3
84 Book reviews/Acta Psychologica 88 (1995) 79-86 D.J. Miller and M. Hersen (Eds.), Research Fraud in the Behavioural and Biomedi- cal Sciences. Wiley, New York, 1992. Due to a number of highly publicized cases, the issue of research fraud has become a major concern for both scientists as well as the general public. Although the interest in this topic is undoubtedly also due to its sensationalistic nature, there is a growing concern that certain aspects of contemporary research practice are conducive to the occurrence of fraud. In this edited volume, Miller and Hersen have tried to bring together a comprehensive treatment of such aspects. According to Miller and Hersen, their aim was to (1) elucidate various problems in the current scientific system, (2) provide guidelines to researchers and institutions for preventing such occurrences, and (3) recommend to administrators and govern- mental officials policy development that will enable them to monitor more effec- tively the huge behavioural and biomedical research enterprise. It is questionable whether they have indeed succeeded in these ambitious aims. There are a number of problems. First, some of the chapters are not concerned with or focussed on the central topic of research fraud. Instead, they deal with more general issues that are only tangentially related to the fraud issue. For example, the chapter by Schaffner on "Ethics and the nature of empirical science" has little specific to say about the issue of fraud. Although it does describe one specific case of fraud (the Summerlin affair), it does not do anything with it other than just mentioning it. Similarly, the chapter on "Institutional Review Boards" by Cohen and Ciocca deals mainly with the ethics of using human subjects and not with the issue of how such review boards may fulfil a role in handling cases of (suspected) fraud or misconduct. Second, some of the chapters are a bit superficial. I personally found the chapter by Charrow and Saks on "Legal responses to allegations of scientific misconduct" disappointing in that it describes how two institutions, the US Food and Drug Administration (FDA) and the National Institutes of Health (NIH) tried to deal with the issue of fraud and misconduct. Although the authors acknowledge the differences between the research supervised by the FDA and that sponsored by the NIH, they do not advance a clear view of how the law should deal with research fraud and indeed whether there should be separate legislation to deal with fraud. They seem to be most sympathetic to the FDA approach that involves, among other things, a formal, almost trial-like, procedure and a program of random data audits. The latter, however, are clearly not very practical for basic research and would quickly result in simply a lot of red tape and bureaucracy. In his chapter, Freedman describes this as "the judgment of the FDA's former sardine factory inspectors or fiscal auditors" (p. 185). Miller's chapter on "Per- sonality factors in scientific fraud and misconduct" does little more than describe a number of factors that might be involved. What is lacking, however, is some kind of empirical evaluation of how important these really are in the occurrence of fraud. The most interesting chapters are those that give an inside view of specific cases of fraud and those that analyze some of the factors that may trigger such

Research fraud in the behavioural and biomedical sciences: D.J. Miller and M. Hersen (Eds.), Wiley, New York, 1992

Embed Size (px)

Citation preview

84 Book reviews/Acta Psychologica 88 (1995) 79-86

D.J. Miller and M. Hersen (Eds.), Research Fraud in the Behavioural and Biomedi- cal Sciences. Wiley, New York, 1992.

Due to a number of highly publicized cases, the issue of research fraud has become a major concern for both scientists as well as the general public. Although the interest in this topic is undoubtedly also due to its sensationalistic nature, there is a growing concern that certain aspects of contemporary research practice are conducive to the occurrence of fraud. In this edited volume, Miller and Hersen have tried to bring together a comprehensive treatment of such aspects. According to Miller and Hersen, their aim was to (1) elucidate various problems in the current scientific system, (2) provide guidelines to researchers and institutions for preventing such occurrences, and (3) recommend to administrators and govern- mental officials policy development that will enable them to monitor more effec- tively the huge behavioural and biomedical research enterprise.

It is questionable whether they have indeed succeeded in these ambitious aims. There are a number of problems. First, some of the chapters are not concerned with or focussed on the central topic of research fraud. Instead, they deal with more general issues that are only tangentially related to the fraud issue. For example, the chapter by Schaffner on "Ethics and the nature of empirical science" has little specific to say about the issue of fraud. Although it does describe one specific case of fraud (the Summerlin affair), it does not do anything with it other than just mentioning it. Similarly, the chapter on "Institutional Review Boards" by Cohen and Ciocca deals mainly with the ethics of using human subjects and not with the issue of how such review boards may fulfil a role in handling cases of (suspected) fraud or misconduct.

