The Differences Are Real
Prof. Arthur Jensen
|IN 1969, in the appropriately
academic context of The Harvard Educational Review I questioned
the then and still prevailing doctrine of racial genetic equality in intelligence.
I proposed that the average difference in IQ scores between black and white
people may be attributable as much to heredity as environment. Realizing
that my views might be wrongly interpreted as conflicting with some of
the most sacred beliefs of our democracy, I emphasized the important distinction
between individual intelligence and the average intelligence of populations.
Moreover, I presented my research in a careful and dispassionate manner,
hoping that it would stimulate rational discussion of the issue as well
as further research.
Much to my dismay, however, my article set off an emotional furor in the world of social science. Amplified by the popular press, the furor soon spread beyond the confines of academia. Almost overnight I became a cause celebre, at least on college campuses. I had spoken what Joseph Alsop called "the unspeakable." To many Americans I had thought the unthinkable.
Science Vs. the Fear of RacismFor the past three decades the scientific search for an explanation of the well-established black IQ deficit has been blocked largely, I feel, by fear and abhorrence of racism. In academic circles doctrinaire theories of strictly environmental causation have predominated, with little or no attempt to test their validity rigorously. The environmentalists have refused to consider other possible causes, such as genetic factors. Research into possible genetic influence on intelligence has been academically and socially taboo. The orthodox environmental theories have been accepted not because they have stood up under proper scientific investigation, but because they harmonize so well with our democratic belief in human equality.
The civil-rights movement that gained momentum in the 1950s "required" liberal academic adherence to the theory that the environment was responsible for any individual or racial behavioral differences, and the corollary belief in genetic equality in intelligence. Thus, when I questioned such beliefs I, and my theories, quickly acquired the label "racist." I resent this label, and consider it unfair and inaccurate.
The Real Meaning of RacismSince the horrors of Nazi Germany, and Hitler's persecution of the Jews in the name of his bizarre doctrine of Aryan supremacy, the well-deserved offensiveness of the term "racism" has extended far beyond its legitimate meaning. To me, racism means discrimination among persons on the basis of their racial origins in granting or denying social, civil or political rights. Racism means the denial of equal opportunity in education or employment on the basis of color or national origin. Racism encourages the judging of persons not each according to his own qualities and abilities, but according to common stereotypes. This is the real meaning of racism. The scientific theory that there are genetically conditioned mental or behavioral differences between races cannot be called racist. It would be just as illogical to condemn the recognition of physical differences between races as racist.
When I published my article in 1969, many critics confused the purely empirical question of the genetic role in racial differences in mental abilities with the highly charged political-ideological issue of racism. Because of their confusion, they denounced my attempt to study the possible genetic causes of such differences. At the same time, the doctrinaire environmentalists, seeing their own position threatened by my inquiry, righteously and dogmatically scorned the genetic theory of intelligence.
Thankfully, the emotional furor that greeted my article has died down enough recently to permit sober and searching consideration of the true intent and substance of what I actually tried to say. Under fresh scrutiny stimulated by the controversy, many scientists have reexamined the environmentalist explanations of the black IQ deficit and found them to be inadequate. They simply do not fully account for the known facts in the comprehensive and consistent manner we should expect of a scientific explanation.
The Black IQ DeficitFirst of all, it is a known and uncontested fact that blacks in the United States score on average about one standard deviation below whites on most tests of intelligence. On the most commonly used IQ tests, this difference ranges from 10 to 20 points, and averages about 15 points. This means that only about 16 percent of the black population exceeds the test performance of the average white on IQ tests. A similar difference of one standard deviation between blacks and whites holds true for 80 standardized mental tests on which published data exist.
A difference of one standard deviation can hardly be called inconsequential. Intelligence tests have more than proved themselves as valid predictors of scholastic performance and occupational attainment, and they predict equally well for blacks as for whites. Unpleasant as these predictions may seem to some people, their significance cannot be wished away because of a belief in equality. Of course, an individual's success and self-fulfillment depends upon many characteristics besides intelligence, but IQ does represent an index, albeit an imperfect one, of the ability to compete in many walks of life. For example, many selective colleges require College Board test scores of 600 (equivalent to an IQ of 115) as a minimum for admission. An average IQ difference of one standard deviation between blacks and whites means that the white population will have about seven times the percentage of such potentially talented persons (i.e., IQs over 115) as the black population. At the other end of the scale, the 15-point difference in average IQ scores means that mental retardation (IQ below 70) will occur about seven times as often among blacks as among whites.
