Much of the controversial research has been summarized in great detail in The Bell Curve, published in 1994 by Richard J. Herrnstein and Charles Murray. It immediately attracted much media attention, and was denounced by some as thinly veiled racism. The authors have repeatedly been publicly denounced as racists. In response to the debate, a public statement circulated by fifty two internationally known scholars was published in The Wall Street Journal, (December 3, 1994), which summarized what they considered to be the mainstream views on race and IQ. These scholars held that the reasoning and data in the book were reliable, and that most of the conclusions were valid.
Since then, many other scientists have disputed the analysis in The Bell Curve, and have found what they see as serious methodological flaws. A critique of the book can be found in the revised and expanded edition of The Mismeasure of Man, by Stephen Jay Gould (1996, W. W. Norton and Co., ISBN 0393039722.)
In the first edition of that book, published in 1981, Gould made a number of critical points concerning many of the studies Herrnstein and Murray were to draw on. Gould’s larger point is that most scientific studies of the relationship between race and human behavior have been heavily biased by the assumption that human behavior is best explained by heredity. He criticizes studies of the relationship between race and intelligence on several grounds. One thing he points out is that much of the data that earlier scientists relied on may have been falsified. The prime example is Cyril Burt’s famous study of the IQs of twins separated at birth. Serious questions have been raised as to the actual extent of Burt’s research (but recent studies have yielded reliable estimates of the heritability of intelligence, which fall within the same quantitative range as Burt’s figures - and it should also be noted that the specific charges against Burt by Leon Kamin about “invariant correlations” have since been shown to be seriously flawed).
Most of Gould’s criticisms pertain to cases where the data seems to be legitimate. Most of his arguments have to do with the value of statistical correlations (the measure of the co-occurrence of two different things). Most arguments around IQ center on the issue of correlation – the claim that the test measures an actual thing requires that the answers to various questions will correlate highly; the claim that this thing is inherited requires that the scores of respondents who are closely related will correlate significantly more highly than results of those distantly related.
First, he points out that correlation is not the same as cause. As he puts it, measures of the changes, over time, in “my age, the population of Mexico, the price of Swiss cheese, my pet turtle’s weight, and the average distance between galaxies” will have a high positive correlation – but that does not mean that Steven Jay Gould’s age goes up “because” the population of Mexico goes up. Second, and more specifically, a high positive correlation between parents’ IQ and children’s IQ can be taken as evidence that IQ is inherited – OR that IQ is determined by social and environmental factors. Since the same data can be used to argue either side of the case, the data in and of itself is not useful. This is why studies of twins separated at birth, and of adopted children, are given so much attention.
Furthermore, Gould makes the subtle and often ignored point that even if it were demonstrated that the correlations in IQ within a group were completely determined by heredity, this tells you nothing about the causes in differences in IQ between unrelated groups or whether those differences can be changed by environment. One example that Gould brings up is height which is known to be highly heritable. Knowing that differences in height within a single group are due to heredity tells you nothing at all about why there are height differences between different groups.
According to Gould, a good example of the confusion of heritability is found in the statement of international scholars published in the Wall Street Journal (see web-link above): “If all environments were to become equal for everyone, heritability would rise to 100% because all remaining differences in IQ would necessarily be genetic in origin.” He says that this claim is at best misleading and at worst, false. First, it is very hard to conceive of a world in which everyone grows up in the exact same environment; the very fact that people are spatially and temporally dispersed means that no one can be in exactly the same environment (a simple example will illustrate how complex social environments are: a husband and wife may share a house, but they do not live in identical environments because each is married to a different person). Second, even if people grew up in exactly the same environment, not all differences would be genetic in origin. This is because embryonic development involves chance molecular events and random cellular movements that alter the effects of genes.
Gould argues that heritability is not a measure of phenotypic differences between groups, but rather differences between genotype and phenotype within a population. Even within a group, if all members of the group grow up in exactly the same environment, it does not mean that heritability is 100%. All Americans (or New Yorkers, or upper-class New Yorkers – one may define the population in question as narrowly as one likes) may eat exactly the same food, but their adult height will still be a result of both genetics and nutrition. In short, heritability is almost never 100%, and heritability tells us nothing about genetic differences between groups. This is true for height, which has a high degree of heritability; it is all the more true for intelligence. This is true for other reasons besides ones involving “heritability”, as Gould goes on to discuss.
Gould’s most profound criticism is his rejection of the very thing that IQ is meant to measure, “general intelligence” (or “g”). IQ tests, he points out, ask many different kinds of questions. Responses to different kinds of questions tend to form clusters. In other words, different kinds of questions can be given different scores – which suggests that an IQ test is really a combination of a number of different tests that test a number of different things. Proponents of IQ tests assume that there is such a thing as general intelligence, and analyze the data so as to produce one number, which they then claim is a measure of general intelligence. Gould argues that this one number (and therefore, the implication that there is a real thing called “general intelligence” that this number measures) is in fact an artifact of the statistical operations psychologists apply to the raw data. He argues that one can analyze the same data more effectively and end up with a number of different scores (but valid, meaning they measure something) rather than one score.
Finally, Gould points out that he is not opposed to the notion of “biological variability” which is the premise that heredity influences intelligence. He does criticize the notion of “biological determinism” which is the idea that genes determine destiny and there is nothing we can or should do about this. Many people who study race intelligence hold that Gould is not representing their views correctly, and is effectively engaging in straw-man attacks on their work.
The book The Bell Curve Debate: History, Documents, Opinions, edited by Russell Jacoby and Naomi Glauberman, offers a range of responses to the book and these issues.