Mutation Research 589 (2005) 17–30 www.elsevier.com/locate/reviewsmr Community address: www.elsevier.com/locate/mutres
Review
Scientific authorship Part 1. A window into scientific fraud? Larry D. Claxton* Environmental Carcinogenesis Division, National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Mail Drop B143-06, Research Triangle Park, NC 27709, USA Received 26 September 2003; accepted 30 July 2004 Available online 23 September 2004
Abstract The examination of a single scientific manuscript seldom alerts scientists, reviewers, editors, and scientific administrators to the fabrication and falsification of data and information. This review shows that most documented cases of scientific fraud involve falsification (altering truthful information) and fabrication (inventing information where none previously existed). Plagiarism is much less frequent. The review of published accounts also shows that the publication of scientific papers containing recognizable fraudulent material is very low, probably less than 0.02% and extremely difficult to detect. Because most reported cases of fraud have involved research done at prestigious organizations with distinguished co-authors, and that is published in journals with exacting review processes, it becomes evident that some unscrupulous scientists are adept at fabricating and falsifying data. However, ‘‘significant’’ scientific fraud is detected when scientists repeatedly report results that cannot be independently verified, when colleagues report suspicious behavior, or scientific audits are performed. This review documents and compares many of the better-known cases of scientific fraud. Fraudulent behavior has served as the impetus for the scientific community to develop publication procedures and guidelines that help to guard against not only fraudulent behavior but also against other types of unethical or undesirable behaviors. A companion paper reviews the non-fraudulent issues associated with scientific publication. Published by Elsevier B.V. Keywords: Falsification; Fabrication; Plagiarism; Fraud; Misconduct; Scientific publication
Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Abbreviated history of scientific fraud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Purpose of review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
doi of the original article: 10.1016/j.mrrev.2004.07.002. * Tel.: +1 919 541 2329; fax: +1 919 685 3281. E-mail address:
[email protected]. 1383-5742/$ – see front matter. Published by Elsevier B.V. doi:10.1016/j.mrrev.2004.07.003
18 18 20
18
L.D. Claxton / Mutation Research 589 (2005) 17–30
2. The 2.1. 2.2. 2.3. 2.4.
issues and problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fraudulent practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples of scientific fraud . . . . . . . . . . . . . . . . . . . . . . . . Estimating the frequency and impact of fraud and plagiarism . Fraud and other unethical conduct . . . . . . . . . . . . . . . . . . . .
. . . . .
20 20 22 23 24
3. Impetus to develop guidelines and oversight groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
4. The impact of guidelines and review upon the occurrence of fraud . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
5. A testable hypothesis for determining the level of fraudulent behavior and identifying practical deterrents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
You may escape detection by falsifying an insignificant finding, but there will be no award. You may falsify an important finding, but then it will surely form the basis for subsequent experiments and become exposed. Daniel Koshland
1. Introduction 1.1. Abbreviated history of scientific fraud As the necessity of publication for professional opportunity and advancement became apparent in the 1970s and 1980s, increasing news of scientific misconduct began to appear, and concerns over matters of authorship (including fraud, plagiarism, and unethical conduct) became topics of open discussion [1–8]. The names of accused scientists and associated institutions became the objects of intense scrutiny. Examples of well-known cases of scientific misconduct are given in Table 1 and in other references [8–15]. In response to mounting concerns, the Federal government in 1989 established two oversight offices: Office of Scientific Integrity (OSI) and Office of Scientific Integrity Review (OSIR) [16]. In 1992, the Office of Research Integrity (ORI)
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
replaced these two offices [16]. An analysis of ORIs efforts during its first 5-year period (1993–1997) showed that ORI: (a) received about 1000 misconduct allegations, (b) conducted 150 investigations, and (c) had 76 findings of scientific misconduct and 74 findings of no misconduct [16]. In the 2001 annual report [17], ORI reported receiving 196 new allegations, opening 35 new cases and closing 25 cases. More than 50% of the latter cases resulted in adverse action by ORI resulting from 10 cases of falsification, 3 cases of plagiarism in combination with falsification and/or fabrication, and 1 case of plagiarism. The other 11 cases were closed without a finding of misconduct. In ORIs most recent annual report [18], 191 new allegations were received, 32 cases were closed, 41 new cases were opened, and 50 cases were carried forward. ORI found misconduct in 13 (41%) of the closed cases, and all misconduct findings involved falsification and/or fabrication of data. The percentage of negative findings for 2001 (56%) and 2002 (41%) was higher than the historical average of 33%. In addition, the 2002 report stated that in 2001, 61 institutions reported 72 new scientific misconduct cases. Obviously, not all misconduct allegations are substantiated. In a 1998 report prepared for ORI, Lubalin et al. [19] stated that 70% of the scientific misconduct cases examined by ORI resulted in exoneration. However, 60% of the exonerated individuals experienced one or more negative con-
