A true pro and his cons

This is the fourth book written about the allegations of fraud against Professor Sir Cyril Burt; and it is by far the most detailed and objective. Whatever the prior opinions of the reader, this examination of the available evidence must surely be accepted as scrupulously fair and lucidly presented.

When Burt died in 1971 at the age of 88, he had long been regarded as one of the major figures in British psychology. During his long career he made many significant contributions not only to both basic and applied psychology but also to educational policy and its realisation. He was clearly a man of many talents. As Arthur Jensen points out in the introductory chapter, "there are only three characteristics about which one finds complete agreement: Burt's exceptional intellectual brilliance, his extraordinary general erudition and his untiring industry". To these might be added his remarkable mastery of the English language (among others) and the great ability, even in his later years, to grasp ideas new to him and then lucidly communicate them to others.

It was three years after his death that attention was drawn to some apparently puzzling features of his data concerning the inheritance of intelligence: by Arthur Jensen, who admired Burt and shared his views on the importance of genetic factors; by Leo Kamin, an acknowledged expert in the theory of animal learning, in a strongly worded book that questioned the scientific validity of all research publications, then available, supporting a significant role of heredity in measured IQ; and, finally, by two former PhD students of Burt's, Alan and Anne Clarke, in a textbook on mental deficiency.

But actual allegations of fraud were first made public in a Sunday Times article by Oliver Gillie on October 24 1976. This appeared just as Leslie Hearnshaw, who had given a moving eulogy at Burt's memorial service, was near to completing his official biography Cyril Burt: Psychologist. Hearnshaw, reluctantly no doubt, had to consider the allegations by Gillie, later supported by Jack Tizard, an avowed environmentalist, and the Clarkes. When Hearnshaw's book finally appeared in 1979, Burt's many admirers were distressed that some, but not all, of these allegations were accepted. Worse for them, but perhaps raising doubts about the motives of Burt's detractors, the Council of the British Psychological Society endorsed Hearnshaw's conclusions in a pamphlet, A Balance Sheet on Burt, although the authority and credibility of this is questionable given that the directly relevant contributions came from Hearnshaw and from Gillie and the Clarkes.

To understand the initiation and continuation of the controversy about these allegations, it is necessary to appreciate an underlying complex of scientific, political and personal factors and events. Reference to this inevitably appears in all the contributions to the book under review, but the most intellectually satisfying feature of the book under review is the attempt to draw conclusions about the allegations on the basis of careful analysis of the publications of Burt and of his detractors.

Jensen, in the opening chapter, after summarising biographical details, outlines a history of the controversy, describing the allegations of Hearnshaw and of others and the rebuttals offered in post-Gillie books by Robert Joynson, The Burt Affair (1989), and Ronald Fletcher, Science, Ideology and the Media: The Cyril Burt Scandal (1991). His summary is what one might expect of a cautiously sceptical admirer of Burt but it clearly presents the issues that are considered in later chapters.

One of Hearnshaw's charges was that Burt falsified the early history of factorial analysis - a method for describing a matrix of correlations (for Burt, usually scores of individuals on different ability tests) in terms of a small number of dimensions - to enhance the importance of his own contributions at the expense of those of Charles Spearman, his predecessor as head of the psychology department at University College, London. Joynson advanced strong reasons in favour of Burt's account; and, given Hearnshaw's lack of expertise in this field, this accusation was never very convincing as Richard Rawles and I pointed out in a review of Joynson's book for The Psychologist in 1990.

In Cyril Burt: Fraud or Framed? Steven Blinkhorn, in a convincing scholarly analysis of the history of factorial analysis - and it is easy to understand why he expects it to "raise hackles" - not only concludes in Burt's favour but draws attention to other notable contributions that seem to have been overlooked. At the same time, Blinkhorn expresses some unease about Burt's tendency to gloss over significant detail and he draws attention to a relevant appendix referred to by Burt but yet to be found.

The major charge against Burt is that additional data on kinship correlations of IQ, especially for monozygotic twins reared apart, reported to have been gained over the period 1955 to 1966, were fabricated. The failure to trace named assistants in this research, one of whom appears as an author of one of the papers, was a principal feature of Gillie's accusations. In chapter three, Nicholas Mackintosh, the editor and prime contributor to this book, pursues this matter in great depth and in an admirably balanced way. He meticulously follows up Joynson's explanation of the apparently mysterious invariance of certain highly relevant kinship correlations when sample sizes appear to have been increasing; and he convincingly demonstrates that the only real mystery lies in the data reported in the last paper, published in 1966, and, even here, he is almost prepared to accept that the anomalies could be explained by obvious errors in transcribing data from three different sources. The "smoking gun", to use Leo Cronbach's phrase, is a reference in Burt's diary to "calculating data on twins for Jencks" which relates to Christopher Jencks's request for details of the IQ scores on which the correlation was based. Joynson had already demonstrated that the diary entries, to which Hearnshaw gave some weight, were far more intermittent and tended to report day-to-day trivia rather than significant professional activities. Yet "calculate" suggests that the required data are not to hand. Mackintosh examines the data forwarded to Jencks in great detail and identifies inconsistencies with statements in the main body of the 1966 paper and other papers. Although expressing his obvious suspicions, he does not claim that he can be certain that fraud had been committed.

The defence scenario he envisages is a muddle in copying data which can clearly be seen elsewhere (and admitted to his sister by Burt) and that the data were real, although Burt's description of their source disguises their possible collection at a much earlier date. It should perhaps have been mentioned that the suspected paper appeared when Burt was 83 and working in comparative isolation.

