Like rain on election day, crime is good for the Republicans. Whenever crime seems to be increasing, significant numbers of Americans tend to blame liberal permissiveness and turn to conservative political candidates, partly because they endorse a sterner approach to raising children, policing the streets, and punishing criminals, and partly because they oppose government “giveaways” to the poor, blacks, and other groups that commit a lot of crimes. While orthodox liberals answer that “getting tough” won’t really help and that the way to reduce crime is to make society more just and opportunity more equal, this response to crime has seldom moved the electorate. When crime rates rise, liberals almost always find themselves on the defensive.
The political effect of crime on the public may be the result of an intellectual mistake, but if so it is an understandable one. Modern liberalism is a product of the eighteenth century, and as its name suggests, its most consistent and powerful impulse has been to expand personal liberty, or, as we often say today, “opportunity.” In recent decades American liberalism has been primarily concerned with making sure that minorities, women, the poor, and other disadvantaged groups have the same opportunities as affluent white males, so it has acquired an increasingly egalitarian cast, but its strongest impulse is still to eliminate constraints and provide people with more choices.
Liberals have traditionally hoped that more freedom would lead to more of almost everything else they valued. Many Americans viewed the 1960s and early 1970s as a test of this hypothesis. Restrictions on personal behavior diminished dramatically during this period, altering everything from sexual habits and hair styles to relations with the police and employers. Deference to authority in all its guises also declined, making people feel they had more choices.
Among blacks, the end of de jure segregation and the advent of affirmative action opened even more opportunities. Black college-entrance rates almost doubled between 1960 and 1975, and large numbers of young black adults moved into professional and managerial jobs for the first time. Yet for blacks as for whites, the most important change was probably subjective rather than objective. The influence of both black and white authority figures declined precipitously during this period, leaving young blacks with the feeling that there were no clearly defined limits on the choices open to them. They could become anything—or nothing.
In the early 1960s most liberals had hoped that this kind of liberalization would make us all feel a stronger sense of solidarity (or “fraternity”) with one another. Once our own rights were more fully recognized, we were supposed to become more attentive to the rights of others. Increasing affluence and opportunity were also supposed to give those at the bottom of the social pyramid a stronger feeling that they had something to lose if they broke the law. The liberal innovations of the years from 1960 to 1975 were therefore supposed to reduce the frequency of murder, rape, assault, robbery, and burglary. Instead, all these crimes became far more common. Rightly or wrongly, many Americans concluded that increased liberty, especially for the poor, had actually caused the decline in “fraternity.”
Public concern about crime is obviously selective. The mere fact that some behavior is illegal does not worry most Americans much. Every year, for example, millions of Americans defraud the Internal Revenue Service by underreporting their income or overstating their deductions. The amounts stolen in this way almost certainly exceed the amounts stolen by muggers on the streets. Yet very few Americans view tax fraud as a serious threat to themselves or to the republic. The reason seems obvious. Unlike robbery, tax evasion has no individual victims. It forces the rest of us to pay higher taxes than we otherwise would, but it does not create the same kind of fear or the same sense of personal violation as being raped or even having your house burgled. We react to most other white-collar crimes with equal indifference. Given a choice, almost everyone would rather be robbed by computer than at gunpoint. This does not make white-collar crime morally preferable to blue-collar crime, but it does explain why white-collar crime is not a major political issue.
The politics of criminology mirror the politics of crime. Any complete account of why people commit crimes must include two elements: a description of how criminals differ from noncriminals and an explanation of how they got that way. In most cases, however, conservative criminologists emphasize the psychological differences between criminals and noncriminals, while liberals emphasize the social circumstances that produce these psychological differences. Indeed, extreme liberals sometimes argue that there are no stable psychological differences between criminals and the rest of us, and that criminals are just ordinary people who find themselves in especially trying or especially tempting circumstances. The extreme conservative position, in contrast, seems to be that social circumstances have nothing to do with crime; some people are just born rotten.
The political motives behind these differing approaches to crime are obvious. Traditional moral reasoning holds us responsible for our psychological states but not, in most cases, for the circumstances in which we find ourselves. Attributing crime to psychological deficiencies is thus compatible with holding criminals morally accountable for their acts, while attributing crime to the circumstances in which criminals find themselves is to some degree morally exculpatory.
Portraying crime as a product of psychological deficiencies also encourages us to place the ultimate blame on parents rather than on society. If we hear a social scientist say that black teen-age muggers are unusually “aggressive,” “egocentric,” or “impulsive,” for example, we instinctively impute these defects of character to poor upbringing. Since Americans see upbringing as the responsibility of the family rather than society, the ultimate villains more often turn out to be the mugger’s parents (who are also black), not the federal government, capitalism, white racism, or other targets of liberal reform.
James Q. Wilson has had a major part in discrediting liberal ideas about crime among both scholars and public officials over the past fifteen years. Today he is probably the most influential single writer on crime in America. In 1977 he began teaching a course on crime at Harvard with Richard J. Herrnstein, an experimental psychologist whom the left had bitterly attacked in the early 1970s for his views on the role of heredity in determining IQ scores and economic success. Wilson and Herrnstein have now turned their lecture notes into a book, Crime and Human Nature, which tries to summarize what we know about criminal behavior. Unlike Wilson’s earlier books on crime, this one is both discursive and inconclusive; indeed, it reads more like a textbook than a trade book. Nonetheless, it has become controversial, largely because it argues that genetic variation helps to explain why some people commit serious crimes while others do not.
Most liberals (and radicals) are instinctively suspicious of any claim that genes influence human behavior. In part, no doubt, this reflects the fact that we know almost nothing about how genes influence behavior. Indeed, we know very little about how they influence physical development. Confronted with the DNA of a dinosaur, for example, geneticists could not even tell you whether it would grow up to be large or small, much less how it would behave. But liberal resistance to genetic explanations is more than just a matter of scholarly caution. Genetic explanations of human behavior arouse opposition because we think they imply that undesirable behavior is a product of forces society cannot control. If we are told that genetic variation explains much of the variation in children’s performance on cognitive tests, for example, we take this to mean that schools can do very little to improve slow learners’ performance on such tests. In the same way, if we are told that criminal behavior is traceable to genetic differences between criminals and the rest of us, we treat this as an assertion that some people are born criminals and that there is nothing either their parents or society can do to make them better citizens.
This response to genetic explanations of human behavior, while nearly universal, is fundamentally wrong. Wilson and Herrnstein recognize that it is wrong, but they nonetheless say very little about the ways in which social institutions can alter genes’ effects on behavior. As a result, their book has helped to perpetuate liberal opposition to genetic explanations. In what follows I will argue that such opposition is misguided. Since I will use Wilson and Herrnstein’s genetic explanations of crime as examples, I may leave the reader with the impression that Crime and Human Nature is devoted exclusively to such matters. In fact, two thirds of the book has nothing to do with genes, and even the remaining third concerns traits that have nongenetic as well as genetic determinants. Nonetheless, it is the genetic argument that has made the book controversial.