Table 1 Examples of alleged cases of major scientific misconduct and controversy Type of case
Articles involved (retracted)
Comments
Reference
Jan Hendrik Scho¨ n, 2003
Fabricated and falsified research findings Fabricated and falsified research findings
25 (28)
See account in text for description
[32,35,37,85,86]
1 (1)
University of North Carolina ad hoc committee implicated author; Leadon denied misconduct but resigned position Committee at Gottingen University found him guilty of scientific misconduct; he left his university hospital position Reviewer noted irregularities in submitted paper; Collins checked data, repeated experiments; and quickly informed university, ORI, colleagues; Hajra barred from federally funded research and denied doctorate Involved 1980–1984 studies while at U. Pittsburgh; 1987: NIMH discloses misconduct. 1988: pleaded guilty in court Notifications sent to journals, support-granting bodies, and co-authors
[87–89]
Steven A. Leadon, 2003
Alexander Kugler, 2002
Lack of proper data handling and record keeping
1 (1)
Amitov Hajra, 1996
Fraudulent data
6 (3) plus bogus nucleotide sequences in GenBank
Stephen E. Breuning, 1987
Fraud: reported studies never done
20 (2)
Robert Slutsky, 1985
Fraud: fabricated data
John Darsee, 1982
Fraud: fabricated data
55 questionable; 13 considered fraudulent; 18 retracted; reports vary on these numbers At least eight journal articles and many abstracts were fraudulent
Elias Alsabti, 1980
Fraud: Plagiarism
Waksmans–Schatz 1940’s
Waksman did not share credit
Copied articles of other authors, replaced names of authors with his, and sent articles to obscure journals Schatz, a Ph.D. student, was co-discoverer of streptomycin. Waksman alone won Nobel Prize due to lack of mention of Schatz
Lost Harvard position; banned from NIH research grants for 10 years; Harvard repaid grant money Asked to leave several different institutions when practice discovered. Finally left the U.S. Schatz took legal action to obtain royalties and won an out-of-court settlement. Many in scientific community were antagonistic toward Schatz. Debate continues on whether Schatz should have shared Nobel Prize or if prize was for Waksman’s larger efforts
[90]
[34,91,92]
[10]
[93–95]
[28,96–100]
[9,49]
L.D. Claxton / Mutation Research 589 (2005) 17–30
Identifier, year of disclosure
[101]
19
20
L.D. Claxton / Mutation Research 589 (2005) 17–30
sequences of being accused [19]. Most of the allegations and findings center upon publication issues, because scientific publication documents the actions of the researcher. Most discoveries of fraudulent science, in the past, have occurred after manuscripts or reports were published. Therefore, some ask, ‘‘Why did peer review not uncover the fraud before the manuscript was published?’’ Some also ask, ‘‘What was the responsibility of the co-authors and the sponsoring organizations? And did these co-authors and organizations fail in their responsibilities?’’ 1.2. Purpose of review This paper and its companion paper are offered not because there is a crisis in ethical behavior but because an informed scientific community can better meet the challenges surrounding authorship issues. Although there may be no crisis in ethical behavior, the problems surrounding scientific authorship continue to grow because of the increase in numbers of scientists, the complexity of modern science, the increase in competition for support and recognition, and the influence of entrepreneurial endeavors. Most of the literature on scientific fraud is anecdotal or informal. The reason is that most papers, editorials, commentaries, and reports dealing with authorship arise out of the need of editors and others to change practices in order to diminish abuses (fraudulent papers, papers with improper attribution of authorship, etc.) [5,20]. Historically, most articles, comments, and editorials dealing with authorship issues were published in medical journals (e.g., British Medical Journal (BMJ), Journal of the American Medical Association (JAMA), and Lancet) and broadly based journals (e.g., Science and Nature) [21]. In addition, most references deal with or emerge from some type of abuse or problem, and they are written by those confronting the problem (e.g., editors) or by those recommending specific changes or viewpoints. These companion papers, written from a different perspective, strive to inform scientists (especially younger scientists), editors, reviewers, and scientific administrators about the bad news and the good news regarding scientific publication. This awareness helps all participants to understand the increasing efforts aimed at regulating and creating guidelines for scientific conduct and publishing. Because most
scientists probably believe ‘‘it won’t happen to me’’ [22], awareness of the bad news (past abuses, struggles, and divergent opinions) creates a needed awareness as one collaborates upon, writes, and reviews scientific papers. The good news (seen in the emerging consensus among published recommendations and guidelines) reassures everyone that the scientific community actively pursues truth, the betterment of humanity and the environment, and ethical conduct. Although Part 1 of the review provides an overview that others can use in their deliberations, the primary focus of this paper is explore whether or not normal manuscript preparation and review processes tend to discover fraudulent science. Secondly, I will show that fraudulent misconduct has led the scientific community to develop mores and guidelines that prescribe professional conduct in the routine developing and publishing of scientific works (the topic of the companion paper).
