The <i> g </i> whizz in IQ

The Bell Curve: Intelligence and Class Structure in American Life by Richard Herrnstein and Charles Murray (hereafter the BC) precipitated a flood of critical articles, and the 100 contributions in these two books are mainly reprints: those in The Bell Curve Wars (hereafter Wars) are from The New Republic, while those in The Bell Curve Debate (hereafter Debate) are from various sources, even quoting Francis Galton.

The BC's premise is that individual differences in cognitive ability are well described by a single, immutable, general factor of intelligence (g), whose distribution in the population - the bell curve (the shape of a graph of population proportion having a given IQ) - has a substantial genetic basis. While this is controversial, what really inflamed passions was the use of the bell curve to analyse American society and draw three conclusions:

1. economic and social power is increasingly determined by g, and the resultant stratification of society will have at its extremes a powerful cognitive elite and a deprived underclass;

2. most social problems have their root cause in low intelligence; 3. the 15-point IQ difference between blacks and whites has genetic origins, so that blacks proportionately outnumber whites in the underclass. The book warns of an apocalypse in which the elite gains increasing political power and establishes a custodial state to maintain social order. Its panacea is devolution of federal control to "neighbourhoods"b where all cognitive strata live in mutual respect.

Given its thesis, it is not surprising to read, in Wars, that Stephen Jay Gould calls the BC "discredited Social Darwinism" and that Jacqueline Jones even regards it as "hate literature with footnotes". But Thomas Sowell, while rejecting the conclusions, considers that the book "deserves critical attention, not public smearing and private acceptance" and Alan Wolfe welcomes it "because it brings back to life a rich tradition of sociological speculation".

A major problem is to disentangle serious social analysis from conservative ideology. For Richard Nisbett, though the BC has "the trappings of a scientific work", it "wouldn't be accepted in a peer-reviewed journal", while Howard Gardner calls it, appositely, "scholarly brinkmanship". Many agree with Orlando Patterson's comment: "Alas must take seriously, 1/3 million have bought."

Recalling earlier controversies aroused by Arthur Jensen, Herrnstein himself and William Shockley, some writers wonder why the BC should now receive such prominence. Michael Lind fears "an ongoing transformation of conservatism: a new culture based on immigration, race and sex", in which, according to Salim Muwakkil (in Debate) the BC and the Republican senate majority are a dangerous tandem reflecting national xenophobia. But others relate this coming together to welfare costs. It is fuelled by the hereditarian Pioneer Fund, which is funding "professors of hate", say Charles Lane, John Sedgwick and Magnus Linklater. Lane and Randall Kennedy indict a publishing culture that feeds on controversial works, and managers of public opinion who allow journalists to show disrespect to blacks. "Be interesting, outrageous and timely", as The Economist succinctly described the impact of the book.

A few exonerate the BC from charges of reactionary racism. Nathan Glazer believes Herrnstein was unhappy with the explanations for racism commonly given in America; he imagines Herrnstein as saying: No, that gives the us the bum rap, it's the distribution of IQ. In Wars, Martin Peretz, publisher of The New Republic, considers that the BC's focus on groups is a reflection of federal policy that assumes these groups' salience and new norms that allot equal rights to groups, regardless of whether their members are gifted or not. In Debate, a conservative, Charles Krauthammer, is interested to see that liberals who insist on every institution keeping data on group membership and every conceivable activity choose to pillory the BC for including IQ scores.

So, looking as dispassionately as possible, how valid is the original book's thesis? On IQ, it draws six provocative conclusions, which are almost dogma from the classical tradition of research on intelligence but which Herrnstein and Murray coyly prefer to term "cognitive ability".

The first conclusion is that g, posited by Charles Spearman, exists, and nothing angers Gould more than the BC's claim that g "captures a real property in the head". Gould far prefers Leon Thurstone's multifactor approach; and Gardner offers his own similar supermarket of component abilities.

