Skip to main content

Fallacies in Discussions of Race

[Editor’s Note: This essay is an abridged version of a chapter in The Real American Dilemma: Race, Immigration, and the Future of America, available at the American Renaissance store.]

It is an embarrassment to defenders of the intellectual parity of the races that all the evidence goes against them. Black children fall behind white children in school from the earliest grades. No black society has ever produced a written language or mathematics, white whites and Asians have done so man times. Whites consistently outscore blacks on IQ tests.

Lacking any case for racial equality in intelligence, egalitarians are forced to attack the abundant evidence against it. This negative, reactive approach encourages far-fetched reinterpretations of data — e.g., blacks do badly on tests because of worry that poor performance will reflect on their race — as well as purely definitional and conceptual objections. Thus, until recently, the favored criticism of IQ tests were that “intelligence” is meaningless and that the tests are biased. (These criticisms are inconsistent, by the way, since a test for a non-existent trait cannot be biased, that is to say, measure the trait more accurately for one group than another.) The wide publicity given the validity of intelligence testing by Richard Herrnstein and Charles Murray in The Bell Curve seems to have lessened carping about IQ per se, at least for now, but egalitarians have simply redirected their attacks elsewhere. The present essay looks at some newer but no sounder efforts to discredit inegalitarianism.

I have sorted the fallacies discussed below into the purely verbal and those concerning heritability.

The Purely Verbal

Since it is now widely agreed that “intelligence” names an intrinsic property of the mind, only one purely verbal ploy is currently used by egalitarians, namely the alleged vagueness of “race.” No race can be more or less intelligent than any other, we are assured, since there is no such thing as race. Why, allegedly, is there no such thing as race? Because genetic differences between human groups are insignificant, human groups have cross-bred, and given the unlimited number of ways to classify human beings any one classification is arbitrary.

Scientific friends of the concept of race will rush to defend it because of its fruitlessness and predictive validity, but I fear that defense just sustains the imbroglio. Decades of dealing with disputes about “free will,” “knowledge” and kindred philosophical terms have convinced me that as soon as someone starts to make heavy weather over a word, one should abandon it. Disavow its use, and then, since the storm was about language rather than facts, restate the facts without it. Your interlocutor should be delighted — he’s the one unhappy with the term, after all — although in practice you may find him aggrieved. When people don’t want clarity, as Sam Goldwyn might have said, nothing can stop them. No other tactic really has any chance of moving the debate forward. An invitation to mud-wrestle should not prompt a search for the best hold; the smart move is to stay out of the mud.

The smart move in the present instance is to drop the word “race” and reformulate whatever you wish to say without it. What people usually have in mind by “race” is geographic ancestry, as indicated by the interchangeability of “black” with “African-American,” and by liberal indignation at “Eurocentrism” when they really mean to attack whites. Consequently, instead of saying that whites are more intelligent than blacks, say that people of European ancestry are more intelligent than people of sub-Saharan African ancestry. To allow for the intermarriage that has occurred in the U.S. over the past three centuries, refine this to: People of European ancestry are more intelligent than people, 75 percent or more of whose ancestors 10 generations back, were born in sub-Saharan Africa. These paraphrases, using only notions like “ancestor” that even obscurantists accept, engage the substantive issues. Nothing is lost but a word.

Nor do the seemingly small genetic differences between Africans and Europeans nullify talk of ancestry. Cavalli-Sforza, Menozzi, and Piazza report1 that the typical European and African share 99.84 percent of their genes. But one cannot reason a priori that an absolute difference of 0.0016 percent is too small to support genetic divergence in IQ. Genetic effects tend to be nonlinear; even minute genetic differences can bear large phenotypic consequences. Humans and chimpanzees have 98.4 percent of their genes in common (This is so simply because most human and chimp genes are dedicated to building hearts, lung, and other shared structures.). This does not make chimpanzees as intelligent as humans. The human/gorilla genetic difference, at 2.5 percent2, is 1.5 times the human-chimp difference, but the human/gorilla IQ gap is not 1.5 times the human/chimp gap. It would not be surprising if world-class golfers differ from the average duffer in the genes controlling coordination by less than 0.0016 percent. One must first determine the contribution of genetic difference to phenotypic differences, individual or group, and then decide whether it is “significant.”