The IQ difference between blacks and whites, then, clearly has considerable social significance. Yet the environmentalists dismiss this difference as artificial and claim it does not imply any innate or genetic difference in intelligence. But as I shall show, the purely environmental explanations most commonly put forth are faulty. Examined closely in terms of the available evidence, they simply do not sustain the burden of explanation that they claim. Of course, they may be possible explanations of the IQ difference, but that does not necessarily make them the most probable. In every case for which there was sufficient relevant evidence to put to a detailed test, the environmental explanations have proven inadequate. I am not saying they have been proven 100 percent wrong, only that they do not account for all of the black IQ deficit. Of course, there may be other possible environmental explanations as yet unformulated and untested.
Arguments for the Genetic HypothesisThe genetic hypothesis, on the other hand, has not yet been put to any direct tests by the standard techniques of genetic research. It must be seriously considered, however, for two reasons: 1) because the default of the environmentalist theory, which has failed in many of its most important predictions, increases the probability of the genetic theory; 2) since genetically conditioned physical characteristics differ markedly between racial groups, there is a strong a priori likelihood that genetically conditioned behavioral or mental characteristics will also differ. Since intelligence and other mental abilities depend upon the physiological structure of the brain, and since the brain, like other organs, is subject to genetic influence, how can anyone disregard the obvious probability of genetic influence on intelligence?
Let us consider some of the genetically conditioned characteristics that we already know to vary between major racial groups: body size and proportions; cranial size and shape; pigmentation of the hair, skin and eyes; hair form and distribution; number of vertebrae; fingerprints; bone density; basic-metabolic rate; sweating; consistency of ear wax; age of eruption of the permanent teeth; fissural patterns on the surfaces of the teeth; blood groups; chronic diseases; frequency of twinning; male-female birth ratio; visual and auditory acuity; colorblindness; taste; length of gestation period; physical maturity at birth. In view of so many genetically conditioned traits that do differ between races, wouldn't it be surprising if genetically conditioned mental traits were a major exception?
The Heritability of IntelligenceOne argument for the high probability of genetic influence on the IQ difference between blacks and whites involves the concept of heritability. A technical term in quantitative genetics, heritability refers to the proportion of the total variation of some trait, among persons within a given population, that can be attributed to genetic factors. Once the heritability of that trait can be determined, the remainder of the variance can be attributed mainly to environmental influence. Now intelligence, as measured by standard tests such as the Stanford-Binet and many others, does show very substantial heritability in the European and North American Caucasian populations in which the necessary genetic studies have been done. I don't know of any geneticists today who have viewed the evidence and who dispute this conclusion.
No precise figure exists for the heritability of intelligence, since, like any population statistic, it varies from one study to another, depending on the particular population sampled, the IQ test used, and the method of genetic analysis. Most of the estimates for the heritability of intelligence in the populations studied indicate that genetic factors are about twice as important as environmental factors as a cause of IQ differences among individuals.
I do not know of a methodologically adequate determination of IQ heritability in a sample of the U. S. black population. The few estimates that exist, though statistically weak, give little reason to suspect that the heritability of IQ for blacks, when adequately estimated, should differ appreciably from that for whites. Of course the absence of reliable data makes this a speculative assumption.
What implication does the heritability within a population have concerning the cause of the difference between two populations? The fact that IQ is highly heritable within the white and probably the black population does not by itself constitute formal proof that the difference between the populations is genetic, either in whole or in part. However, the fact of substantial heritability of IQ within the populations does increase the a priori probability that the population difference is partly attributable to genetic factors. Biologists generally agree that, almost without exception throughout nature, any genetically conditioned characteristic that varies among individuals within a subspecies (i.e., race) also varies genetically between different subspecies. Thus, the substantial heritability of IQ within the Caucasian and probably black populations makes it likely (but does not prove) that the black population's lower average IQ is caused at least in part by a genetic difference.
What about the purely cultural and environmental explanations of the IQ difference? The most common argument claims that IQ tests have a built-in cultural bias that discriminates against blacks and other poor minority groups. Those who hold this view criticize the tests as being based unfairly on the language, knowledge and cognitive skills of the white "Anglo" middle class. They argue that blacks in the United States do not share in the same culture as whites, and therefore acquire different meanings to words, different knowledge, and a different set of intellectual skills.
