This article originally appeared in The Skeptic, Volume 7, Issue 4, from 1993.
Ray Ward says (The Skeptic, 7.2), the ‘Cyril Burt Affair’ is of significance to studies of paranormal belief. However, it is of rather more significance than the reasons he gives, which are that research can alter apparently well founded views, and that strong prejudices can bias understanding. The further significance is that Burt published extensively on extra-sensory perception, and these writings, although properly skeptical in many ways, in general took a rather benign view of a number of paranormal phenomena. He held for instance, that ‘the experimental evidence now provides a fairly strong presumption in favour of so-called telepathy’, and seems to take the view that ‘spirit messages from beyond the grave’ were unlikely to have arisen from chance coincidences, but to represent some other phenomenon. His general reliability and credibility are therefore of some interest.
Science deals with what happens: history and law deal with what happened. Standards of proof in these fields are necessarily rather different, for, in history and law, it is not possible to repeat the event under consideration as part of an experiment. It is therefore difficult to prove that Burt’s behaviour in his own field of research was deliberately fraudulent, rather than profoundly flawed through incompetence, since this deals with motivation. Fortunately, it is not necessary to decide which is most applicable – that can remain a matter of opinion – if it can be demonstrated that it must be one or the other.
As Ray Ward says, two books have tried to rehabilitate Burt. These are Science, Ideology and the Media: the Cyril Burt Scandal by Ronald Fletcher (Transaction 1991), and The Burt Affair by Robert B Joynson (Routledge, 1989). Of the two, Joynson’s book is the more measured in tone. Fletcher sets out to defend Burt and some of the views he espoused, while Joynson has no such additional aim. Since Ray Ward is right to say that prejudices can distort perception, I should perhaps say that I have no position to defend on the interrelationship of intelligence, inheritance and the environment – my interest in this subject relates only to the methodology of analysing evidence in science.
Since strong disagreements plainly arise over this matter, I think it is best to concentrate on those matters on which there is a degree of agreement on facts. For this reason, I will follow Fletcher and Joynson’s accounts for evidence of Burt’s wrongdoing, and leave the reader to decide if it was deliberate or accidental.
The Missing Ladies
Ray Ward says that ‘There now seems no doubt that the “missing ladies” existed’. This is not the case. A number of hearsay descriptions of the ‘ladies’ have been made, not all of which are mutually consistent. Joynson says ‘There is still no certain documentary evidence for the existence of either’, though a Miss M A Howard was a member of the British Psychological Society in 1934. He certainly admits a degree of uncertainty as to whether or not they existed. Fletcher identifies a ‘Miss H M J Conway’ who certainly existed in the right place at the right time, but properly cannot confirm her identity as the ‘J Conway’ cited by Burt. Miss Howard he takes to be the M A Howard mentioned above. This can hardly be said to admit of no doubt as to their existence.
However, both Joynson and Fletcher concede that Burt himself wrote papers and letters to journals under false names, and ‘Miss J Conway’ was employed in this way as a pseudonym in one significant work. This is not a minor peccadillo. Personally, I would count it as outright fraud, but it may merely have been extraordinarily unwise. The reason I consider it fraudulent is that the essence of testing scientific work is its independent repeatability. To have worked with someone does not compromise a researcher’s reliability, and their subsequent work may count as a verification. Writing a paper and attaching a false name to it gave Burt the opportunity of citing this work as if it was an independent confirmation.
It can be quite proper to add someone’s name to a paper – even as first author – without their participation in writing it, if they made a significant contribution to the work. Ideally, they should see and approve the manuscript before publication, but it may well happen that this is not possible. What is wrong is for the true author to delete their own name in such circumstances. For minor publications such as correspondence and book reviews, the reader may also decide whether it is right for an editor to write such items under false names and insert them in a scientific journal, particularly when they comment on the editor’s own work or that of his rivals.
Invariant Correlations
A number of papers were published in which the numbers of separated twins changed, but the correlations between their intelligence, and/or other properties, remained the same. Burt is defended against this charge in a number of ways. One is that in some papers, he was merely repeating tables published in previous work. This is plainly the case. However, in such cases it is incumbent on the author to state that this is being done. In particular, tables including new and old data together must indicate which is which, and in all papers in which a statistical argument is made, it is necessary to state explicitly the number of cases (n) underlying each category. Burt did not do so in the papers under consideration in this regard. This is unwise, and likely to mislead. In the case of the key invariant correlation between intelligence of separated twins and of non-separated twins (0.771 and 0.925 respectively, in two separate papers involving different numbers of twins in each case), it is argued that this is not impossible, that Burt did not draw particular attention to it, and that he was too clever to fake the results in this particular way. This last argument is not wholly reliable: to say that ‘no sensible person would do such a thing’ is not of itself sufficient to demonstrate that it was not done, or that the person doing it was not otherwise sensible. Nor can anything be made of Burt not drawing particular attention to it. To any mathematically literate analyst it requires no attention drawn to it. That it was not impossible is of course true. It is unlikely but possible. It bolsters a lifetime philosophical position that Burt espoused in the most convincing way imaginable: it was produced without adequate documentation when he was over 80 and of admittedly waning powers. But it is indeed possible.
Inadequate Description
The last paragraph raises another category of criticism: that Burt described his work inadequately. For the crucial twin studies, Burt provides no clear details of when or how or by whom the twins were identified and assessed. It is not enough to say that his methods were clear enough in general: in papers of such vital import the methodology must be clearly and explicitly spelled out, or clearly referenced to an accessible source. Not to do so is a sign of extraordinary sloppiness. It does not necessarily imply fraud: but clear descriptions of methodologies are one of the indications that fraud or error had not taken place. His defenders suggest that Burt’s standards were no worse than those of other psychologists at the time; whether or not this startling accusation aimed at a large field of scientific endeavour is true is irrelevant in this instance.
Inadequate Measurement
In his first ever paper, Burt set out to investigate the relationship between the intelligence of parents and their children, with the aim of estimating the relative contribution of heredity and environment. Tests were administered to small numbers of children, and the results validated by subjective impressions. The intelligence of their parents (in essence their fathers) was not tested at all: it was inferred from their social position. Burt interpreted the results as showing that intelligence was inherited, and he referred to these results with evident satisfaction in his later work. This parody of the scientific method would be laughable, if this were not an important topic.
School Attainment Levels
Burt was opposed to comprehensive education and modern teaching methods, and contributed to the second ‘Black Paper’ on education which promoted a traditionalist view. In this he said that modern attainment standards were appreciably lower than those achieved previously. No figures were given here, but a reference led to an article in the Irish Journal of Education. A table was published in this journal, the compilation of which Burt ascribed to a Miss MG O’Connor, who has also not been traced. Curiously, this table gives rather flimsy evidence for a fall in standards. However, the essential problem with this table is that it gives no indication of where, or by whom the data was collected, nor does it give the numbers involved. Some of the earlier data might well have been available to Burt; the later data would have been harder for him to collect. Of course, this does not prove that he invented it. He might just have been completely incompetent at reporting it.
In conclusion, Burt may have been guilty of deliberate fraud, or of merely failing to conform to the most basic scientific standards. His biographer concluded that some of his activities were fraudulent, and speculated on what may have driven him to this. These speculations, which Ray Ward attacks so vigorously, are irrelevant to the assessment of the value of Burt’s work. It is quite possible for a fair-minded person to read the books by Fletcher and Joynson, and still believe that Burt’s reputation had been grossly overestimated during his lifetime. There is certainly, for skeptics, no compelling reason to accept Burt’s valuation of the worth of parapsychological events.
