Kinks in the curve

When Richard Herrnstein and Charles Murray's book The Bell Curve: Intelligence and Class Structure in American Life appeared in 1994, it rapidly became a bestseller, but also excited feelings of moral outrage. Epithets such as "pseudo-scientific racists" were hurled at the authors, and the book was described as "alien and repellent". At the last count, the book had given rise to more than 200 publications, including some half a dozen books with titles such as Measured Lies and Debunking the Bell Curve and Scientific Racism. A quite extraordinary flow of polemic with, it has to be said, as much hyperbole and bias among the critics as that in the original book. Why? Was the science so bad, or were the political solutions proposed so abhorrent, or did the problem lie in the connections between the two? Indeed, one may ask if the book was so poor, why give it so much attention? Should it not be simply ignored, as happens with most bad science?

Given that the Bell Curve wars show no sign of subsiding, perhaps the time is ripe for a more measured reappraisal of the evidence and arguments based on it, together with a critique of the broader issues involved in making the journey from scientific evidence to policy recommendations. This new book provides much the best attempt so far at a reasoned, dispassionate consideration of the issues and findings. That it does so owes much to the wide range of disciplines represented, but most of all to the (largely successful) commitment to be fair to what Herrnstein and Murray (hereafter referred to as H and M) actually said, and to approach the topic in a balanced, empirically based fashion. No easy task, given all that has gone before.

Stephen Fienberg and Daniel Resnick set the scene for the rest of the book with their very succinct opening chapter on the historical background. They note that during the 1920s, in a programme that had no British counterpart, 24 American states passed sterilisation laws and that by the mid-1930s some 20,000 Americans had been sterilised against their will. Nazi Germany carried things even further with some 320,000 suffering the same fate between 1934 and 1939. A reminder that this immoral implementation of supposed genetic findings occurred helps explain the outraged response to H and M's argument that the lowest social caste, which includes a disproportionate number of African-Americans, is largely genetically determined and is expanding due to over-reproduction of the least able. Before we become smug regarding British avoidance of the extremes of United States and Nazi eugenics, we need to remember that their roots are to be found in our own Francis Galton, a pioneering genius with an enormous amount to his credit, but whose views on eugenics were very close to those of H and M.

In the next chapter, Terry Belke presents a factual summary of what H and M wrote. That is essential reading for the huge number of people who think that they "know" what the book said, but who have never actually read the tome (it runs to more than 850 pages!) and who have based their opinions on impassioned reviews. It provides a useful reminder of the innumerable caveats and cautions in the book (eg that IQ scores tell you little about whether you will admire a person, and that the evidence is inconclusive on whether black/white differences in IQ are genetically determined), as well as H and M's dubious claim that their main aim is "the quest for human dignity" and the need for "valued places for everyone". Nevertheless, while being scrupulously fair to The Bell Curve, Belke also brings out clearly the highly tendentious nature of some of the science and of the questionable leaps to policy recommendations.

The remainder of the book is made up of three chapters on the genetics of intelligence, two on the concept of general intelligence and its measurement, five chapters presenting reanalyses of the National Longitudinal Study of Youth on which H and M based much of their arguments, and a final three chapters discussing the public policy arguments in The Bell Curve. As the topics indicate, Devlin and his colleagues have as their prime objective a critique "designed for quantitative readers who want to gauge for themselves the soundness of the statistical argument I to look at how hypotheses are stated and tested I to weigh the effect of using different statistical procedures I and to argue about whether the authors' understanding of heritability respects the findings of modern genetics". This might appear to suggest that the book is highly technical and difficult to follow, but it is not. Some understanding of quantitative techniques is essential for some (not all) parts of the book, but the book is clearly and interestingly written and is generally easy to follow. Moreover, the critique is also concerned with the story about scientific research and its translation into public policy. Even the least quantitative readers need to understand what is involved in that story because it has a much wider applicability.

I am puzzled why this set of critical essays starts with the question of how much of the variance with respect to individual differences in IQ can be attributed to genetic influences. H and M opted for a figure of 40 per cent to 80 per cent, whereas Michael Daniels and his colleagues settle on per cent to 54 per cent (depending on what is included in heritability). Either way, it is clear that environmental influences are important; also the heritability figure is population- and context-specific so that it is uninformative on whether or not changes in environment would help raise IQ scores. The latter point is taken up in the next chapter, by Douglas Wahlsten. He points to the evidence that IQ scores are rising and that intervention programmes have brought benefits. Most reviewers would consider the latter as more modest than Wahlsten claims, but his general point on the potential malleability of intelligence is valid.