A more rigorous dissection of the relationship between the cognitive psychologist's view that cognition is a process and the psychometric conception of cognition as a collection of abilities is made by Earl Hunt in a review published in Scientific American. On a show of hands, five critics in Wars favour g and six are against it. The concept has simplicity and robustness on its side, and some verdigris not yet acquired by the variegated concepts of Gardner, Robert Sternberg and others.

The BC's next four conclusions require an acceptance that IQ scores are indices of intelligence. Gardner warns that the BC has succumbed to professional "intoxication with the IQ test". None the less, Sowell, Glazer and Leon Wieseltier (of The New Republic) accept the IQ test as a valid measure of intelligence.

The book's sixth conclusion is that the heritabilities of cognitive ability lie between 40 and 80 per cent. Broad heritability, h2, estimated at 60 per cent by Herrnstein and Murray, is the ratio of the additive genetic variance to the phenotypical variance of a trait. Patterson reminds us of the complications when genotype and environment co-vary.

He also refers to the argument given elsewhere by Jack Vale, that "any trait under strong selection for a long evolutionary period will demonstrate little additive genetic variance". Thus, assuming selection for intelligence, h2 should be near zero. Here is a genuine paradox, of which one resolution, seemingly absurd, is to doubt that IQ scores reflect intelligence. Furthermore, h2, a within population parameter, cannot be used to test assertions about genetic differences between black and white populations.

On the vexed question of genetically determined racial differences in IQ, Herrnstein and Murray claim to be "resolutely agnostic". Many critics doubt the authors' seriousness. John Judis points out that although the BC acknowledges that "neither American blacks nor Latinos represent distinct races", the book "proceeds to describe American blacks as a race and to talk about 'genetic differences between the races'." We agree with Judis that the BC imputes "to the difference between two social groups certain prehistoric genetic traits, making it more plausible to assert that whites are inherently smarter than blacks".

The BC makes much of the average 15-point IQ difference between American blacks and whites, terming it "the B/W gap".

The book has little to say about the evidence that the gap has narrowed over the past two decades. Nisbett summarises every relevant study reported in the BC and concludes that the median reduction in the gap is six IQ points. The BC contrives no more than a two-to-three-point reduction and pessimistically hints at a reversal.

The narrowing of the B/W gap may have been the effect of interventions to boost black cognitive skills. The BC will have none of this and belittles programmes designed to improve the educational performance of blacks. The book claims that whatever were the gains, they later vanished rapidly. No one denies this, but the interpretation put on it differs hugely: to Patterson (who is aware of the irony) it is "persuasive proof of the deleterious impact of the environment".

Douglas Besharov, categorised in Debate as a conservative, reports a study showing strong positive effects of improving a child's neighbourhood environment. He advocates the continued search for methods of raising intelligence and avers that "we are morally bound to keep plugging away" at this most seemly task for psychologists.

Several authors draw attention to Robert Flynn's finding of a near worldwide increase of three IQ points per decade, which both argues for environmental factors and challenges the BC's fears that dysgenesis will lower America's average IQ. Herrnstein and Murray attribute the increase to growing sophistication about IQ tests among the less intelligent. They argue that there is an emerging class structure in which high intelligence is universally advantageous: entry into top colleges will become dramatically more dependent on IQ; professionals and senior business executives will come from the high-IQ end of the bell curve and their financial advantage will grow; IQ will be the best predictor of job performance and employers will gain economic advantage by selecting for it; and residual wage variation, after controlling for education and so on, will rise and reflect intelligence. The outcome will be that the cognitive elite will be increasingly isolated and homogeneous, and the social "churning" of America will diminish.

Most commentators disagree. Wolfe regards Herrnstein and Murray as followers of Robert Reich and Christopher Lasch. He argues that a cognitive meritocracy would face an adverse cultural climate. On the political right, Reagan's populism denigrated brains, and on the left affirmative action dealt a blow to merit. Neither right nor left likes brains: the right prefers breeding (in the United Kingdom) or the forgotten man (in the United States); radicals dislike social distinctions.

