On September 2, 2003, the Washington Post published a front page article by Rick Weiss called “Genes’ Sway Over IQ May Vary With Class,” which argued that “environmental factors—not genetic deficits—explain IQ differences among poor minorities.” Citing a recent study, Mr. Weiss claimed that “the impact of genes on IQ varied depending on a child’s socioeconomic status,” and that “the influence of genes on IQ was significantly lower in conditions of poverty, where environmental deficits overwhelm genetic potential.” Here, in the face of increasingly irrefutable evidence of the power of genes, was a conclusion liberals could love: Heredity may account for intelligence differences in middle-class whites but it is environment that mainly determines intelligence for poor blacks.
Basing his article on a not-as-yet-published study by Eric Turkheimer of the University of Virginia and several colleagues, Mr. Weiss drew the inevitable conclusion: “The results suggest that early childhood assistance programs such as Head Start can help the poor and are worthy of public support.” (The timing was perfect. Congress had just been debating an overhaul of Head Start that would turn it over to the states, some of which were likely to scale it back.)
The article compared Prof. Turkheimer’s work to a trivial experiment done by the famed anti-”biological race concept” scientist Richard Lewontin, who showed that genetically identical seeds grow differently in different soils. Poor minorities, explained Mr. Weiss, perform badly on intelligence tests “not because of their genes but because they are raised in an environment lacking in resources and poisoned by racist attitudes.” The article quoted Marcus Feldman of Stanford University, who said the study should “draw us somewhat away from genetics and back into the importance of the social sciences.”
Mr. Weiss also quoted Robert Plomin of King’s College, London, who praised the new study even though his own work does not find that intelligence is any more affected by environment among the poor than among the rich. Prof. Plomin wondered whether the groups he studied were “not as poor” as Prof. Turkheimer’s, or if they might have benefited from “Britain’s superior social safety net,” implying that the billions America has spent on blacks have not been enough.
Mr. Weiss’s article was, in other words, the perfect account of what appeared to be the perfect study with which to confound those who doubt that environmental intervention can raise the intelligence of blacks and poor people. If genes make a big difference to middle-class white children, but it is the environment that matters most for poor blacks, then all the uplift programs of the 1960s can be justified after all.
Needless to say, it was all far too perfect. The Weiss article is a startling distortion of the Turkheimer study that only exposes how desperate the Washington Post has become in the face of mountains of evidence that the liberal-egalitarian project has failed and has no hope of success.
The actual research paper (“Socioeconomic Status Modifies Heritability of IQ in Young Children,” Psychological Science, Vol. 14, issue 6, pp. 623-628.) finally appeared in November, two months after the Weiss article. It does report data that are, in some respects, surprising, but in contrast to the Post’s breathlessness, Prof. Turkheimer is quite restrained. His paper hardly mentions race, and certainly does not dismiss genetic explanations for the results. The differences between it and the Post account are so striking, it is clear Mr. Weiss must have assumed his readers would never see the original Turkheimer paper, and would therefore believe whatever he wrote about it.
What does the paper say? Prof. Turkheimer conducted a twin study, which is the classic way to determine heritability. Such studies are important in discovering the effects of genes because identical twins are genetic copies of each other, whereas fraternal twins are no more similar (50 percent) than ordinary siblings. However, the environments shared by twins, both before and after birth, are as similar as those any two different individuals can ever experience. If the environments are thought to be identical, the greater similarity between identical as opposed to fraternal twins should show just how much a trait is affected by genes.
The Turkheimer data were from the National Collaborative Perinatal Project (1959-1974), which recruited expectant mothers from 12 urban hospitals across the country, followed their babies from birth, and tested their IQs at age seven. Forty-three percent of the twins were white, 54 percent black, and three percent were “other.” Twenty-five percent of the families were below the poverty line ($4,500 in 1973 for a family of four), and the median annual income for the group was between $6,000 and $7,000, or the equivalent of around $22,000 a year in 1997 dollars. This was clearly not a typical sample of Americans.
The Turkheimer paper set out to determine whether the effects of genes and environment on IQ differ according to socioeconomic status, specifically, whether the IQs of poor children are more heavily influenced by environment than the IQs of rich children. The authors took the standard approach of breaking down the influences on IQ into three components. The first was genetic heritability. This was estimated by comparing the variation of IQ between monozygotic (MZ) “identical” twins, and dizygotic (DZ) “fraternal” twins. The greater the heritability—the influence of genes—on a trait, the greater the tendency for MZ twins to be more similar to each other than are DZ twins.
Obvious examples of highly heritable traits are hair- and eye-color, in which MZ twins are usually indistinguishable, whereas DZ twins differ from each other as much as ordinary siblings. An equally extreme example of something unaffected by genes is language. Children brought up in the same home speak the same language. MZ twins are no more similar to each other than are DZ twins, so this “trait” is not heritable.