In order to illustrate both the basic logic of genetic explanations and their potential pitfalls it is helpful to begin with an example in which the causal links are relatively clear, such as hair length. In our society variation in hair length is largely attributable to the fact that some people have their hair cut shorter than others. In most cases, moreover, men cut their hair shorter than women. This means that if you are born with two X chromosomes your hair usually ends up longer than if you are born with an X and a Y chromosome. In a statistical sense, therefore, the presence or absence of a Y chromosome predicts much of the variation in hair length—let us say 60 percent. But the fact that genes currently predict hair length fairly accurately tells us nothing about society’s ability to alter hair length. If men and women became convinced that equal hair length was important, achieving this result would be no harder than making hair length equal among males alone. Likewise, if some zealot decided that shorter (or longer) hair would contribute to human happiness, he would be a fool to abandon his campaign simply because someone pointed out that people’s genes currently “explained” 60 percent of the variation in hair length. And if some social scientist read a study showing that genes explained 60 percent of the variation in hair length, he would be an even greater fool to conclude, as many now do, that environmental influences explained only 40 percent. Environmental variation (in the way people have their hair cut) would explain virtually all the variation in hair length, despite the fact that genetic variation explained 60 percent.
Similar problems arise when we try to analyze the effects of genes on crime. As the authors of one of the most important empirical studies of heredity and crime point out:
There are no genes for criminality, but only genes coding for structural proteins and enzymes that influence metabolic, hormonal, and other physiological processes, which may indirectly modify the risk of “criminal” behavior in particular environments.1
Two seemingly innocuous points in this formulation require special emphasis: the notion that genes exert their effects “indirectly” and the notion that their effects depend on the “particular environments” in which individuals find themselves. The effects of gender on crime illustrate both how genes’ effects can be indirect and how they can vary considerably from one particular environment to another.
As Wilson and Herrnstein point out, men are five to ten times as likely as women to commit almost every crime on which American society keeps records. Men also commit more crimes than women in all other societies that keep records, though the magnitude of the difference varies somewhat. This statistical association obviously means that an individual’s genes affect his or her chances of committing a crime. But the important question is why this happens. Does a Y chromosome exert physical effects that somehow make men less law-abiding than women no matter how society treats them? Or does having a Y chromosome make us commit more crimes simply because it alters the way society expects us to behave? Wilson and Herrnstein are reticent on this question.
Most feminists argue that sex differences in crime, like all sex differences in social behavior, are a product of culture rather than of physiology. This is a false distinction. It is certainly true, for example, that human societies have traditionally tolerated or even encouraged more aggression among males than among females, but this cultural pattern itself demands an explanation. It is not enough to say that male aggression is part of a systematic pattern of male domination. We must ask why male domination has been universal and female domination essentially unknown.
One possible explanation is that men are usually bigger and stronger than women. This difference appears to arise largely because having a Y chromosome influences physical growth, though in many societies social norms accentuate the difference by encouraging men to exercise more than women. Because of these physical differences, men have been able to beat or overpower their wives without much risk of retribution. To survive in such a world women have had to accommodate rather than confront, seduce rather than overwhelm. That is what daughters have been taught to do the world over. Parents seem especially likely to emphasize these virtues in cultures where domestic violence is commonplace. Liberated women who treat their spouses as equals can survive only in cultures where men do not exploit their physical advantages.
Differences in size may not be the only reason we tolerate more aggression among boys than among girls. There has been a lively debate in recent years, for example, about whether hormonal differences between men and women contribute to male aggressiveness, independent of their effects on size and strength.2 My point is not to defend any particular physical explanation of cultural norms, but merely to argue that cultural support for the prevalence of male rather than female aggressiveness is so nearly universal that it can hardly have arisen simply by chance. If male aggression and violence are not just historical accidents, they must be directly or indirectly attributable to the genetic difference between the sexes.
Certainly culture also has a life of its own, which is only partially constrained by physiology. Cultures usually encourage men and women to act in ways that accentuate the effects of physical differences, but this is not a universal law. Suppressing physical aggression among young males is often construed as an attempt to “sissify” them in America, for example, but many European and Asian societies do this routinely. Most European and Asian societies also have fewer violent crimes than we do, though there may be other explanations for this, and the pattern is neither uniform nor consistent over time.
Wilson and Herrnstein recognize that a Y chromosome can have social as well as physical consequences, but they largely ignore the possibility that other genes have equally complex and ambiguous effects. Their discussion of race is a striking example, both because it is a sensitive topic about which I would expect them to be especially cautious, and because race is widely regarded as similar to gender.
As Wilson and Herrnstein note, blacks currently account for about half of all arrests for rape and murder and two thirds of all arrests for robbery in the United States, even though they constitute less than one eighth of the population. Since about two thirds of all robbery victims also say that their assailant was black, we cannot blame these arrest statistics on police prejudice. The conclusion that blacks are five to ten times as likely as whites to commit most violent crimes is almost inescapable. This means that the genes determining skin color are as closely correlated with criminal violence in the US as the genes determining gender. Once again the question is why. The traditional liberal view has been that your skin color affects your behavior by affecting the way others treat you, which then affects the way you treat others. Conservatives often suspect that skin color is a proxy for other unidentified physiological differences that somehow make blacks less law-abiding than whites, though they seldom say this in print.
Crime and Human Nature devotes an entire chapter to assessing various explanations of black crime. The authors begin with a discussion of what they call “constitutional” differences between blacks and whites. A “constitutional” difference between two individuals or groups, they tell us, is “present at or soon after birth” and its “behavioral consequences appear gradually during the child’s development.” They give three examples of differences between blacks and whites that they think might be “constitutional” in this sense: differences in body type, differences in personality, and differences in IQ scores. They conclude that while such differences exist, there is no way of knowing whether racial differences in personality traits or IQ scores are “heritable.” By this they mean that there is no way of knowing whether such differences would persist in a colorblind world.
Unfortunately, Wilson and Herrnstein ignore the most obvious—and most obviously heritable—constitutional difference between blacks and whites, namely physical appearance. We don’t know for sure that appearance affects criminality, but the possibility surely deserves serious consideration. Professor Sandra Scarr is currently studying the social adjustment of black children adopted into white families, for example. Almost all these children grew up in white neighborhoods and attended predominantly white schools. If constitutional factors don’t matter, these children should commit the same number of crimes as whites adopted into similar families. If black adoptees commit more crimes than white adoptees, constitutional factors must matter in some way.