2. The issues and problems 2.1. Fraudulent practices Fraudulent practices are deceptive practices in which an author with forethought sets out to deceive others. Protti reviews the legal aspects of scientific misconduct [23]. Legally, scientific fraud is a deliberate misrepresentation by someone who knows the truth [23]. This definition does not include contradictory or misguided interpretations, mistakes, poor scientific and unprofessional practices, or even negligence, which can also cause truth to be misrepresented or misinterpreted [23]. Because fraudulent behavior involves some type of deliberate dishonesty and unethical behavior, these practices are the ones that do the most damage to the name of science and are the hardest to guard against. For example, one who fabricates data in order to circumvent the normal processes of scientific inquiry and reporting may misdirect the research of others, damage the reputations of co-workers, sponsoring organizations, societies, and journals, and lessen the image of scientists in the eyes of the public [10,24– 32]. A reviewer and/or an editor, however, may not be able to discern whether or not a graph, for example, was produced from fabricated data or was the result of honest, laborious research.
L.D. Claxton / Mutation Research 589 (2005) 17–30
21
Table 2 Summary of Office of Research Integrity 2001 and 2002 findings of misconduct Name, degree, position, organizationa
2002 ORI report T. Arichi, Ph.D., NCI/NIH D.C. deSales, SFDPH M.B. Ganz, M.D., CWRU A. Handa, M.D., Ph.D., Visiting Fellow, NHLB/NIH
Typeb
Fal, Fab Fal Fal, Fab Fal, Fab
M.A. Lipski, WUSL A.J. Morrow, B.S., Grad. Student, SLU H.J. Muenchen, Ph.D., Postdoc, UM
Fal, Fab Fal, Fab
S.F. Munjee, Res. Fellow, WFU-SM J.C. Pennington, Grad. Stud., BU M.R. Prasad, Professor, UK-SM M. Shishov, M.D., Lab tech, BWH R.B. Tracy, Ph.D., Student and Postdoc, USC/UC-D Z. Yao, M.D., Ph.D., Assoc. Prof., UNC-CH
Fal, Fab
2001 ORI report S.F. Arnold, Ph.D., Assist. Prof., TU J. Elster, Undergrad. Res. Assist., SLU D.R. Jacoby, M.D., Ph.D., Instructor, HU, MGH
Fal, Fab
Comments
No.c
C-Ad
Journal(s)
1
7
PNAS
2
4
J. Gen. Virol.
3
5
Clinical Cancer Res., Urology, J. Biol. Chem., 2 draft papers
1
1
Brain Res.
Fal, Fab Fal, Fab Fal, Fab
Arichi is first author Contractor report Grant application One paper retracted by single co-author; other flagged but not listed as retracted Grant report Poster presentation, Draft manuscript, NIH application First and corresponding author; both retracted; J. Biol. Chem. retracted by another author NIH application, abstract for AACR Grant report Retracted by co-author Laboratory entries; some mitigating circumstances Both papers retracted; also for thesis
Fal, Fab
2
2
Science, EMBO J.