Incidentally, Cavalli-Sforza, Menozzi, and Piazza also report that the average genetic “distance” between humans is 0.0012 percent of the genome, which puts the black/white distance a third again as great as the average, a surplus anything but trivial.

An allied and quite popular statistical fallacy is that within-race genetic variations so exceed between-group differences that talk of group differences is “meaningless.”3 This is like arguing that, since the variation in weight among dogs and among cats is greater than the mean dog/cat differences, it is meaningless to say that dogs on average outweigh cats.

The Question of Heritability

Roughly speaking, the heritability of a phenotypic trait is the degree to which individual differences in that trait are due to individual genetic differences. As heritability fallacies play on the more technical aspects of heritability, however, more precise definition is needed.

The total variation of a trait in a group, over the entire range of environments to which the group has been exposed is the trait’s phenotypic variance for that population. The total variation in the population’s genes is the population’s genetic variance. The trait’s heritability is then the ratio of the genetic variance to the phenotypic variance. Thus, in a population of genetically identical clones, the heritability of weight must be zero, since all variation in weight depends on variation in the environment. If in another population there is as much variation in the genes that influence weight as there is variation in weight, the heritability of weight reaches its other limiting value of one.

Twin studies consistently demonstrate a high heritability for intelligence. Identical twins raised apart, differing only in their rearing environments, are far more alike in IQ (and many other traits) than random pairs of individuals and more alike even than fraternal twins raised in the same household. Environmentalists object that placement agencies may give separated twins to similar families and that reared-apart twins reunited by psychologists influence each other, but these factors cannot begin to explain the extent of the similarity between twins. The best current estimate of the heritability of IQ is 0.7, increasing to 0.8 toward the end of the lifespan as individuals increasingly shape their own environments. The higher the heritability of a phenotype, the greater the environmental disparity needed to explain a phenotypic gap of a given size, hence the less likely the gap is to be caused solely by environment. In my book Why Race Matters, I calculated that for an IQ of 15 points between two individuals to be wholly environmental in origin, those individuals must occupy environments more than two standard deviations apart, an improbably large separation.

Given that a high heritability for IQ is beyond dispute, all an environmentalist can do is challenge the significance of heritability itself. And many environmentalists dully contend that heritability is a scientifically meaningless or “lousy” concept.4

The environmentalist challenge to heritability is based on relativity of heritability to environments. We saw earlier that the heritability of a trait, for a given population, depends on how the trait varies. Because of gene/environment interaction, however, a trait may show much less variation in one set of environments than in another. For instance, weight in a population may vary between 140 and 270 lb. when food is abundant, but vary between 120 and 150 lb. for a genetically identical group when food is scare. The heritability of weight is high in the first case, where each individual’s weight is determined by his genetically determined metabolism, but low in the second, where the scarcity of food cancels genetic differences in metabolization.

According to Richard Lewontin, this shows that:

the linear model [phenotypic variance = genetic variance + environmental variance] is a local analysis. It gives a result that depends upon the actual distribution of genotypes and environments in the particular population sampled. Therefore, the result of the analysis has a historical (ie., spatiotemporal) limitation and is not in general a statement about functional relations . . . [T]he particular distribution of genotypes and environments in a given population at a given time picks out relations from the array of reaction norms that are necessarily atypical of the entire spectrum of causative relations . . . The analysis of causes in human genetics is meant to provide us with the basic knowledge we require for correct schemes of environmental modification and intervention. . . . Analysis of variance can do neither of these because its results are a unique function of the present distribution of environment and genotypes.5

Lewontin moves much too quickly. The ratio of genotypic to phenotypic variance in a limited group of environments is a perfectly well-defined statistic. It does not tell us everything we want to know about a genotype, but it tells us something. Without infallibly predicting the behavior of a genotype everywhere, heritability over a varied range of environments permits informed guesses about phenotypes in unexamined ones. When a trait such as intelligence has appeared in the same form in every ecological niche yet colonized by humans, it may reasonably be expected to emerge in like form in new niches. Ensembles of environments do not become “necessarily atypical” simply by falling short of exhaustiveness.