In chapter four, Nicholas Mascie-Taylor considers the allegation by D. D. Dorfman in 1978 that the data in Burt's paper on "Intelligence and social mobility" (1961) were fabricated. He agrees with his co-authors that Burt often concealed information on the size of samples and when they were collected. After a most detailed analysis that also questions some aspects of Dorfman's conclusions, as others did at the time, he finds himself unable to find an interpretation of Burt's account of his procedures that can be reconciled with all aspects of his published data. Yet, as for several papers by Burt, he acknowledges that this paper stimulated other studies on IQ and social mobility and these have confirmed Burt's conclusions.

In chapter five, Mackintosh investigates a paper published by Burt in 1969 on the decline of educational standards from 1914 to 1955. He gives Burt the benefit of the doubt over whether he had obtained the data on which his conclusions were based. His main criticisms are as follows: first, that it is not at all clear how Burt could have employed comparable IQ tests for the years in question, and his descriptions of his procedures are very inadequate; second, that the decline in IQ reported by Burt is in marked contradiction to other major studies that have shown steady increases in measured IQ from 1930 or even before. Mackintosh is painstakingly fair in exploring alternative explanations for this latter discrepancy. He leaves us with two alternative conclusions: Burt used tests for his post-1945 data that were not made comparable with the older tests; or some of the data were fabricated. He leans towards the latter conclusions but as a probable and not a certain one. In chapter five, Hans Eysenck, a former student of Burt, gives a highly illuminating account of the complex of factors surrounding this controversy: Burt's complicated personality and ambivalence towards his students; the bias of the media in regarding heredity as having no relevance to intelligence, and a reluctance to report strong evidence to the contrary; the misrepresentation by his critics of Burt's actual position on education selection; and, above all, using an attack on one man to suppress truths, important for education policy, based on a wealth of evidence about IQ, selection and genetics.

Eysenck also has a section on the fudging of data by great scientific figures - Newton, Kepler, Ptolemy, Mendel and Pasteur - to arrive at conclusions later shown to be correct and speculates on the reasons for this. He cannot, of course, resist the opportunity to cite Freud as a lesser figure also guilty of fraud.

In the final chapter, Mackintosh both considers the scientific and political impact of Burt's work and weighs up the pros and cons of the case against him. In assessing the impact of Burt's later papers on science and educational policy, Mackintosh refers to his much earlier and very thorough review of Kamin's 1974 book, which I assume stimulated his interest in the controversy. Then, although critical of Kamin's presentation of his case, he praised him for showing that the quality and quantity of the then-vailable evidence on the heritability of IQ was very much less than the consensus view. Mackintosh now comes full circle in this book to the question of the heritability of IQ and gives his personal conclusion that we can say no more than that this is certainly greater than zero and probably greater than 0.3. One suspects that this is the beginning of a future debate between him and Eysenck and Jensen, who surely would argue for a much higher minimum value.

It is a truth about psychology that there remains an inadequately bridged gulf between the distinct fields of the search for general laws and the investigation of individual differences. I speculate that those concerned with the laws of learning - and Mackintosh and Kamin are both distinguished contributors to our knowledge - must be optimistic that these laws will come to help significantly in explaining individual differences in measured ability. But, at this time, the real problem for all those who favour a dominant role for environmental factors in determining intelligence is that they have no positive model for making quantitative predictions about kinship correlations comparable to the kind (admittedly often requiring guestimates of important parameters) advanced by such as Burt. On whether Burt committed fraud, his own personal judgement is "that the cumulative weight of evidence makes it difficult to sustain Burt's innocence". He is strongly influenced here by the sometimes vague and obscure descriptions of his samples and test procedures that lead to the suspicion that there is something to hide. In the final paragraph of the book, Mackintosh concludes that none of the cases against Burt, taken alone, is sufficient to establish guilt beyond reasonable doubt. In that way he concurs with the "not proven" verdict of Eysenck and Jensen. However, he states that we are not in a court of law and, on balance, the evidence makes it probable that some of the data were fabricated.

Although Mackintosh, the chairman of this book's examining magistrates, justifiably concludes that Burt is at least guilty of deception, for an overall appraisal of Burt's work after his retirement (and this is what is under examination) I would draw upon the conclusions of Eysenck and Jensen, and a comment by Mascie-Taylor. Eysenck states that "at bottom, he preferred theory and statistical analysis to experimental rigour and hypothesis testing along deductive lines".

I more or less agree but would stress that Burt, often from the pages of the British Journal of Statistical Psychology which he edited until the age of 82, was mostly concerned to show how certain important problems could be tackled and to draw attention to the available quantitative methods. Jensen writes: "A talented scientist who works largely alone makes a good many personal enemies. He is sometimes careless and eccentric in the presentation of his studies. He becomes a public figure. More important, he develops politically incorrect theories on socially sensitive topics. This combination of factors gives his opponents - aided by sympathetic journalists -ample ammunition to attack him."

During his later years Burt was deeply concerned that the environmentalist position was gaining ascendancy in educational policy without much significant empirical evidence in its favour. Few who knew him would consider that he was motivated by any political ideology, indeed it seems clear that this was much more the case for his detractors. Some of the responsibility for the imperfections of his later publications - on the inheritance of IQ - must lie with the psychological community. Mascie-Taylor expresses some surprise that the normal editorial process had not required Burt to revise his papers; for his own journal this is obviously no mystery, but he also published elsewhere.

The fact is that many psychologists were in awe of him; and his detractors, instead of confronting him during his lifetime, bided their time. Although some of the evidence he published in his later years is of doubtful scientific value, the contributions he made during a long professional life remain impressive, and I believe it is misleading to continue to hold him up as the icon of scientific fraud.

Robert Audley is emeritus professor of psychology, University College, London.