Fal, Fab
1
2
J. Cell. Mol. Cardiol.
Plus other multiple applications; cited by seven papers as late as 2004
Fal, Fab
1
5
Science
Arnold is 1st author; paper was retracted
Fal, Fab Plag, Fal
K.-F. Lin, DVM, Grad. student, MUSC
Fal data
S.F. Munjee, M.S., Res. Fellow, WFU-SM D.A. Padgett, Ph.D., Assist. Prof., OSU R.S. Pandurangi, Ph.D., Assist Prof., UM-C K.M. Ruggiero, Ph.D., HU A. Saleh, Ph.D., Postdoc, UP
Fal, Fab
D.D. Sanchez, Res. Assist., PHFE M. Sarker, MBBS, MPH, Doctoral fellow, UA-B
Journal articles
3
3
Hypertension, Hypertension Res., Human Gene Therapy
Grant applications and presentation; cover-up attempted Errata rather than retractions were submitted to journals
Plag
Grant application; abstract to AACR Grant application
Plag, Fal
Grant application
Fab data Fal data
Fal and Fab data Fal data
2 1
3 9
Nature
Grant applications Did not accept or admit misconduct but published erratum PHS report For thesis
22
L.D. Claxton / Mutation Research 589 (2005) 17–30
Table 2 (Continued ) Name, degree, position, organizationa
Typeb
Journal articles c
S. Smith, Res. Technician, UC-SF
Fal and Fab data
V. Valentin, Counselor, BU-SM M. Xiong, Ph.D., UT-HSCH
Fab data Plag and Fab data
d
Comments
No.
C-A
Journal(s)
10
Not provided
Report says all 10 retracted Cooperative agreement report Grant application
a BU: Brown U.; BU-SM: Boston U. School of Medicine; BWH: Bringham and Women’s Hospital; CWRU: Case Western Reserve U.; MGH: Mass Gen. Hospital; MUSC: Medical U. South Carolina; NCI/NIH; NHLB/NIH; OSU: Ohio State U.; PHFE: Public Health Foundations Enterprises, Inc.; SFDPH: San Francisco Dept. Public Health; SLU: St. Louis U.; TU: Tulane U.; UA-B: U Alabama, Birmingham; UC-D: U. of California, Davis; UC-SF: U. California, San Francisco; UK-SM: U. Kentucky School of Medicine; UM: U. Michigan; UM-C: U. Missouri, Columbia; UNC-CH: UNC-Chapel Hill; UP: U. Pittsburgh; USC: U. Southern California; UT-HSCH: U. Texas Health Science Center at Houston; WFU-SM: Wake Forrest U. School of Medicine; WUSL: Washington U. St. Louis. b Fal: falsification of data; Fab: fabrication of data; Plag: plagiarism. c Number of journal articles shown to be affected. d C-A: total number of co-authors on implicated papers.
2.2. Examples of scientific fraud Tables 1 and 2 identify some instances where researchers were shown to be guilty of fraudulent misconduct. Table 1 provides examples of highly visible cases examined by the popular and scientific press. Table 2 summarizes the cases of misconduct as reported in the 2001 and 2002 ORI Annual Reports. While Table 1 reports the more flagrant and perhaps more interesting cases, Table 2 represents the more common types of occurrences. The majority of cases were incidences of falsification (substituting for or altering truthful information) or fabrication (creating records, data, or information where none previously existed). Plagiarism has also been reported (Table 1). Because the scientists listed in Table 1 were able to defraud their colleagues, reviewers, and editors in order to have their papers published in highly visible journals, they are examples of how fraud can go undetected for at least a significant period of time. For example, Wade [33] points out that during John Darsee’s 13 years of successful fakery, he spent 2 years in two leading heart centers in the U.S. I will highlight two cases of fraud as case studies. Because Table 2 presents a broader sampling of fraudulent activity, I will provide a summary characterization of these cases. In the Amitov Hajra case (Table 1), Dr. Francis Collins found himself as the unwitting co-author of fraudulent scientific manuscripts [34]. A Ph.D. candidate had fabricated data that appeared in a paper submitted to Oncogene on which Dr. Collins and his associates were co-authors. An astute reviewer for the
journal Oncogene recognized that protein markers in a Western blot appeared fabricated. After being informed by the editor, Dr. Collins and a colleague spent 2 weeks checking the work of the Ph.D. candidate. They concluded that the student had falsely manufactured the data. Upon further examination, they determined that Mr. Hajra fabricated data that had appeared in five published papers. Two of these previous papers had to be retracted, and the other three had to be corrected. All other co-authors were exonerated. The expertise of the laboratory, peer reviewers, and journal editors was not sufficient to recognize any suggestion of fraud for the five published papers. Only careful laboratory review and confessions of the scientist himself identified all of the illegitimate information. Between 1998 and 2002, the German scientist Dr. Jan Hendrik Scho¨ n (Table 1) established himself as a highly visible and productive scientist in the field of nanotechnology [26,35–37]. During this period while employed by Bell Labs, he was an author on over 90 papers, including 15 papers in Science and Nature [32]. He was the lead author on 74 papers [32]. His efforts energized new areas of physics, and it was estimated that over 100 laboratories worldwide were trying to follow his lead [32,38]. Some thought he was headed for the Nobel Prize [26,32]. However, his results could not be independently verified in spite of millions of dollars being used to duplicate his reported results [32]. Then in April 2002, Dr. Lydia Sohn, Princeton University, receiving a tip from a colleague at Bell Labs, examined separate papers found in Science and Nature [35,39]. Upon finding problems