Scarr has not yet published her research but let us assume, for purposes of illustration, that she finds high crime rates among black adoptees. This could have at least three explanations. One possibility is that parents, teachers, and friends treat black adoptees differently from white adoptees, and that black adoptees turn to crime as a result. A second possibility is that people treat black and white adoptees the same way, but that blacks nonetheless see themselves as “different,” seek out black friends, emulate these friends’ behavior, and end up in trouble as a result. A third possibility is that people treat blacks and whites similarly, but that blacks react differently for some physiological reason.
The first of these explanations is compatible with traditional liberalism. The last is compatible with scientific racism. Yet in all three cases the difference between black and white adoptees’ behavior would be “constitutional” in Wilson and Herrnstein’s sense of the term, since it would be a direct or indirect consequence of physical differences that are “present at birth” and whose “behavioral consequences appear gradually during the child’s development.” Unfortunately, nothing in Crime and Human Nature warns the reader that a “constitutional” explanation of black crime could also be a social explanation. Wilson and Herrnstein treat the two kinds of explanation as if they were mutually exclusive.
The examples of gender and race suggest that heredity and environment are not mutually exclusive explanations of human diversity, since genes can influence behavior by influencing the environment. Most of us recognize this ambiguity in the cases of gender and skin color, but we tend to assume that if other genes affect human behavior they must exert their effects directly, rather than influencing the environment individuals encounter. If a study shows that genetic differences among white males are associated with differences in crime rates, IQ scores, or alcoholism, for example, we assume that this is because genetic variation causes variation in the way white males respond to the same environment, not because white males with different genes are treated differently. As we shall see, this assumption is unwarranted. The ambiguities that plague efforts to disentangle the contributions of heredity and environment to differences between males and females or between blacks and whites recur in virtually every other example of genetic influence.
If we set aside race and sex, how much of the variation in criminal behavior do genes appear to explain? The best data come from a study of Danish boys adopted by nonrelatives between 1927 and 1947. Almost all these boys were adopted into families where neither parent had ever been convicted of a crime. Of the 3,718 boys adopted by apparently law-abiding families, one third had a biological parent—usually the father—who had been convicted of one or more crimes, and one ninth had biological parents with three or more convictions. Table A shows what had happened to these boys at the time of the follow-up.
The more often the biological parents had been convicted, the more often their sons had been convicted.3
There have been three other studies of adopted children, including a fairly large study in Sweden. All suggest that sons resemble their natural fathers more than we would expect if genes had no connection with criminality. Such studies are not entirely conclusive. To begin with, adopted children have usually had some contact with their natural parents, or at least with their natural mother, prior to adoption. The authors of the Danish study report, however, that the age at which children were adopted did not affect the degree of resemblance between them and their biological parents. This suggests that early childhood contact with the natural parents was not critical, perhaps because such contact was usually with the natural mother, and it was usually the natural father who had a criminal record.
A second potential problem with adoption studies is that adoption agencies often place children from advantaged backgrounds in advantaged homes, creating artificial resemblance between biological and natural parents. The authors of the Danish study report that statistical adjustment for this kind of socioeconomic matching did not appreciably change their results. The psychologist Leon Kamin has argued that subtler forms of selective placement might conceivably explain the Danish findings, but the constraints under which the Danish adoption officials worked make this seem unlikely.4
The most plausible explanation of the adoption data is therefore the one that Wilson and Herrnstein accept, namely that adopted sons resemble their natural fathers because adopted sons and natural fathers have half their genes in common, and genes exert some influence on the likelihood that both fathers and sons will be convicted of crimes.5
While the adoption studies certainly suggest that genes matter, they do not suggest that genes matter very much. To begin with, genes seem mainly to influence men’s chances of committing minor offenses. There is no solid evidence that, if we set aside skin color, men’s genes affect their chances of committing violent crimes. Even for less serious offenses, genetic influences are comparatively unimportant. One way to assess genes’ importance is to ask how closely the behavior of adopted sons resembles that of their natural parents. If we do this we find that natural parents’ criminal records account for only about 2 percent of the variation in their sons’ criminal records.6
Another way to assess the importance of genes is to compare the effects of having a criminal parent to the effects of various sociological determinants of criminal behavior. Table A shows that children of repeat offenders committed 2.4 times as many crimes as children of nonoffenders. Wilson and Herrnstein report that adolescents enrolled in poorly run London comprehensive schools have crime rates three times those of students from similar backgrounds who enrolled in well-run schools. The crime rate in American cities and suburbs is also about three times that in rural areas, and the crime rate for America as a whole in 1980 was roughly three times that in 1960.7 These comparisons suggest that genetic effects are neither trivial nor crucial. When we compare the effects of heredity to the effects of growing up in one country rather than another, however, genes look relatively unimportant. The murder rate in New York City in the late 1960s was at least twenty times that in Madrid, Dublin, Paris, or Brussels, for example. And the murder rate in the United Statesas a whole was about five times that in Australia, despite the fact that a significant fraction of the Australian population was descended from English convicts.
While adoption studies certainly suggest that a man’s genes have some influence on the number of crimes he will commit, they do not tell us why this is the case. In this respect they are logically analogous to studies showing that gender or skin color influences criminal behavior. The existence of such a statistical association is intriguing, but it does not have any practical meaning unless we know why it arises. At present, we don’t.
The association between body type and crime, with which Wilson and Herrnstein begin their discussion of the “constitutional” determinants of crime, illustrates the importance of identifying causal mechanisms. Wilson and Herrnstein report a series of studies showing that “heavy boned and muscular” mesomorphs are overrepresented in American prisons, while stringy ectomorphs and flabby endomorphs are underrepresented. They do not discuss the extent to which body type depends on heredity, but everyday experience suggests that the influence cannot be negligible. It seems to follow that genes influence criminal behavior partly by influencing body type.
How can this be? Wilson and Herrnstein assume that body type cannot influence criminal behavior directly, so they look for other factors that could be correlated with both. Their candidate is personality. Mesomorphs, they suggest, are likely to be dominated by “the spirit of unrestrained, impulsive self-gratification.” This may conceivably be true, but the studies they cite certainly do not convince me of the fact. In any event, the association between body type and crime may have a much simpler explanation they do not consider.
American society makes relatively little collective effort to discourage physical aggression among young males. Pre-adolescent boys who are larger and stronger than average may therefore find that they can “throw their weight around” without much fear of collective retribution. Some of these boys—perhaps especially those with low IQ scores—are likely to find that violence, or a threat of violence, is the best way of getting what they want from others. A stringy ectomorph or a flabby endomorph is less likely to find that violence pays, so he is more likely to develop other strategies for getting what he wants. Given these differing incentives, Wilson and Herrnstein’s own model of the determinants of crime predicts that mesomorphs should be especially tempted to become playground bullies, then to be drawn into teen-age street gangs, and eventually to end up in prison.