Needless to say, the effects of genes and environment on intelligence and personality are not as clear-cut as on eye-color or language. Twin studies, including those of twins separated at birth and reared in different homes (and therefore different environments), have produced a range of estimated heritabilities for different characteristics. Intelligence, with a heritability of 60-80 percent, is a trait almost always found to be largely independent of environment. This is why the Turkheimer results are significant if, indeed, the effects of environment really are greater for poor families. Other personality traits, such as attitude toward the death penalty (51 percent heritable), divorce (40 percent), socialism (26 percent), and pajama parties (0.8 percent) appear to be the result of increasingly large environmental effects with a diminishing genetic heritability.
The second influence on IQ Prof. Turkheimer studied was “shared environment,” that is, environmental factors common to a family and shared equally by the children. Shared environment includes such things as family wealth and social standing, quality of housing, neighborhood, presence of two parents, whether a parent suffers from depression, is unemployed, in jail, etc. The National Collaborative Perinatal Project determined the quality of “shared environment” by measuring the socioeconomic status (SES) of the mothers according to standard scales.
The third influence on intelligence was calculated from the variance in IQ that remained after genes and shared environment were taken into consideration. This remainder is often attributed to measurement error but also to what is called “unshared environment.” This refers to those parts of a child’s surroundings that are different from those of his siblings, even though they grow up in the same family. Many of the different experiences children have are random events, but others reflect different preferences that cause children to seek out different experiences. These preferences often have a substantial genetic component.
The Turkheimer study calculated the amount of IQ variation due to the three components, as well as their interaction with socioeconomic status (SES). The authors found that as SES increases, genes become the dominant influence on intelligence, whereas at low SES, shared environment is the dominant influence.
In practical terms, this means that at the higher end of the socioeconomic scale, there was a clear difference between MZ and DZ twins: MZ twins were considerably more similar to each other in IQ (less variance) than were DZ twins. At the low end of the SES scale, MZ twins were not much more similar to each other in IQ than DZ twins, which suggests a strong environmental rather than genetic effect.
Prof. Turkheimer and his colleagues looked at the data in a number of sophisticated ways, but the simplest was to separate the twins into one group that was above the median SES level and another group below it. They found that, indeed, in the low-SES group, MZ twins were only slightly more similar to each other in intelligence than were DZ twins (correlations of .68 and .63 respectively) whereas in the high-SES group, the DZ twins were considerably more similar to each other than were MZ twins (correlations of .87 and .51).
This is the long and the short of the Turkheimer findings. They are interesting, but do not come anywhere near justifying Mr. Weiss’s conclusions. First, Prof. Turkheimer and his colleagues say nothing about how race affects any of these outcomes, presumably because the effect of SES they found on IQ was the same for blacks and whites. Thus, there is no justification for saying anything about minorities at all, much less asserting that their environments are “poisoned by racist attitudes.” This is not just liberal Post spin; it is pure misrepresentation designed to buttress the liberal view that whites are responsible for the poor performances of non-whites.
Nor do the findings necessarily mean the IQs of poor children can be raised by environmental intervention, as Prof. Turkheimer and his colleagues understand perfectly. Like all serious scholars and unlike most journalists, Prof. Turkheimer knows that poor children do not grow up in bad environments because of pure bad luck. Their parents largely make their environments for them, and if parents have low IQs they pass on a genetic tendency towards low intelligence while, at the same time, they build bad environments that are likely to drag it down further. As Prof. Turkheimer explains:
“It would be naïve . . . to interpret SES strictly as an environmental variable. Most variables traditionally thought of as markers of environmental quality also reflect genetic variability . . . Children reared in low-SES households, therefore, may differ from more affluent children both environmentally and genetically . . . and the models we employed in this study do not allow us to determine which aspect of SES is responsible for the interactions we observed.”
The paper goes on to point out that there is an inherent flaw in trying to determine what effect SES has on the intelligence of children when SES itself can be a result of the genetic makeup of the parents. SES is not acting independently but reflects, at least in part, the very genes the children are inheriting. As the authors explain in somewhat convoluted language, this skews their results:
“SES and IQ are correlated, and that correlation is potentially mediated both genetically and environmentally. Therefore the models [we built to understand the data] are attempting to detect an interaction between genotype and environment in the presence of a correlation between genotype and environment, raising the concern that the presence of the correlation might introduce bias in the estimation of the interaction.”
The authors do not pretend fully to understand what they have found. As they have the good grace to concede: “[M]aybe outcome simply becomes less predictable in poor environments.” Not only is this an honest admission of the limits of their theoretical model, it completely undercuts Mr. Weiss’s happy prediction that “childhood assistance programs such as Head Start can help the poor and are worthy of public support.” It is hard to imagine more flagrant disregard for reservations expressly included in a scientific paper.