Lewontin’s criticism of heritability is simply the old philosophical problem of induction applied to genetic variation. That something has always happened a certain way is no guarantee that it will continue to. It does not follow from the fact that copper has always conducted electricity on Earth that it conducts electricity in other galaxies, or that it will conduct electricity on Earth tomorrow. Likewise, intelligence is relatively fixed in known environments but might conceivably vary wildly in so-far unknown ones. These points of logic, however, have nothing to do with what is likely to happen. No one has yet figured out how to prove that the future will resemble the past, but scientists have no hesitation in extrapolating from the known to the unknown absent specific reasons not to. Without some specific reason to believe in environments over the horizon with which intelligence genes interact in novel ways, the present heritability of intelligence is a good guide to the overall importance of genes.

A less extreme environmentalist argument runs that high between individual heritability, the sort of heritability discussed so far, is irrelevant to heritability between groups. Average body weight in an environment with adequate food might be, say, 180 lb. while the average weight of a genetically identical population in an impoverished environment might be 135 lb. Although weight is substantially heritable in both groups, the difference in average weight is due entirely to the difference in food availability. According to a similar argument, individual heritability of IQ might be high for both whites and blacks, but the average black and white IQs differ solely because white children experience more stimulating environments.

I believe hereditarians have too readily conceded the within group/between-group distinction — not that this concession destroys hereditarianism. Arthur Jensen points out that while the two kinds of heritability may be logically distinct, high individual heritability for a trait is evidence that genes contribute to group variation.6J. Philippe Rushton shrewdly notes environmentalist special pleading the assumption that, since environmental factors explain some individual variation in IQ, they must explain group variation as well. But there is a more forceful reply. The logic of estimating the role of genes in group differences is precisely the same as that of estimating the role of genes individual differences.

To ask whether environment alone can explain a phenotypic difference in IQ between two given individuals is in effect to ask how likely the given gap would be if those individuals were identical twins. The higher the (individual) heritability of IQ, as explained earlier, the more widely separated their environments would have to be, and the wider that separation, the less likely it is. An identical analysis applies to mean group phenotypic differences. For environment alone to explain the race difference, the average genotypes of the two races must be identical. To say that the average black/white gap is totally environmental in origin is, in other words, to say that the average white and the average black are identical twins. Assuming the individual heritability of IQ to be the same for blacks and whites, calculations like those done for individuals indicate how far apart the races’ mean environments must be, and how common such widely spaced pairs of environments are. The higher the individual heritability of IQ is, the less likely it becomes that the mean black/white difference is purely environmental.

Proceeding as in the individual case, an environmental cause for the 15-point racial IQ gap requires a gap of more than two standard deviations between typical black and white environments, a gap not often encountered. Of course, the issue cannot be resolved by means alone. The actual dissimilarity of black and white environments is a factual question, as is the actual dissimilarity of the environments of any two individuals. The points on which I am insisting are that within- and between-group heritabilities are structurally similar, and that the probability of a genetic contribution to a group difference rises as the individual heritability of IQ does.

A last effort to undermine the significance of heritability focuses on gene/environment correlation. The basic idea, again accepted by all sides, is that genes influence not only the bodies and minds of their home organisms but the kinds of environment these organisms occupy. A man genetically disposed to exercise may choose an environment enabling him to, perhaps by building a tennis court or living in an environment suitable for jogging. As a consequence, the “exercise gene” is not sprinkled randomly across environments, but occurs disproportionately often in those that favor exercise. You will find lots of exercise genes in the vicinity of tennis courts. This gene/environment relation is not necessarily interactive, since the genes correlated with an environment need not be the ones shaping it. Genetically high-IQ children usually have books handy, not because the children purchase books, but because their high-IQ parents, who bequeath their offspring good genes, also buy books for them. Giftedness genes correlate with reading matter without causing the correlation.