While such a linkage between body type and crime seems natural, it is obviously not inevitable. Only a small percentage of mesomorphs, even from poor families, become criminals. If body type affects criminality by affecting the costs and benefits of aggressive behavior, moreover, it should have less effect in societies where adults consistently punish physical aggression among schoolboys. (Japan and France come to mind as possible examples.) The same should also be true in any American school that consistently punishes physical aggression. Body type should also make less difference for girls, among whom physical aggression is far less acceptable than it is among boys. I know of no effort to test theories of this kind. Scholarly research remains divided between those who believe that biology is destiny and those who prefer to ignore it. Even Wilson and Herrnstein, who are interested in both the biological and the social determinants of behavior, have very little to say about how the two interact.
The association between heredity, IQ, and crime provides a more important and more fully documented example of the way genes might influence criminal behavior. Wilson and Herrnstein believe that criminal behavior depends partly on IQ, and that IQ scores depend partly on heredity. It obviously follows that criminal behavior must depend partly on heredity. Yet while the premises and logic of this syllogism seem to me impeccable, the conclusion need not imply what most readers are likely to think it does.
Before discussing the links between heredity, IQ, and crime, I must say something about the vexed question of what IQ tests measure. Despite the claims of some early twentieth-century psychologists, “intelligence” is not a fixed, one-dimensional trait. Nor does it mean the same thing in all cultures. All modern societies label people “intelligent” if they are good at solving problems that require thought. But since different kinds of thought require different skills, people who are good at solving one kind of mental puzzle may not be good at others. This means we can define and measure intelligence only if we can agree on which sorts of thinking we value most. In a society dominated by engineers, intelligence would be largely associated with quantitative skills. In a society dominated by lawyers or theologians, it would mean verbal skills. In a society dominated by politicians and diplomats, it might mean skill at figuring out what other people think and want. In our society it means all this and more. Measuring “intelligence” is thus no different from measuring how much people know about medieval history or molecular biology: it depends on social convention. A useful test is one that accurately mimics the demands that some particular set of social conventions makes on us.
Alfred Binet designed the first IQ tests to estimate the ability of French children to do schoolwork, so he included many problems that demanded the kinds of skills that schools teach and reward. Many IQ tests now contain other kinds of problems as well, but an IQ test’s power to predict performance still depends on its resemblance to schoolwork. The Wechsler Intelligence Scale for Children (WISC), for example, has a “verbal” and a “performance” section. Despite its name, the “verbal” section includes not only a vocabulary test and a test of verbal analogies but tests of arithmetic reasoning, general information, and ability to solve hypothetical problems. Children’s scores on this section of the WISC tend to predict a multitude of things, including how long they will stay in school and whether they will end up in prison. The “performance” section, which bears less resemblance to schoolwork, adds almost nothing to the predictive power of the “verbal” section. For practical purposes, therefore, the WISC is just a measure of academic skills. Whether we should call it an “intelligence” test is a political question. My own preference would be to abandon the term, since IQ tests seem to me to define intelligence far more narrowly than adult society does. But whatever we call the test, we need to recognize that it has no magical significance. IQ scores provide essentially the same information as a battery of conventional achievement tests covering vocabulary, reading comprehension, arithmetic reasoning, and general information.
The claim that genetic differences account for much of the variation in IQ scores is even more controversial than the claim that such scores measure intelligence. By the late 1930s studies of adopted children had provided strong evidence that genes influenced IQ scores. But these studies threw no light on how genes exerted their influence. Both hereditarians and their critics almost always assumed that if genes affected IQ scores they must do so by affecting the physical capacity to learn, independent of environmental influences. But there were—and are—a multitude of other possibilities. Since no one could identify most of the genes that affected IQ scores, much less specify how they exerted their influence, the adoption studies proved less than either hereditarians or their critics assumed. Such studies suggested that the causal chain leading from genes to cognitive skills was a fairly strong one, but they told us nothing about how it worked.
The Nazis’ abuse of genetic theories put an end to public discussion of hereditary differences in human beings for a generation. Throughout the 1950s and 1960s American liberals talked and wrote as if IQ and achievement scores varied for purely environmental reasons. Arthur Jensen revived public debate about the question in 1969 when he published a controversial article in the Harvard Educational Review summarizing the statistical evidence that genes affect IQ scores.8 Jensen’s presentation guaranteed liberal opposition, since he argued that genes could explain not only IQ differences among members of the same racial group but also differences between racial groups. In a statistical sense, the latter claim was a tautology. Even if IQ differences between blacks and whites were wholly attributable to white racism and disappeared in societies dominated by blacks, black–white differences in societies such as our own would still be indirectly attributable to genetically determined differences in appearance.
But that was not what Jensen—or his readers—meant when they talked about genetic differences between blacks and whites. Jensen meant that even if blacks and whites encountered identical environments, blacks would end up with lower IQ scores than whites, presumably because blacks’ genes made it harder for them to learn the skills and information that IQ tests measured. This theory could only be tested in a colorblind society. No such society has ever existed, and it is difficult to imagine how one could exist. Jensen’s critics therefore argued, quite correctly, that his racial case was based on speculation, not evidence. This allowed most liberals and radicals to dismiss his entire hereditarian argument, on the grounds that anyone who thought whites genetically superior to blacks was bound to be wrong about almost everything.
Hereditarian theories suffered another political setback in the 1970s, when Leon Kamin demonstrated that Sir Cyril Burt, the leading English advocate of the genetic hypothesis, had fabricated most of his data. As a result, many liberals still believe that heredity has no effect on IQ scores. But the hereditarian case does not rest on Burt’s data alone. American studies of resemblance among relatives also indicate that genes have a major influence on IQ scores. Wilson and Herrnstein suggest that 60 percent of the variation in IQ scores is in some way attributable to genetic variation. This is less than the 80 percent that Herrnstein cited in the early 1970s,9 before Burt’s data had been exposed as fraudulent, but it still strikes me as a bit high. The exact number is far less important than its interpretation, however. The crucial point is that all such figures include not only genes’ effects on what children learn from a given environment, but also their effects on the way parents, teachers, schools, and peers treat a child, as well as on the environments children select for themselves. Wilson and Herrnstein fail to mention this.