Culture-Fair Vs. Culture-BiasedHowever commonly and fervently held, this claim that the black IQ deficit can be blamed on culture-biased or "culture-loaded" tests does not stand up under rigorous study. First of all, the fact that a test is culture-loaded does not necessarily mean it is culture-biased. Of course, many tests do have questions of information, vocabulary and comprehension that clearly draw on experiences which could only be acquired by persons sharing a fairly common cultural background. Reputable tests, called "culture-fair" tests, do exist, however. They use nonverbal, simple symbolic material common to a great many different cultures. Such tests measure the ability to generalize, to distinguish differences and similarities, to see relationships, and to solve problems. They test reasoning power rather than just specific bits of knowledge.
Surprisingly, blacks tend to perform relatively better on the more culture-loaded or verbal kinds of tests than on the culture-fair type. For example, on the widely used Wechsler Intelligence Scale, comprised of 11 different subtests, blacks do better on the culture-loaded subtests of vocabulary, general information, and verbal comprehension than on the nonverbal performance tests such as the block designs. Just the opposite is true for such minorities as Orientals, Mexican-Americans, Indians, and Puerto Ricans. It can hardly be claimed that culture-fair tests have a built-in bias in favor of white, Anglo, middle-class Americans when Arctic Eskimos taking the same tests perform on a par with white, middle-class norms. My assistants and I have tested large numbers of Chinese children who score well above white norms on such tests, despite being recent immigrants from Hong Kong and Formosa, knowing little or no English, and having parents who hold low-level socioeconomic occupations. If the tests have a bias toward the white, Anglo, middle-class, one might well wonder why Oriental children should outscore the white Anglos on whom the tests were originally standardized. Our tests of Mexican-Americans produced similar results. They do rather poorly on the culture-loaded types of tests based on verbal skills and knowledge, but they do better on the culture-fair tests. The same holds true for American Indians. All these minorities perform on the two types of tests much as one might expect from the culture-bias hypothesis. Only blacks, among the minorities we have tested, score in just the opposite manner.
Intelligence Tests Are ColorblindThose who talk of culture bias should also consider that all the standard mental tests I know of are colorblind, in that they show the same reliability and predictive validity for blacks and whites. In predicting scholastic achievement, for example, we have found that several different IQ tests predict equally well for blacks and whites. College-aptitude tests also predict grades equally well for blacks and whites. The same equality holds true for aptitude tests which predict job performance.
We have studied culture bias in some standard IQ tests by making internal analyses to see which kinds of test items produce greater differences in scores between blacks and whites. For example, we made such an item-by-item check of the highly culture-loaded Peabody Picture Vocabulary Test, on which blacks average some 15 points lower than whites. The PPVT consists of 150 cards, each containing four pictures. The examiner names one of the pictures and the child points to the appropriate picture. The items follow the order of their difficulty, as measured by the percentage of the children in the normative sample who fail the item.
California Vs. England; Boys Vs. GirlsTo illustrate the sensitivity of this test to cultural differences in word meanings, we compared the performance of white schoolchildren in England with children of the same age in California. Although the two groups obtained about the same total IQ score, the California group found some culture-loaded words such as "bronco" and "thermos" easy, while the London group found them difficult. The opposite occurred with words like "pedestrian" or "goblet." Thus the difficulty of some items differed sharply depending on the child's cultural background. A similar "cultural" bias shows up when comparing the performance of boys and girls, both black and white. Though boys and girls score about equally well over all, they show significant differences in the rank order of item difficulty; specific items, e.g. "parachute" versus "casserole" reflect different sexual biases in cultural knowledge.
Yet when we made exactly the same kind of comparison between blacks and whites in the same city in California, and even in the same schools, we found virtually no difference between the two groups in the order of items when ranked for difficulty, as indexed by the percent failing each item. Both groups show the same rank order of difficulty, although on each item a smaller percentage of blacks give the correct answer. In fact, even the differences between adjacent test items in terms of percent answering correctly show great similarity in both the black and white groups.
If this kind of internal analysis reflects cultural bias between different national groups, and sexual bias within the same racial group, why does it not reflect the supposed bias between the two racial groups? If the tests discriminate against blacks, why do blacks and whites make errors on the same items? Why should the most and least common errors in one group be the same as in the other?
Another way internal analysis can be used to check for bias involves looking for different patterns of item intercorrelations. For example, if a person gets item number 20 right, he may be more likely to get, say, item 30 right than if he had missed item 20. This follows because the test items correlate with one another to varying degrees, and the amount of correlation and the pattern of intercorrelations should be sensitive to group differences in cultural background. Yet we have found no significant or appreciable differences between item intercorrelations for blacks and whites.