Second, some of the chapters are a bit superficial. I personally found the chapter by Charrow and Saks on "Legal responses to allegations of scientific misconduct" disappointing in that it describes how two institutions, the US Food and Drug Administration (FDA) and the National Institutes of Health (NIH) tried to deal with the issue of fraud and misconduct. Although the authors acknowledge the differences between the research supervised by the FDA and that sponsored by the NIH, they do not advance a clear view of how the law should deal with research fraud and indeed whether there should be separate legislation to deal with fraud. They seem to be most sympathetic to the FDA approach that involves, among other things, a formal, almost trial-like, procedure and a program of random data audits. The latter, however, are clearly not very practical for basic research and would quickly result in simply a lot of red tape and bureaucracy. In his chapter, Freedman describes this as "the judgment of the FDA's former sardine factory inspectors or fiscal auditors" (p. 185). Miller's chapter on "Per- sonality factors in scientific fraud and misconduct" does little more than describe a number of factors that might be involved. What is lacking, however, is some kind of empirical evaluation of how important these really are in the occurrence of fraud.

The most interesting chapters are those that give an inside view of specific cases of fraud and those that analyze some of the factors that may trigger such

Book reviews/Acta Psychologica 88 (1995) 79-86 85

behaviour. These include "Cardiology: The John Darsee experience" by Braun- wald, "Plagiarism: The case of Elias A.K. Alsabti" by Miller, "The consequences of fraud" by Poling, "Academic pressures" by Thelen and DiI~renzo, and "Edi- torial processes, safeguards, and remedies" by Freedman. Each of these is not only interesting to read but also contains important lessons for anyone involved in conducting and supervising scientific research.

The strangest, and in fact somewhat out-of-place, chapter is that by Jensen on the Cyril Burt case. I assume that the editors had in mind a chapter that describes the effect of a highly publicized fraud case on the general evaluation and appreciation of the subject area involved, and on those working in it. What Jensen has delivered, however, can only be characterized as a partisan review of the Burt case that is intended to leave the impression that there is little that Burt can be blamed for and that the whole affair is a witch-hunt by egalitarian communists and anti-hereditarians (these are not my terms; Jensen repeatedly uses such ad hominem arguments). All of this is perplexing and disturbing since the chapter itself does give ample evidence for the fact that Burt held quite unconventional views (to say the least) regarding research ethics. Jensen argues that it is not proven beyond any doubt that Burt deliberately faked his data. However, as described in other chapters, fraud or serious misconduct includes other activities beside making up the data.

For example, Jensen acknowledges that Burt published papers under the names of others, including some collaborators that no one has been able to track (which of course does not prove in a legal sense that they do or did not exist, even though the latter hypothesis seems likely given the fact that the Butt case including the issue of the "missing ladies" was well publicized world-wide). He also concedes that Burt suggested to have collected new data when he most probably did not. A corollary of this is that Burt repeatedly used the same material in different empirical papers, another type of behaviour that is generally considered unethical.

Jensen makes it appear as though such behaviour was "not much out of keeping with the general style of reporting studies in British journals at that time". Moreover, "Burt's main articles ... obviously passed muster with the journal editors and referees at that time". The latter statement seems to me to be incredibly naive.

Much of Jensen's discussion is concerned with the well known issue of the miraculous constancy of Burt's correlations, despite large increases in sample size. The discussion makes a number of points, such as the fact that these correlations do not involve independent samples and that the true correlation is probably in the order of 0.75--0.80. This however misses the main point. What Jensen should have shown (by statistical analysis) is that such a constancy is not in fact unlikely given the above considerations. To me, it is surprising that he has not performed such an analysis.

Jensen discusses a number of points that are too complicated to be evaluated given the few details mentioned. For example, the question of whether Burt's data show a fit to the normal curve that is too good to be true, a point raised by Dorfman (1978). The discussion of this issue is not really clear in Jensen's chapter.

86 Book reviews/Acta Psychologica 88 (1995) 79-86

Jensen argues that Burr normalized his data and expressed them as relative frequencies to a common base. This cannot be true, however, since the latter would involve a linear transformation which does not affect the normality of the distribution. What probably happened, is that Burt transformed the data to normality, a nonlinear transformation. This seems to be the point made by two statisticians, cited by Jensen. However, they as well as Jensen do not seem to realize that such transformations are not allowed if one assumes that IQ is measured on an interval scale.

All in all, this chapter is quite out-of-place in a volume that seeks to create a better understanding of ethical issues in research. The implicit message of Jensen (as long as you do not completely invent the data, anything is allowed), is not the type of message that one would like to convey to young researchers. The editors would have been wiser not to have included this chapter given that its message deviates so strongly from the general attitude expressed in the other chapters.

In sum, I have mixed feelings about this volume. Apart from the poor editing (compare e.g. Jensen's description of the Burt affair with that by Poling), what is lacking most is a clear empirical analysis of the frequency of occurrence of various types of research fraud. Although the editors mention some estimates (such as 12% of the FDA audits prior to 1985), it would have been much more informative if these data had been broken down into various subcategories of misconduct which would have led to a clearer picture of the nature and extent of fraud and serious misconduct in contemporary research.

Jeroen G.W. Raaijmakers Department of Psychology

University of Amsterdam Roetersstraat 15

1018 WB Amsterdam The Netherlands