There is, then, a critical difference between the way we estimate the “heritability” of a trait and the way we usually interpret such estimates. Since there is no practical method for separating the physical and social effects of genes, heritability estimates include both. This means that heritability estimates set a lower bound on the explanatory power of the environment, not an upper bound. If genetic variation explains 60 percent of the variation in IQ scores, environmental variation must explain the remaining 40 percent, but it may explain as much as 100 percent. If, for example, genes affected IQ scores solely by affecting children’s appearance or behavior, and if their appearance or behavior then affected the way they were treated at home and at school, everything genes explained would also be explicable by environmental factors. In such a world, environmental differences would explain 100 percent of the variation in IQ scores. But if genetic variation explained 60 percent of the variation in children’s environments, there would be no contradiction between the claim that genes explained 60 pecent of the variation in IQ scores and the claim that the environment explained 100 percent. My earlier example of hair length illustrates the same idea. In that case individual preferences explained 100 percent of the variation in hair length, but having an X rather than a Y chromosome explained 60 percent of the variation in individual preferences.10
Unfortunately, most social scientists and laymen interpret heritability statistics as if they set an upper bound on environmental influences rather than a lower bound. In everyday language, the statement that something is hereditary means that it is “not environmental.” In the same way, if someone says that 60 percent of the variation in IQ scores is genetic, we take this to mean that even if we treated all children exactly alike, their IQ scores would still vary greatly. This is simply wrong. The proposition that genes explain 60 percent of the variation in IQ scores is entirely compatible with the proposition that treating all children alike would result in their all having identical IQ scores. I do not mean to suggest, of course, that treating all children alike really would make their IQ scores identical. No sensible person believes this. My point is merely that hereditarians cannot refute such claims simply by showing that genes influence IQ. Hereditarians can only discredit the extreme environmental position by showing that when we treat children exactly alike their IQ scores still differ. Since we never treat children exactly alike, the claims of extreme environmentalists are in practice irrefutable, even though often implausible.
The link between IQ scores and crime is almost as controversial as that between genes and IQ scores. Surveys of teen-agers find that those with low IQs report having committed a greater number of serious crimes than those with high IQs. Teen-agers with low IQs also get arrested more often than those with high IQs. And adults with low IQs are overrepresented in prisons. Until quite recently, however, most criminologists dismissed the association between IQ and crime as meaningless. They did not deny that criminals who had been arrested or convicted mostly had below-average IQ scores. Nor did they usually suggest that this was solely the result of police or judicial prejudice against low-IQ suspects and defendants. Instead, they mostly argued that IQ tests were culturally biased against the poor and that convicts had low scores on such tests because they mostly came from disadvantaged backgrounds. There is some truth to this argument, but not much. To begin with, class background explains only 10 to 20 percent of the variation in adolescents’ IQ scores, so the idea that IQ scores are simply disguised measures of social advantage is exaggerated.11 Furthermore, the association between IQ scores and criminal behavior persists, at least among teen-agers, even when we look at people from the same class background.12
Once again, the question that ought to be explored is not whether IQ scores predict the statistical likelihood that individuals will engage in murder, rape, robbery, and the like, but why they do so. The prevailing view among sociologists is that children with low IQs learn less in school, earn low grades, and react negatively to this experience. In order to protect their self-respect, many reject the standards of the adult world that defines them as incompetent. Unable to win at being smart in school, some of them turn to being cool and tough. As they get older they also find that the adult world offers them no clear, legitimate route to material success. They know they are not “college material.” Unless they have family connections, the more demanding blue-collar crafts are also closed to them. They can see that they would have to work extremely hard to support a family on the wages available in the unskilled or semiskilled jobs they can expect to get. Confronted with this bleak prospect many look for an alternative. At least in the short run, crime is the best-paid job open to an unskilled teen-ager, even though most of those who make their living this way can expect to spend time in prison, and many die violently.
Wilson and Herrnstein give qualified support to this explanation of the connection between IQ scores and crime, but they do not seem interested in its social or political implications. If the conventional sociological view is correct, adolescents with low IQs commit crimes not because they are inherently more hostile, amoral, or impulsive than their high-IQ classmates, but because both schools and labor markets treat them in ways that make them hostile, amoral, and impulsive. If this is true, the connection between IQ and crime probably varies considerably from one society to another. In societies like our own, youngsters with low IQs cannot do much that others value. This almost inevitably means that they will be treated like dirt and will react accordingly. In societies where youngsters with low IQs can make a more valuable contribution to the common good—through conscientious or courageous performance of simple duties, for example—they might be more likely to become self-respecting, law-abiding citizens. We do not have any data on IQ and crime in such societies, so this hypothesis is hard to test. But Wilson and Herrnstein do not even raise the question of whether low-IQ youngsters might be more law-abiding in a less “academic” (or a less competitive) society. Crime and Human Nature is a book about the way individual differences affect criminality in one particular kind of society, namely our own, not about the way societies can alter—or eliminate—the effects of individual differences.
The notion that having one set of genes rather than another can lead, however indirectly, to variations in criminal behavior of the same magnitude as the variations attributable to attending one school rather than another, living in one community rather than another, or growing up in one decade rather than another, does not strike me as either startling or alarming, but it certainly alarms most people who think of themselves as liberals or radicals. So far as I can tell, this reaction has two sources.
First, as I have already noted, many people assume that if your genes affect your behavior then your behavior is immune to environmental modification. As we have seen, and as Wilson and Herrnstein also emphasize, this is nonsense. To say that men commit more crimes than women, blacks more than whites, or slow learners more than fast learners is not to say that we can’t reduce crime among blacks, males, or slow learners. Such persons did not, after all, commit anything like as many crimes a generation ago as they do today.
Genetic explanations of crime also alarm us because we fear they will lead to more brutal treatment of criminals. The notion that criminals are “different” has been used to rationalize horrifying abuses in the past, and the same thing could happen again. The danger here, however, is not that a realistic understanding of genetic influences will lead us to think of criminals as subhuman, but that the mythology surrounding genetic explanations will do so. The most serious risk is that we will come to think of criminals as incorrigible. If people commit crimes not because of the situations in which they find themselves but because of what they personally bring to the situation, then the fact that they have committed crimes in the past can create a strong prima facie expectation that they will commit more crimes in the future. Such reasoning leads many people to favor locking up even first offenders more or less indefinitely.
But the same misguided logic can also come into play if you think that criminals differ from the rest of us for purely environmental reasons. If a child has been neglected or abused for many years, this experience is as irreversible as having inherited the wrong genes. The consequences of having been abused in childhood may be reversible, of course, but this is equally true for the consequences of having inherited the wrong genes. A medical analogy may be helpful here. Suppose you have a deaf child. The disorder may be inherited, or it may be the result of childhood disease or accident. But the prospects of curing the disorder do not depend in any direct sense on whether it is a product of nature or nurture. Rather, they depend on what is actually wrong, and how much your doctors know. Furthermore, if the disorder cannot be cured, the question of how you should educate such a child does not depend, at least in any simple way, on what caused the problem. The same principle holds for behavioral problems.
In the right setting, moreover, genetic explanations can be helpful rather than damaging. Research on adopted children in Sweden shows, for example, that if a Swede’s natural father was an alcoholic, his own chances of becoming an alcoholic are significantly greater than if his father was not an alcoholic.13 This does not mean that there is an “alcoholism gene,” any more than that there is a “crime gene.” But it does suggest that your genetic makeup can somehow influence your susceptibility to alcohol addiction. Far from resisting this notion, many alcoholics seem to find it quite helpful, and Alcoholics Anonymous has endorsed it. Genetic explanations of alcoholism dramatize the fact that alcoholics cannot expect to engage in social drinking just because others do so. Telling alcoholics that they are genetically different is also a way of transforming their problem from a character defect into an illness. This not only makes society less punitive but makes the alcoholic feel less guilty. In many cases, of course, guilt is both an appropriate and a productive response to past behavior. But it does not seem to be a good thing for alcoholics, who often drink to escape it.