In summary, we have found no discriminant features of test items that can statistically separate the test records of blacks and whites any better than chance, when the records are equated for total number correct. We could do so with the London versus California groups, or for sex differences within the same racial group. Thus, even when using the PPVT, one of the most culture-loaded tests, black and white performances did not differ as one should expect if we accept the culture-bias explanation for the black IQ deficit. I consider this strong evidence against the validity of that explanation.
The Effect of the TesterWhat about subtle influences in the test situation itself which could have a depressing effect on black performance? It has been suggested, for example, that a white examiner might emotionally inhibit the performance of black children in a test situation. Most of the studies that have attempted to test this hypothesis have produced no substantiation of it. In my own study in which 9,000 black and white children took a number of standard mental and scholastic tests given by black and white examiners, there were no systematic differences in scores according to the race of the examiners. What about the examiner's language, dialect, or accent? In one study, the Stanford-Binet test, a highly verbal and individually administered exam, was translated into black ghetto dialect, and administered by a black examiner fluent in that dialect. A group of black children who took the test under these conditions obtained an average IQ score less than one point higher than the average IQ score of a control group given the test in standard English.
The Question of "Verbal Deprivation"To test the popular notion that blacks do poorly on IQ tests because they are "verbally deprived," we have looked at studies of the test performances of the most verbally deprived individuals we know of: children born totally deaf. These children do score considerably below average on verbal tests, as expected. But they perform completely up to par on the nonverbal culture-fair type of tests. Their performances, then, turns out to be just the opposite of the supposedly verbally deprived blacks, who score higher on the verbal than on the nonverbal tests.
If one hypothesizes that the black IQ deficit may be due to poor motivation or uncooperative attitudes of blacks in the test situation, then one must explain why little or no difference in scores occurs between blacks and whites on tests involving rote learning and memory. Such tests are just as demanding in terms of attention, effort and persistence, but do not call upon the kinds of abstract reasoning abilities that characterize the culture-fair intelligence tests. We have devised experimental tests, which look to pupils like any other tests, that minimize the need for reasoning and abstract ability, and maximize the role of nonconceptual learning and memory. On these tests black and white children average about the same scores. Therefore, the racial difference clearly does not involve all mental abililities equally. It involves mainly conceptual and abstract reasoning, and not learning and memory.
Another factor often cited as a possible explanation for the black IQ deficit is teacher expectancy—the notion that a child's test score tends to reflect the level of performance expected by his or her teacher, with the teacher's expectation often based on prejudice or stereotypes. Yet numerous studies of teacher expectancy have failed to establish this phenomenon as a contributing factor to the lower IQ scores of blacks.
Testing the Environmental HypothesisTo test the environmentalist hypothesis, we have examined the results of those tests that most strongly reflect environmental sources of variance, and they turn out to be the very tests that show the least difference between blacks and whites in average scores. The greatest difference in scores between the two racial groups occurs on the tests we infer to be more strongly reflective of genetic variance. If the cultural environmental hypothesis were correct, just the opposite would be true.
The "Sociologist's Fallacy"In an at tempt to disprove the genetic hypothesis for the black IQ deficit, environmentalists frequently cite studies that compare IQs of socioeconomic ally matched racial groups, and find considerably less difference in test scores than the usual 15 point difference between races. Here we have a good example of the "sociologist's fallacy. "Since whites and blacks differ in average socioeconomic status (SES), the matching of racial groups on SES variables such as education, occupation, and social class necessarily means that the black group is more highly selected in terms of whatever other traits and abilities correlate with SES, including intelligence. Therefore the two groups have been unfairly matched in terms of IQ.
Those who cite the socioeconomic matching studies also fail to take account of the well-established genetic difference between social classes, which invalidates their comparison. For example, when the two races are matched for social background, the average skin color of the black group runs lighter in the higher SES groups. This difference indicates that genetic characteristics do vary with SES. Thus SES matching of blacks and whites reduces the IQ difference not only because it controls for environmental differences, but because it tends to equalize genetic factors as well.
Variables That Don't BehaveA host of other environmental variables don't behave as they ought to according to a strictly environmentalist theory of the black IQ deficit. For example, on practically all the socioeconomic, educational, nutritional and other health factors that sociologists point to as causes of the black-white differences in IQ and scholastic achievement, the American Indian population ranks about as far below black standards as blacks do below those of whites. The relevance of these environmental indices can be shown by the fact that within each ethnic group they correlate to some extent in the expected direction with tests of intelligence and scholastic achievement. Since health, parental education, employment, family income, and a number of more subtle environmental factors that have been studied are all deemed important for children's scholastic success, the stark deprivation of the Indian minority, even by black standards, ought to be reflected in a comparison of the intelligence and achievement-test performance of Indians and blacks.