Other criticisms of the emergence of a cognitive elite take the view that IQ is not a good predictor of performance. Yet although the Herrnstein/Murray case has weaknesses, the critics often offer opinions rather than arguments: "IQ can't substitute for training, expertise, motivation, creativity" (Gardner), "at best the BC identifies a testocracy, with little relevance to human endeavours" (Andrew Hacker). Wolfe states more perceptively: "They don't document statistical trends." And Judis is surely right to point to factors other than IQ responsible for increasing income disparity: "Declining unions, competition from low wage countries and high CEO salaries due to the identification of income and status." No contributor to Wars or Debate advances the telling economic arguments against employers selecting for high IQ, raised by Arthur Goldberger and Charles Manski.

The consensus view in these two books is probably Wolfe's: "If they [Herrnstein and Murray] wrote in modesty they would reject the IQ- performance relationship." But we feel the case made out by the BC cannot be entirely dismissed.

The book argues that socially undesirable behaviour is "associated with" low cognitive ability more than socioeconomic status (SES). The authors analyse data on white males from a major longitudinal study to determine the relative strength of association between various kinds of behaviour and SES with IQ kept constant and between various kind of behaviour and IQ with SES kept constant.

Gould accepts that the best database and an appropriate statistical analysis were used. But he (and others) has two main criticisms: that the statistical associations are weak (a fact deliberately hidden away in appendices); and that a few per cent of statistical determination is not a causal explanation. Sowell joins Gould in attacking the BC's uncritical approach to correlations largely ignoring the problem of multicollinearity, in which an unknown x may account for the correlation between measures. (Gardner thinks x is likely to be "social class and luck".) Judis notes that Herrnstein and Murray admit the problem of inferring causation from their statistics, yet throughout the book use such phrases as "low IQ a factor in", "significant determinant of".

Goldberger and Manski, referred to already, advance three major criticisms not mentioned in Wars and Debate: 1. the BC's techniques are inadequate to evaluate the role of education; 2. its index of SES is astonishingly arbitrary; 3. its comparison of the roles of SES and IQ expresses these in terms of standard deviations from average values. On education, they wrongly castigate the book for not citing references for its claim that "each year of schooling increases IQ score by about one point" but their general criticism stands. We agree that it is impossible to interpret the BC's analysis meaningfully: "Their measures give incomparable units the appearance of being comparable" (Goldberger and Manski).

Many contributors agree with Patterson that "something is horribly wrong in the condition of one-third of all black people" in the United States. Some document the long history of oppression suffered by blacks and the relatively short period that even the most fortunate have had to participate in American society on close to fair terms; and Hugh Pearson fears that blacks will accept that they cannot compete on equal terms with whites and not have the confidence that is needed to succeed in modern society. Patterson sees a dangerous contradiction in the thinking of black leaders and sympathetic whites - they cite slavery and its destructive legacy as an environmental argument against hereditarianism but they are quick to traduce anyone who in other contexts points to the cultural deficiencies of their group.

Faced by The Bell Curve and its 100 critics, what should we conclude? Only stalwart hereditarians (such as Arthur Jensen and Richard Lynn in Debate) can accept its analysis of the relationship between race - whatever that is - and intelligence; and as Jones points out, a monocausal theory that neglects gender (except poor women as procreators of the underclass) and ignores history must be inadequate. No commentators regard their political panaceas, which they see largely as Jeffersonian or even Jim Crow fantasies, as feasible in a modern society. But social problems, particularly among blacks, continue to increase in the US, with reverberations in the United Kingdom - and we lack the research needed for an informed social policy. Perhaps Christopher Winship in Debate is right when he points to the greatest peril in the BC controversy: that scholars will back away from research on cognitive skills and social outcomes and from reporting results inconsistent with environmental determination.

Robert Audley is emeritus professor of psychology and Richard Rawles is senior lecturer in psychology, University College, London.