Burton Singer and Carol Ryff focus on the importance of environmental influences on diseases such as hypertension that involve an important genetic component but yet show major differences in incidence among ethnic groups. They argue for an interplay between nature and nurture and note that the causes of differences between population groups may not be the same as the causes of differences within each group. The parallel with IQ is reasonable, but it is surprising that they do not take issue more directly with the genetic argument put by H and M. Almost none of the genetic evidence on IQ has involved African-Americans; accordingly, we know next to nothing about the heritability of IQ in African-Americans and especially not in those living in socially disadvantaged inner-city areas. Also, compared with white groups, family disadvantage in black groups is more likely to coincide with seriously adverse area influences. In any case, the heritability of IQ within population groups has no necessary implications for the role of genetics in explaining between-group differences.

Many of the critics of The Bell Curve have focused, in my view wrong-headedly, on the uncertainties over whether there is such a thing as 'g' (general intelligence) and over whether, even if there is, IQ tests measure it. John Carroll in a thoughtful, brief review of the issues and findings, concludes that H and M were correct in presuming that "there is such a thing as a general factor of cognitive abilities on which human beings differ". H and M underplayed the reality of specific cognitive skills, and it remains uncertain how far "g" is a basic feature or rather the result of a combination of specific skills. But, in truth, it does not really matter very much. At a practical level, there is such a thing as general intelligence and IQ tests do a reasonable (albeit imperfect) job of measuring it.

The issue of the extent to which IQ accounts for success or failure in life is another matter. The five chapters presenting re-analyses of the longitudinal data set do a good job in raising important questions. In particular, they note that the predictions from IQ to wages or crime (to take but two examples of outcomes) vary by gender and ethnic group. Christopher Winship and Sanders Korenman reject H and M's conclusion that education has no effect on IQ; and Clark Glymour notes the many limitations of H and M's statistical approach for testing causal hypotheses. Clearly, IQ has an important influence on life success and failure, but H and M exaggerate its importance and fail to consider the various alternative causal mechanisms that might be involved.

In many ways, the most dubious of all of H and M's claims is that the relationship between IQ and life success has increased over time in the US, that cognitive elites and their reverse have become more marked and separate, and that despite the overall rise in IQ scores the higher fertility of the less intelligent has had a serious dysgenic effect such that the "cognitive capital" of the US is declining. In a brief chapter, Nicholas Lemann casts doubt on the claims, but it is a pity that this fundamental point in H and M's line of reasoning was not subjected to more rigorous empirical scrutiny. It is most unlikely that it would hold up and, if it does not, most of the policy recommendations lose their empirical basis.

The final chapter by Resnick and Fienberg comes to the heart of H and M's claims that the scientific evidence points to their policy recommendations, which include putting more money into programmes for the gifted; ending attention to equal opportunities and affirmative action; providing better protection for wealthier neighbourhoods; and reducing the role of government. Undoubtedly, Resnick and Fienberg are correct in their assertion that H and M's recommendations are more a product of a moral persuasion and libertarian ideas than solid empirical findings. Nevertheless, it is regrettable that they did not provide a closer scrutiny of the steps in the argument from supposed scientific findings to policy recommendations. Thus, for example, if IQ is largely genetically determined, why invest more in education for the gifted? But, also, what was needed was more critical discussion of the interface between empirical findings and values in coming to policy recommendations. Thus, let us suppose that there is an increasing tendency for there to be a cognitive elite and a low IQ underclass (although the evidence that this is so is highly questionable), how should one decide how to respond? Even H and M accept that environmental influences account for 20 per cent to 40 per cent of population variance, so should one focus on environmental manipulations to effect change? If we should, what goals ought there to be, and what is the place of values and of scientific findings in coming to that decision?

Without question, Intelligence, Genes and Success is a most useful, thought-provoking book on an important topic. Both quantitative and non-quantitative readers will gain much from reading it. Personally, I do not consider the quality of The Bell Curve warrants the attention that is has received. The science is so-so, many of the values are objectionable, and the links between science and policy recommendations are feeble. The important issue is not the reality of intelligence or the importance of genetic influences (both well supported), but rather the complex considerations involved in bringing scientific evidence to bear on policy recommendations. The book is valuable in opening that crucial issue, but it is a pity that it does not take it further.

Sir Michael Rutter is honorary director, Social, Genetic and Developmental Psychiatry Research Centre, Institute of Psychiatry, London.