Enter the environmentalist, objecting that gene/environment correlation leads heritability estimates to include environmental effects. Take exercise once again. How often someone exercises depends on unarguably biological factors like vigor, but also on environment: residents of Southern California can and will get more outdoor exercise than genetically identical habitants of the Arctic. However, since an athletic genotype tends to choose environments conducive to exercise, it will be associated with exercise not simply by coding for biological vigor, but by encouraging a person to locate where exercise is available. Heritability conflates both sources of the gene/exercise association, counting as genetic such environmental effects as the availability of tennis courts.

As so far stated this is a non-problem, for the effects on an organism of an environment produced by the organism’s genes certainly should count as genetic. Phenotypes need not end at the skin. A tennis court built by a man to indulge his genetically-determined desire to exercise expresses his genes as surely as does his height, and the level of fitness he reaches by using that court daily is as much an effect of his genes as is his genetically determined metabolism.

Ned Block7 and Christopher Jencks8 reply, in effect, that gene-induced environmental factors can affect organisms in quite another way. Suppose — Block’s example — hair color is genetically determined, but non-redheads regularly bang redheads over the head with baseball bats. Genes for red hair will be closely associated with low IQ, and the heritability of IQ correspondingly high, because red hair provokes an IQ-lowering societal reaction, yet it would be perverse (argues Block) to term such group differences in IQ genetic. If genetic race is associated with low IQ because black skin color diminishes the expectations of teachers, which then diminishes the intellectual performance of black children, the race difference in IQ should also be considered socially caused.

How great a problem possibilities like Block’s present for heritability estimates depends on their frequency. They are certainly not prominent in the genetic literature, where most gene/environment correlations studied are of the athletes-building-tennis-courts or parents-stimulating-offspring sort. The increase in heritability of IQ with age, as individuals increasingly control the stimuli they receive, is a typical finding. Such reactive correlations as are discussed tend to be those in which phenotypes are amplified rather than diminished, as when a curious child stirs adult responses that stimulate his curiosity.

With regard to race, Block’s analogy is off-base. To see why, suppose the average redhead in Block’s hypothetical world were found to perform most poorly on the tests of mental ability that were most heritable. The beatings administered to redheads could not explain this correlation, since they would presumably depress redhead mental abilities uniformly. We would conclude, faced with such data, that the redhead/brunette IQ gap probably was genetic in origin after all, the cudgeling notwithstanding. To conclude otherwise would require, in effect, that brunettes hit redheads harder on those parts of the noggin protecting brain functions under genetic control — too great a coincidence to be believed.

Now, concerning race, it appears that the black/white gap on mental tests does in fact increase with test heritability. To conclude that the gap is nonetheless due to racist reactions to skin color is to believe that whites somehow calibrate their abuse of blacks to inhibit most strongly those aspects of mental functioning most under genetic control. Such pre-established disharmony is not impossible, but it is most unlikely.

In any case, the redhead analogy is inappropriate because whites do not egregiously mistreat blacks. American blacks are now, in some respects, a privileged class, enjoying preference in employment, schooling, the right to disrupt the free association of whites, and solicitude when victimized by white offenders. A white who did attack a black with a baseball bat would not only be guilty of assault but face federal civil rights charges as well. Indeed, about 80 percent of interracial violent crime consists of blacks assaulting whites. As for the treatment of blacks in the past, school segregation — exhibit A in the case against whites — was hardly an “assault,” since the schools blacks were permitted to attend were far superior to any educational institution ever created by a black society. It can thus be argued that Jim Crow was a stimulating environment and that the integrated schools blacks now attend are even more stimulating.

But Block’s error is not simply empirical. To see why, let’s “deconstruct” his “discourse.” Why does he draw an analogy involving hair? No doubt because hair is a literally superficial, indeed physically detachable trait. Once race is viewed as equally external, societal reactions to skin color appear entirely irrational, and not only should not count as genetic but deserve no respect at all. As the depth of race differences becomes increasingly apparent, this sort of analogy will seem increasingly silly.

