Rich Harris has shown convincingly in her book “The Nurture Assumption” that parenting has at best a small effect on our adult personality. Most of the child development research is worthless, because it can’t distinguish between the genes you inherit from your parents and the parenting you get from them. Just as an example, consider “I grew up in a house full of books, so I love to read.” But it’s just as likely that you love to read because you got your genes from the type of parents who fill their house with books. The only type of research that can reliably distinguish nature from nurture is studying twins separated at birth, and these studies are not kind to the nurture assumption.
No one wants to believe this. I don’t want to believe it. But what the data actually show is that around half of personality is determined by genes, and the other half by something rather mysterious that is probably much more related to peer pressure than to parenting. The practical implication is that, perhaps the single best way parents can influence their children is to move to a neighborhood where the kids will have admirable peers.
Before you start pooh-poohing, read Harris’ book, please.
30% isn’t very much.
Given the uncertainties in the measurables.
Controls for studying the causes of the drop in crime rates in the 1990s are available from other western countries. Crime rates dropped in nearly all western nations during that time period. Many of the factors identified as causes do not stand up to comparison with other places (e.g., the increase in the abortion rate after Row vs. Wade or increased incarceration rates in the U.S.). The causes are complex, but demographics undoubtedly played a major role. New York City is an outlier where huge drops in violent crime rates argues for special factors there. I found that The Great American Crime Decline by Franklin Zimring provided a good overview.
I don’t think there could have been much stigma against women, pregnant or otherwise, smoking in the fifties and early sixties, at least in the New York metropolitan area. My mom smoked — a lot — back then, and so did many of my aunts, and many of the women who would turn out to be my friend’s moms, once I got old enough to have friends.
The husbands smoked too, with the main difference being they also got to smoke cigars and pipes (when we had to bring cigar boxes to school for a second grade make-a-musical-instrument project, there was no question that most everyone would have one available at home).
We were all solidly middle-class and the moms all did their best to take good care of us, both pre- and post natally. When it was announced that smoking was bad for your health, some of the adults around me quit cold-turkey. The others were too addicted. No one in my circle in my generation took up the habit.
I’ll add, for what it’s worth, that neither I nor my siblings nor any of our cousins turned out to be criminals, unless maybe someone hasn’t been caught yet. The youngest cousins are 51, so that seems unlikely. There does seem to be a fair amount of asthma among us, though.
Regarding nature versus nurture and the effects of parenting. There is a very good article in the New Yorker on the effects of traumatic experience in early childhood, most of which are due to bad parenting. It’s not contradictory to say that most parents have little impact but that repeated traumatic experiences in the form of abuse, foster parenting arrangments and so on can have a lasting negative impact. That’s what the article was about: how childhood trauma can have an impact that is severe enough to cause long-term changes even at a cellular level (e.g., immune system, neurological responses to stress, and so on).