Prof. Turkheimer and his colleagues do not mention this, but there is another possible explanation for their data. The twins in the low-SES group will have had IQs considerably below normal. It may be that genes that code for intelligence have a cumulative effect, in that there is a threshold level of high-IQ genes that must be reached before intelligence is seen to be consistently heritable. In other words, at the low end of the IQ range, individuals have few of the genes that code for high intelligence, and so may fail to achieve the necessary minimum required for IQ to be predominantly determined genetically. Thus, it is conceivable that children with very low intelligence may be influenced to a greater extent by environmental factors because they lack the genetic structure necessary to allow for the higher levels of intelligence that are clearly heritable.
What are some of the possible implications of the Turkheimer paper? The Post would have us believe it is saying some environments are so bad they lower a child’s IQ no matter how good his genetic endowment. In extreme cases, this is obviously true. If Isaac Newton had spent his entire life in a closet he would have lived as an idiot despite his genetic endowments. Mr. Weiss seems to be ready to conclude that poor (black?) environments are so bad they uncouple genes from intelligence in much the same way.
It is true that blacks often degrade their environments in ways that tend to depress the intellectual capacities of their children. An urban ghetto is a terrible place to rear a child. What Mr. Weiss and so many others like him fail to understand is that the ghetto does not arise independently of the genes of the people who live in it.
One in five of the poor mothers in Prof. Turkheimer’s study were younger than 21, and one third did not have husbands. Both these factors, at least to some degree, reflect personal choices and are difficult to blame on “white racism.” Despite liberal assumptions, having an illegitimate child is far more a cause of poverty than a consequence of it. Many of the poor women in this study were poor, at least in part, because of choices they made. These choices, and the cognitive and behavioral characteristics underlying these choices, are at least partially genetically determined.
Another aspect of the Turkheimer study that suggests grounds for caution is the age at which IQ was tested. The IQs of young children are notoriously unstable, and appear to be quite susceptible to temporary environmental influence. A very young child has little control over his environment, and is almost entirely dependent on what his parents provide. As children get older, they change their environments to suit their preferences and abilities. A more genetically intelligent person will change his environment in ways that stimulate his mind, thus improving cognitive development, while a dull person will do the opposite.
This is the basis for the observation that the heritability of intelligence increases with age. A dull child’s abilities may be temporarily boosted by ambitious parents who push him in certain directions, while a bright child may be held back by dull parents who offer a dull environment. Once children are able to make their own environments, their intellectual levels will more closely reflect their own genetic propensities rather than those of the adults who controlled their early environments. As we age, the effects of childhood environment fade, and our cognitive abilities reflect our genetic endowments—intelligence becomes more heritable. Genetic differences present from birth are the constant driving force in this gene-environment interaction.
Just as children eventually create environments that suit their genetic predispositions, early environment can reinforce genes both for better and for worse. As Prof. Michael Levin explained in the March 1994 issue of AR, “One of the ways [high IQ] genes produce high IQs is by producing high-quality environments, which in turn stimulate the development of children raised in them.” At the same time, it is undoubtedly true that abysmal black environments depress black IQs (at least for children), and that those environments are products of a particular genetic structure. Why, after all, is the environment almost always bad wherever there are large concentrations of blacks? Do bad environments just happen to follow blacks around, or do blacks create them?
As Prof. Turkheimer and his colleagues understand, poverty and “low SES” hardly explain everything about differences in achievement. As was reported in the November 2003 issue of AR, whites from families with incomes of less than $10,000 a year (this is less than half the median of $22,000 a year in 1997 dollars for the subjects in the Turkheimer paper) had average SAT scores 123 points higher than the black average, and no fewer than 46 points higher than blacks from families with incomes in the $80,000-$100,000 range! If, for poor children, environment overwhelms genes, why do poor whites outscore rich blacks? SAT scores are not the same as IQ scores, but they track them very closely. In fact, because they reflect the results of training and instruction, they should be biased in favor of wealthy blacks who can offer their children more enriched environments than those of white children from families that make less than $10,000 a year.
If, as Mr. Weiss so rashly claimed in his Post article, environmental intervention can work wonders for the poor even if it has little effect on the rich, why have Head Start programs produced such poor results? J.S. Fuerst of Loyola University examined 684 black students who attended special programs of such intensity they were called “Head Start to the fourth power.” The test scores of these children improved while they were immersed in these expensive programs, but 10 years later their scores were indistinguishable from those of other children from similar backgrounds. People like Mr. Weiss, who claim the Turkheimer study confirms the value of Head Start are ignoring a vast literature to the contrary.