It does not follow, of course, that genetic explanations of crime would have a salutary effect on chronic offenders’ behavior. Much would depend on the mechanisms that actually link genes to crime, about which we still know very little. I can imagine some chronic offenders using simplistic genetic theories to excuse their behavior. (“It’s not my fault I keep breaking the rules. I’m just one of those people who can’t follow the rules no matter how hard I try. Some people are just born to be criminals.”) But social explanations of crime can be abused in much the same way. If we say that crime is a product of poverty, racism, or parental abuse, this too may provide those who are poor, black, or abused with an excuse for committing more crimes. It can also encourage others to believe that poor, black, or abused criminals are incorrigible.
If liberals have trouble with the idea that people’s genes influence their chances of committing crimes, conservatives have trouble with the idea that poverty causes crime. Conservatives do not deny, of course, that the poor commit more crimes than the rich. But instead of assuming that poverty causes crime, conservatives usually assume that poverty and crime have a common cause, namely the deficient character or misguided values of the poor.
Wilson and Herrnstein have surprisingly little to say about the effects of either poverty or social class on crime. They devote chapters to the effects of genes, gender, age, IQ, personality, schools, television, drugs, and even race, but none to the effects of either social class or poverty. Their introduction justifies this omission with the rather lame observations that “class is an ambiguous concept” and that crime and class could have common causes. Fortunately, other writers are less reticent.
Elliott Currie’s Confronting Crime is both short and readable, and unlike Crime and Human Nature it concentrates on what we can do to reduce crime. Nonetheless, Currie’s book has gotten far less attention than Wilson and Herrnstein’s, partly because Currie is not as well known as Wilson or Herrnstein and partly because Currie’s approach to crime is currently out of fashion. Prepublication excerpts from Wilson and Herrnstein’s book appeared in The New York Times; excerpts from Currie’s book appeared in Dissent.
Currie argues that crime is a product of economic inequality. This claim will appeal to anyone who, like myself, favors more equal economic rewards. Egalitarians are as addicted to universal panaceas as everyone else. We want to believe that equality is the answer to every problem for the same reason libertarians want to believe that free markets are the answer to every problem. It is always easier both to remember and to promote a single universal formula than many ad hoc ones.
The notion that equality can reduce crime also has obvious political appeal. Egalitarianism is usually rooted in some combination of sympathy for the disadvantaged and guilt about privilege. But egalitarians know that appeals to compassion and moral principle are seldom enough to move the polity toward more egalitarian policies. We would therefore like to argue that reducing inequality also has practical benefits for the privileged. Even those who benefit from today’s competitive system and have no instinctive sympathy for those who lose the race might, we imagine, endorse egalitarian reform if they thought it was the only alternative to having their house burgled. But precisely because liberals and radicals derive such obvious political and emotional benefits from believing that economic equality can reduce crime, those of us who want to avoid adding self-deception to the list of our sins must look at the evidence for this belief with special care.
Currie adduces a variety of evidence for his claim that economic inequality causes crime, but three facts seem to me central to his argument. First, the economically disadvantaged commit a disproportionate share of all crimes, especially serious crimes. Second, American cities with relatively equal distributions of household income or relatively equal job opportunities for blacks and whites have lower rates of violent crime than cities where job opportunities and incomes are less equal. Third, countries with a high level of economic inequality usually have more homicides than equally affluent countries with less inequality. In each case, however, the facts are subject to more than one explanation.
In the absence of reliable data on individuals, Currie’s argument about class background and crime begins by contrasting rich and poor communities. In Illinois, for example, he contrasts East St. Louis, where nearly 40 percent of all families are poor, with Oak Lawn, where only 3 percent are poor. In East St. Louis one person in every thousand was murdered in 1983—almost twelve times the national average. In Oak Lawn no one was murdered in 1983. While East St. Louis is an extreme case, Oak Lawn is not. The same pattern recurs all across America.
To some extent the difference between Oak Lawn and East St. Louis reflects the fact that Oak Lawn is white while East St. Louis is black. But when we separate blacks from whites we still find large class differences in criminal behavior, even in the same city. Marvin Wolfgang and his colleagues followed a large sample of boys born in Philadelphia in 1946, for example. Boys living in low-income neighborhoods had been charged with almost twice as many criminal offenses by the time they turned eighteen as boys of the same race living in high-income neighborhoods.14 The difference would presumably have been even greater if the “high-income” neighborhoods had been truly well-to-do, like Oak Lawn. In fact, they were merely neighborhoods in which the typical family had an income above the Philadelphia average.
While there is a lot more crime in poor neighborhoods than in affluent ones, the relationship between parental income and crime within any given neighborhood is relatively weak. This reflects the way people choose the neighborhoods in which they live. If a poor family is anxious to keep its children out of trouble and help them join the middle class, it will often make substantial sacrifices to live in a better neighborhood and enroll its children in a better school. Poor families living in middle-income neighborhoods are thus likely to be more ambitious and more law-abiding than those living in poor neighborhoods. Conversely, middle-income parents are unlikely to remain in poor neighborhoods if they are seriously concerned about their children’s behavior. As a result, children from such families get in trouble with the law almost as often as children from poor families in the same neighborhood.15
While the poor commit more crimes than the rich, they do not, of course, commit these crimes solely because they have low incomes. If low incomes alone drove people to crime, graduate students, clergymen, and young schoolteachers would commit a lot of crimes. These examples suggest that we should not exaggerate the effects of income per se. Economic inequality involves much more than just money. Sociologists often find that a father’s occupation has more effect on his children’s behavior than his income does, and this may well be true for teen-age crime. This does not invalidate Currie’s basic argument, but it does suggest that reducing economic inequality would require us to reorganize work as well as redistribute income. While Currie would doubtless agree, conceding this point makes egalitarian reform look considerably harder.
But to what extent do economic disadvantages cause crime? As Wilson and Herrnstein point out, the same defects of character could cause both poverty and crime. Many people who are poor may just be more ignorant, more aggressive, or—as Wilson and Herrnstein claim—more “impulsive” and “present-oriented” than their affluent brethren. If that were the case, closing the economic gap between the rich and poor would not be likely to reduce crime among the poor, since it would not remove the defects of character that caused crime in the first place.