I will be brief in treating the final error about inheritance that has percolated through the environmentalist literature, for the topic is somewhat technical. I have in mind the claim that blacks and whites separated too recently in evolutionary time for an IQ gap to have evolved.

Assessment of this claim depends on how wide the race gap is, how long the races have been apart, and two associated variables called “intensity of selection” and “selection against recessives.” The question is whether it is possible to find in nature the values these variables would have had to attain in order for the known race difference to have evolved in the time during which the races are known to have been apart. The answer, worked out in detail in my 1997 book, Why Race Matters, is “Yes, easily.”

The best estimate of the current mean IQ of black Africans is 73, 27 points below the white mean. The best estimate of the time of the black/white split, when pioneers from Africa entered the Eurasian land mass, is a bit over 100,000 years ago. The mean IQ of both the African and offshoot populations may be assumed to have been 73 at the start of the process. IQ rose to 100 in Eurasia, presumably as the harsher conditions in Eurasia selected against low intelligence more intensely than did conditions in Africa. Using standard methods of quantitative genetics it can be shown that the intensity of selection necessary to achieve a 27-point change is 4,000 generations (one generation = 25 years) is 0.00082, which means, roughly speaking, that each generation in Europe had to be about 1/80 of an IQ point smarter than its predecessor. That this rate of change in turn is genetically possible can be shown by means of a simple additive model in which phenotypic intelligence depends on two alleles at 100 gene loci. In that model, an intensity of selection of 0.00082 corresponds to a rate of selection against “recessives” of 0.000106. Operationally, this requires that, in the Eurasian subgroup, just 11 fewer individuals per 100,000 with a “recessive allele” at one locus reproduced as compared to the African stay-at-homes. As natural processes go, a rate of selection against recessives of 0.000106 is extremely small. In the wild and in laboratory experiments genotypes have been observed to change as much as 1,000 times faster. Dramatic shifts in the coloration of moths and the learning behavior of mice have been achieved in 40 generations, with rates of selection as high as 0.2 or 0.3. There has been plenty of time for race differences to evolve.

The Environmentalist Style

A few general remarks about the style of environmentalist writing remain to be made.

In my experience, the most clearly stated arguments are usually the most cogent ones. Writers fall back on technical jargon when they themselves do not trust their own views to be persuasive when stated clearly. Let the reader beware of environmentalist bluffs.

The reader may wonder why I have neglected that most common environmentalist fallacy, the allegation that hereditarian thinking is dangerous. I am so impatient with this outrageous tactic that I dislike even considering it. I don’t like to see the debate mired in “free speech” issues when the really important subject is race itself. So let the last word be not mine but David Hume’s:

When any opinion leads to absurdity, it is certainly false; but it is not certain that an opinion is false because it is of dangerous consequence. Such topics, therefore, ought entirely to be forborne as serving nothing to the discovery of truth, but only to make the person of an antagonist odious.

Cavalli-Sforza, L., Menozzi, P., and Piazza, A. 1993. “Demic Explanations and Human Evolution.” Science 259: 639-646.

Caccone, A., and Powell, J. 1989. “DNA Divergence Among Hominoids.” Evolution 43: 925-942.

Gould, S. 1995. “Age-Old Fallacies of Thinking and Stinking.” Natural History, June:6-13.

Block, N. 1995. “How Heritability Misleads About Race.” Cognition 56: 99-128.

Lewontin, R. 1976. “The Analysis of variance and the Analysis of Cause.” In Block, N., and Dworkin, G., eds., 1976. The IQ Controversy. New York: Pantheon: 179-193.

Jensen, A. 1973. Educability and Group Differences. New York: Harper and Row.

Block, N. 1995. “How Heritability Misleads about Race.” Cognition 56: 99-128.

Jencks, C. 1992. Rethinking Social Policy. Cambridge, Mass.: Harvard University Press.