The famed “Minnesota Transracial Adoption Study”—initially considered a triumph for IQ egalitarians—was discussed in detail in the March 1994 issue of AR. Here, too, early success faded, as children grew up and reached their own levels of intellectual ability. Sandra Scarr and Richard Weinberg tested the IQs of black and mixed-race children adopted into upper-middle-class families. At age six, these children had substantially higher IQ scores than equivalent children reared by their own parents in poor black homes. However, when the children were retested years later at an average age of seventeen, this IQ gain had vanished.
The fact that a strong environmental influence that was observed in the same age range examined by Turkheimer (six to seven years old) can completely disappear in a decade does not bode well for the long-term stability of Prof. Turkheimer’s findings. It is likely that as his pairs of twins increase in age, the correlations between the MZ and DZ IQ scores will move steadily in directions that indicate heritability of intelligence for children of all SES levels.
Another finding supporting the view that the black-white IQ gap has a genetic origin is the consistent phenomenon of regression to the mean. There is a powerful tendency for extreme or “outlier” values in any normally-distributed variable to “regress” to the average value of that variable. For example, when very tall people marry, they do not generally have children who are even taller, but ones who, though tall, are not as tall as their parents. Likewise, people with very high or low IQs do not continue to produce children with progressively higher or lower IQs. Their children likewise “regress” in the direction of the average.
At the same time, anyone with a very high (or low) IQ is likely to have parents or siblings with IQs closer to the average. For example, any given person with an IQ of 140 may turn out to have a sibling with an even higher IQ, but usually the family members of such a person have IQs closer to the group mean. When very intelligent whites and blacks are matched for the same high IQs and other variables are controlled, the family members of the blacks have lower IQs than family members of the whites. This suggests the black family members are regressing towards a lower group mean (generally found to be 85) than the white family members, who are regressing towards a mean of 100. It is difficult to understand how “racism,” poverty, etc., can explain why the brothers and sisters of a high-IQ black should have lower IQs than those of an equally high-IQ white of similar background.
There is a test that might lay to rest the question of race and IQ. Current studies that compare white and black IQ scores are always open to the theoretical objection that American society is so harshly anti-black that genetic equality with whites is smothered by the effects of “racism.” This objection could be overcome by testing a sample only of black Americans who all, presumably, face the same oppressive environment. Because almost all American blacks are racially mixed to some degree, it would be possible to determine whether IQ rises as the proportion of white genes increases.
It is now easy to determine the precise ancestral mix of anyone, by using autosomal DNA testing. The Ancestry by DNA test offered by DNA Print Genomics, Inc. gives an accurate reading of the percentage of ancestry that is white, Asian, American Indian or black. By focusing on actual racial-genetic ancestry rather than on perceived or self-identified race, the test could measure objective racial correlations with intelligence rather than bog down in debates about “socially defined” groups.
Similar methods have already shown there is a clear link between sub-Saharan African ancestry and obesity. Why not for IQ? Does IQ in the broadly-defined “black” population drop as genetic African ancestry increases? Since black Americans of varying racial proportions are all, socially speaking, equally “black” and equally subject to “racism,” such a connection would be strong evidence of a genetic cause for the racial gap in IQ.
Whites also have varying ancestries. Genetic studies have found that some “whites” have enough African (and American Indian) genes to overlap with the lower end of the “African-American” scale. If it were discovered that within a “socially-defined” white American population IQ varied with African ancestry, that, too, would be strong evidence for a genetic link between race and IQ. One could expand this approach to a country like Brazil, where a large fraction of the “white” upper class has significant African ancestry. Does IQ vary among whites according to percentage of African genes?
Perhaps there is a threshold for African admixture below which the effect is not seen, in which case results would be clearer among blacks than whites. Or, there may be a nearly constant decrease in intelligence as black ancestry increases from zero to 100 percent. And there is always the theoretical possibility that African ancestry has no effect on IQ at all.
To return, finally, to the Turkheimer study, it does offer interesting results. IQ correlations generally thought to indicate heritability of intelligence were far less pronounced for poor children than for rich children. These findings may or may not be replicated. If they are replicated, what do they mean? Prof. Turkheimer does not claim they justify Head Start or anything else, and suggests only that intelligence may develop in poor children in ways we do not fully understand.
Mr. Weiss’s damage, however, is likely to go un-repaired. The Turkheimer paper has not produced the flood of delighted media coverage that would have followed if its contents really justified the Washington Post’s giddy conclusions. Still, thanks to the Post, defenders of Head Start and proponents of the view that whites are to blame for black failure will point with pride and confidence to a study they will never read and would not understand if they did read it. It is constant and—one is tempted to say deliberate—misrepresentation of this kind that makes it impossible to discuss race honestly in this country, much less establish sensible policies.