But in a nationwide survey reported in the Coleman Report, in 1966, Indians scored higher than blacks on all such tests, from the first to the 12th grade. On a nonverbal test given in the first grade, for example, before schooling could have had much impact, Indian children exceeded the mean score of blacks by the equivalent of 14 IQ points. Similar findings occur with Mexican-Americans, who rate below blacks on socioeconomic and other environmental indices, but score considerably higher on IQ tests, especially on the nonverbal type. Thus the IQ difference between Indians and blacks, and between Mexican-Americans and blacks, turns out opposite to what one would predict from purely environmental theory, which of course, assumes complete genetic equality for intelligence. No testable environmental hypothesis has as yet been offered to account for these findings.
Does Malnutrition Affect Intelligence?What about malnutrition, another factor frequently cited by the environmentalists to disprove the genetic hypothesis? Malnutrition has indeed been found to affect both physical and mental development in a small percentage of children in those areas of the world that suffer severe protein deficiencies: India, South America, South Africa, and Mexico. But few blacks in the U.S. show any history or signs of severe malnutrition, and I have found no evidence that the degree of malnutrition associated with retarded mental development afflicts any major segment of the U.S. population.
Nor do I know of any evidence among humans that maternal malnutrition, by itself, can have pre- or postnatal effects on a child's mental development. The severe famine in the Netherlands during the last years of World War II provided an excellent case study of such a possibility. Thousands of men conceived, gestated, and born during the period of most severe famine, were later tested, as young adults, on Raven's Standard Progressive Matrices, a nonverbal reasoning test. Their scores did not differ significantly from the scores of other Dutch youths of the same age who had not been exposed to such maternal nutritional deprivation.
If further research should definitely establish the existence of genetically conditioned differences in intelligence between certain races, what would be the practical implications? It would take several articles to consider the question adequately, but the only morally tenable position in human relations would remain unchanged: that all persons should be treated according to their own individual characteristics, and not in terms of their group identity. Let me stress that none of the research I have discussed here allows one to conclude anything about the intelligence of any individual black or white person.
Equality of rights and opportunities is clearly the most beneficial condition for any society. Acceptance of the reality of human differences in mental abilities would simply underline the need for equality of opportunity in order to allow everyone to achieve his or her own self-fulfillment. In order to take account and advantage of the diversity of abilities in the population, and truly to serve all citizens equally, the public schools should move beyond narrow conceptions of scholastic achievement. They should offer a much greater diversity of ways for children of whatever aptitude to benefit from their education.
Environment Vs. Genetics: Still an Open QuestionI have tried to emphasize the uncertainty of our knowledge of the causes of race differences in mental abilities. I do not claim any direct or definite evidence, in terms of genetic research for the existence of genotypic intelligence differences between races or other human population groups. I have not urged acceptance of a hypothesis on the basis of insufficient evidence. I have tried to show that the evidence we now have does not support the environmentalist theory, which, until quite recently, has been accepted as scientifically established. Social scientists have generally accepted it without question, and most scientists in other fields have given silent assent. I have assembled evidence which, I believe, makes such complacent assent no longer possible, and reveals the issue as an open question, calling for much further scientific study.
Politicizing a Scientific IssueMost of the scientists and intellectuals with whom I have discussed these matters in the past few years see no danger in furthering our knowledge of the genetic basis of racial differences in mental or behaviorial traits. Nor do they fear general recognition of genetic differences in such traits by the scientific world, if that should be the eventual outcome of further research. They do see a danger in politicizing a basically scientific question, one that should be settled strictly on the basis of evidence.
Most of the attempts to politicize the issue, I have found, come from the radical left. True liberals and humanists, on the other hand, want to learn the facts. They do not wish to expend their energies sustaining myths and illusions. They wish to face reality, whatever it may be, because only on the level of reality can real problems be effectively confronted. This means asking hard questions, and seeking the answers with as much scientific ingenuity and integrity as we can muster. It means examining all reasonable hypotheses, including unpopular ones. It means maintaining the capacity to doubt what we might most want to believe, acknowledging the uncertainties at the edge of knowledge, and viewing new findings in terms of shifting probabilities rather than as absolute conclusions.
Also see Black-White IQ Differences and (off-site) Egalitarian Fiction and Collective Fraud. The scientific study of racial differences is a regular subject on PINC.