One way to assess this argument is to look at communities rather than individuals. Poor communities should, of course, have more crime than rich communities, regardless of whether poverty causes crime or character defects cause both poverty and crime. But if we look at communities with the same percentage of poor people, and if those with a large economic gap between rich and poor have more crime than those with a small gap, we cannot easily argue that this is just because such communities attract n’er-do-wells. A more plausible explanation would be that what sociologists call “relative deprivation” leads to crime.
A study by Judith and Peter Blau, cited in both Currie’s book and Wilson and Herrnstein’s, supports this conclusion.16 The Blaus asked what characteristics of America’s 125 largest metropolitan areas were associated with high levels of violent crime in 1970. Because of local differences in whether citizens report crimes to the police and in the way the police record crimes, much of the apparent variation in cities’ crime rates is spurious, but this does not appear to be a serious problem for murder. The Blaus found high murder rates both in cities where whites worked in much more skilled and better paid occupations than blacks and in cities where the difference in incomes between rich and poor was unusually high. In cities where differences in both income and jobs were less marked, such as Utica, New York, or Johnstown, Pennsylvania, the murder rates were lower. The absolute amount of poverty made no difference, except insofar as it was associated with inequality.
But the Blaus also found that ethnic and cultural differences had large effects on murder rates. Even when people in cities had similar economic characteristics, those with more blacks had more murders. Cities with many divorced and separated adults also had more than the average number of murders, even after taking account of the economic and racial mix. This was not just a matter of bad blood between ex-spouses; cities with many divorced adults also had more than their share of robberies, presumably because high divorce rates indicate a more general attenuation of social bonds.
Differences between countries also appear to derive from a mixture of economic and cultural influences. The size of the income gap between the rich and the poor accounts for about a sixth of the variation in countries’ murder rates, for example.17 This is far from trivial, but it suggests that other factors are even more important than economic inequality in explaining variations in murder rates. Table B, on the following page, underlines this point.
In the late 1960s your chances of being murdered in Mexico were twice those in the United States, while your chances in the United States were twice those in India, five times those in Australia, and ten times those in France or Spain. The gap between rich and poor was surely greater in Mexico than in the United States in the late 1960s. But it seems safe to assume that the gap between rich and poor was also greater in India than in the United States, and yet India was much safer.18 This was not just because India was more rural than America. Your chances of being murdered in New York City were four times those in Bombay. And while most European countries were both more law-abiding and more egalitarian than America, France had about as much income inequality as America but had far less violent crime. 19
Comparisons between nations or cities, like comparisons between individuals, are always subject to the objection that the observed patterns of association may not be truly causal. Looking at Table B, for example, many readers may suspect that the differences between Mexico, America, India, Australia, France, and Spain are really attributable to cultural differences between the Mexicans, Americans, Indians, Australians, French, and Spanish who inhabit these nations, not to differences in economic opportunities or outcomes. But we cannot test such intuitions unless we can say precisely what it is about different ethnic groups that leads us to expect their criminal behavior to vary. That turns out to be very difficult. “Cultural” explanations of crime tend to be tautological: “the Japanese commit fewer crimes than the Mexicans because the Japanese have more respect for the law,” for example.
Because of these difficulties, changes over time are probably our best guide to causal connections. The historical experience that still shapes most Americans’ views about crime began in the early 1960s and ended in the early 1970s. It was marked by many egalitarian reforms and by a dramatic increase in crime. Those on the right tend to assume a causal link between the two phenomena. Those on the left deny it. Before looking for causal links, however, we need to ask what really happened.
The increase in crime during the late 1960s is beyond question. The figures on murder in Table B are the most reliable indicator of the trend, because they are not subject to much reporting error, but all other measures of crime show the same increase, often in even more dramatic form. The economic significance of the egalitarian reforms initiated during the late 1960s is harder to assess. The incomes of the poorest fifth of all families rose from 25 percent of the national average in 1963 to 28 percent in 1973—only a modest improvement but nonetheless a step toward equality rather than away from it. The absolute improvement in living standards was greater. Only 11 percent of all Americans were officially poor in 1973, compared to about 20 percent in 1963. But as Currie emphasizes, unemployment among black male teen-agers also rose during this period. Since black male teen-agers were already responsible for a large fraction of all crime in America, any economic change that made it harder for them to join mainstream society could have been expected to produce more crime. If, despite the egalitarian reforms of the 1960s, the economy stopped providing enough jobs for young black males during those years, the most plausible explanation of rising crime rates would be that the reforms were inadequate, not that they were counterproductive. That, indeed, is Currie’s argument.
Currie’s own analysis suggests, however, that the increase in black teen-age unemployment may have been a product of cultural rather than economic change. As Currie sees it, there has never been a dearth of poorly paid, unskilled jobs in which the prospects for promotion were negligible. But, he tells us, black male teen-agers don’t want these “dead end” jobs. Such jobs therefore go to women, illegal immigrants, and others who regard any job as better than none. What black teen-agers want, according to Currie, are “good” jobs—jobs that pay several times the minimum wage and offer a reasonable chance of promotion. Unfortunately, employers who want to fill such jobs are not likely to hire inexperienced, semiliterate teen-agers as long as they have older, more competent, and more responsible applicants. Since black teen-agers don’t want bad jobs and can’t get good ones, they often end up with no job at all.
If this analysis is correct, the increase in black teen-age unemployment may have been caused not by contracting economic opportunities but by rising expectations. Many black teen-agers believed in the 1960s that they had been excluded from good jobs in the past solely because white employers were prejudiced against them. When discrimination became illegal, good jobs were supposed to open up. In fact, however, discrimination had been only part of the problem. Because blacks had been excluded from most good jobs, the culture in which they grew up had not put a premium on either the technical skills or the personality traits that white employers seek when filling such jobs. Nor had the schools that whites provided for blacks prepared their students for such jobs.
The civil rights movement did not change this situation. Indeed, during the late 1960s and early 1970s some civil rights activists were inclined to glorify rather than deplore those features of black culture that made white employers reluctant to hire black teen-agers. Nor did efforts to improve the schooling available to blacks deal with these problems, though some small programs showed very promising results. As a result, black teen-agers’ skills and reliability did not improve much after the passage of the Civil Rights Act of 1964, and demand for their services did not increase in proportion to their expectations. This imbalance could have pushed a lot of young black men into crime.
If this argument is correct—and it is certainly speculative—crime may have increased during the late 1960s because the rhetoric of the times led young black men to feel both that they were entitled to good jobs and that employers who complained about their technical incompetence or social behavior were racists. This was, of course, just what many conservatives, as well as liberals of an older generation, claimed such rhetoric would do. Neither liberals nor conservatives, it should be added, did much to improve the schools provided for blacks.
More recent history also seems to me to contradict Currie’s claim that economic inequality leads to more crime. Since 1980 the real incomes of the poorest American families have fallen, while the real incomes of the more affluent have risen substantially. Inequality has increased as much in the past few years as it fell between 1963 and 1973. Currie’s analysis implies that this increase in economic inequality should have led to rising crime rates. In fact, crime has declined somewhat.
Looking further back in history, we find a dramatic, long-term, worldwide decline in crime during the second half of the nineteenth century. Yet this, too, appears to have been a period of increasing economic inequality. It would be foolish to conclude from this meager evidence that economic inequality helps to reduce crime. Crime rates depend on how people respond to economic inequality rather than on the actual level of inequality, and these responses appear to vary with the historical circumstances—a polite way of saying that we have no clear idea what determines people’s responses.
Yet even if Currie’s claim that egalitarian reforms can reduce crime is overstated, his book still has the enormous virtue of concentrating our attention on societal rather than individual determinants of crime rates. This is, of course, the very “error” that Wilson and Herrnstein seek to correct. Yet the reader who compares Tables A and B can easily see that differences between societies have far more effect on crime rates than the genetic differences between individuals to which Wilson and Herrnstein give a central place in their book.
Liberals and radicals will, no doubt, take this conclusion as a vindication of what they knew all along. But for those who hope to reduce crime, it may not be such good news. Figuring out how genes affect human behavior is a formidable task, but the prospects for significant progress over the next couple of generations seem to me somewhat encouraging. Our methods for identifying specific genes and for tracing their physical effects have improved steadily. While this does not mean that we will also be able to sort out genes’ behavioral consequences, it seems likely to make the task easier.
When we turn from physiology to culture and ask why America, Mexico, and the Caribbean have more crime than most of Europe and Asia, the obstacles to intellectual progress look to me even more formidable than when we try to understand the effects of genes. The record of the past generation is also less encouraging. We are not, I think, any closer to understanding why cultures differ from one another, or why they change over time, than we were thirty years ago. Worse yet, young social scientists are seldom interested in such questions. Most of them seem to prefer mathematical games, which are absorbing and fun to play but have almost no chance of telling us anything useful about anything. Without a clear understanding of why Europeans and Japanese respect one another’s person and property more than Americans do, it is hard to see what practical benefits we can reap from simply knowing that culture and history are important. Yet our failure to understand the deeper causes of the crime rate does not mean that it must remain as high as it is. What went up twenty years ago, for whatever unknown reason, can come down, even though criminologists cannot currently offer any reliable advice about how to make this happen.
February 12, 1987
Michael Bohman et al., “Predisposition to Petty Criminality in Swedish Adoptees,” Archives of General Psychiatry, Vol. 39 (November 1982), p. 1234. ↩
For a skeptical review of the evidence on testosterone and male aggression see Anne Fausto-Sterling’s Myths of Gender (Basic Books, 1985). While Fausto-Sterling shows that the links between testosterone and aggression are more complicated than conventional accounts imply, this certainly does not prove that hormonal differences between males and females have nothing to do with male aggression or violence. ↩
Wilson and Herrnstein also present data from the Danish study suggesting that having biological parents with a criminal record exerts more influence on a son’s behavior than having adoptive parents with a criminal record. This comparison is misleading, however, because natural parents with criminal records had usually committed far more crimes than adoptive parents with criminal records. ↩
See the correspondence between Leon Kamin and Sarnoff Mednick, William Gabrielli, and Barry Hutchings in Science (March 1, 1985), pp. 983–989. ↩
Wilson and Herrnstein also summarize a number of studies comparing small samples of identical twins, who share all their genes, with fraternal twins, who share roughly half their genes. These studies show that identical twins’ criminal records are more similar than fraternal twins’ records. This evidence is not very convincing, however, because identical twins are also likely to have more influence on one another than fraternal twins. ↩
This estimate is derived from data presented by Sarnoff Mednick et al. in Science (May 25, 1984), pp. 891–894. ↩
See Edmund McGarrell and Timothy Flanagan, eds., Sourcebook of Criminal Justice Statistics—1984, US Department of Justice (US Government Printing Office, 1985). ↩
“How Much Can We Boost IQ and Scholastic Achievement?” Harvard Educational Review (Winter 1969), pp. 1–123. ↩
Richard Herrnstein, “I.Q.,” The Atlantic (September 1971), pp. 43–64. ↩
For a more technical discussion of these issues see my paper “Heredity, Environment, and Public Policy Reconsidered,” American Sociological Review (October 1980), pp. 723–736. ↩
The best data on social background and adolescent IQ scores probably appear in William Sewell, Robert Hauser, and Wendy Wolf, “Sex, Schooling, and Occupational Status,” American Journal of Sociology, Vol. 86 (1980), pp. 551–583. Background explains somewhat more of the variation in IQ scores among younger children than among adolescents. ↩
See Travis Hirschi, Causes of Delinquency (University of California Press, 1969), or Albert Reiss and Albert Lewis Rhodes, “The Distribution of Juvenile Delinquency in the Social Class Structure,” American Sociological Review, Vol. 26 (1961), pp. 720–732. ↩
Michael Bohman, “Some Genetic Aspects of Alcoholism and Criminality,” Archives of General Psychiatry, Vol. 35 (1978), pp. 269–276. ↩
Marvin Wolfgang, Robert Figlio, and Thorsten Sellin, Delinquency in a Birth Cohort (University of Chicago Press, 1972), p. 119. ↩
See the paper by Reiss and Rhodes cited in note 12. The studies reviewed by Charles Tittle, Wayne Villemez, and Douglas Smith in “The Myth of Social Class and Criminality,” American Sociological Review (October 1978), pp. 643–656, mostly seem to me to show the same thing, though the authors do not interpret them in the same way that I do. ↩
“The Cost of Inequality: Metropolitan Structure and Violent Crime,” American Sociological Review (February 1982), pp. 114–129. ↩
Steven Messner, “Societal Development, Social Equality, and Homicide: A Cross-National Test of a Durkheimian Model,” Social Forces (September 1982), pp. 225–240. See also John and Valerie Braithwaite, “The Effect of Income Inequality and Social Democracy on Homicide,” British Journal of Criminology (January 1980), pp. 45–53. ↩
Data on the distribution of family income in different countries is seldom comparable, so statistical comparisons are likely to be wrong. Felix Paukert (“Income Distribution at Different Levels of Development: A Survey of Evidence,” International Labor Review, August–September 1973, pp. 97–126) reports, for example, that the distribution of family income was more equal in India than in either the United States or other affluent countries, largely because the Indian poor received a significantly higher fraction of national income than the poor in more affluent countries. I have never found anyone who had spent time in India who believed this. ↩
The paper by Paukert cited in note 18 reports more inequality in most European countries than in the United States during the 1960s, which contradicts Currie’s argument and my own. Malcolm Sawyer provides somewhat more recent and probably better data on Europe in his “Income Distribution in OECD Countries,” OECD Economic Outlook, Occasional Studies (Paris: Organization for Economic Cooperation and Development, 1976). These data show more inequality in the United States than in almost any European country. ↩