L.D. Claxton / Mutation Research 589 (2005) 17–30
with the figures in the paper, she notified the two journals. About the same time, Sohn and others found suspect figures in six other papers [35,40]. Upon becoming aware of these problems, Bell Labs announced that an independent review committee would investigate the matter [32,35,38,40,41]. Scho¨ n denied all allegations of misconduct [35]. Then in September 2002, the panel’s report concluded that Scho¨ n had duplicated, falsified, and destroyed data [36,37,42]. Bell Labs dismissed Scho¨ n upon receiving the panel’s report [37,42]. Although admitting to mistakes, he continued to defend his work [42]. The panel found that there was a ‘‘preponderance of evidence’’ for falsified or fabricated data in 25 papers [31,37,42]. By doing a Current Contents Search1 (5 May 2004) using Dialog1, I was able to locate 28 retractions in a total of seven journals. Again, the expertise of many scientists was not sufficient to recognize any suggestion of fraud for many manuscripts. Discovery of the fraud was enabled when many scientists realized that they could not replicate the studies of Dr. Scho¨ n, a tip was received from one of the scientist’s colleagues, and brilliant ‘‘detective’’ work was done by a group of scientists who compared results found in multiple publications. Within the ORI reports for years 2001 and 2002 [17,18] (summarized in Table 2), 27 cases of verified fraud were reported. Of these 27 cases, 23 involved falsification, 20 involved fabrication, and 4 involved plagiarism. Eleven cases involved a total of 27 articles published in peer-reviewed journals, 8 grant applications, 4 grant or contractor reports, and 2 theses. Most of these cases apparently came to light due to colleague or institutional management suspicions or to data audits [17,18]. Once again, the normal review of manuscripts did not tend to uncover fraudulent behavior. Also, none of the co-authors were implicated in the misconduct. 2.3. Estimating the frequency and impact of fraud and plagiarism ORI classifies approximately 15–20 cases per year as scientific misconduct [17,43–45]. For the years 2000–2002, the PubMed database recorded 78 retracted articles and 60 letters of retraction; and not all of these were for misconduct. Therefore, fewer than 30 scientists published less than 80 scientific papers per year out of over 400,000 (0.02%) in the
23
NIH funded or associated biomedical sciences containing information known to be fraudulent. Such data suggest that the percentage of researchers engaged in this level of scientific misconduct is probably very small. One should note, however, that many of the papers discovered to be fraudulent originated in major research institutes [46,47] and were in high-impact journals (Tables 1 and 2) that have large readerships, making them more likely to be discovered. Using journals that I frequently read, I surveyed (Table 3) the number of published retractions over a 10-year span (1994–2003) for 11 journals. Three journals (BMJ, Nature, and Science) published a total of 20 retractions. Nine of these retractions were for reasons other than fraud. Nine of the 11 fraudrelated retractions were due to a single individual (Jan Hendrik Scho¨ n). Therefore, out of the 75,330 PubMed citations for these journals (assuming that 80% or 60,000 represent scientific manuscripts), the percentage of fraudulent papers is 0.018% (11/75330 0.8). Considering only the three journals with retractions, the retraction rate is approximately 0.022% (11/62050 0.8). An upper limit for the incidence of research misconduct plus scientific inaccuracy can be estimated by observing the results of scientific audits. Shapiro and Charrow [48] report that between October 1985 and April 1988 FDA audits of investigational drug trials detected 7% with serious deficiencies; however, these deficiencies included non-malicious errors, negligence, and/or poorly done science as well as fraud. Some have called these ‘‘disreputable errors’’ because they emanate from careless research procedures and the disregard of established procedural standards and norms [5]. The ORI data (Table 2) shows that reviews of NIH-associated grant applications, reports, and other associated materials (contractor reports and theses) accounted for 7 of the 27 total cases. Therefore, it appears that the incidence of unpublished fraud (that has some potential for publication) is no greater than published fraudulent papers. If we assume, therefore, that that the number of fraudulent papers that go undetected is approximately equal to the number detected, then one would estimate about 40–80 fraudulent papers (0.02%) are published each year in journals covered by PubMed. If this is a 10-fold underestimate, then approximately 0.2% of published papers may contain fraudulent
24
L.D. Claxton / Mutation Research 589 (2005) 17–30
Table 3 Survey of retractions in selected journals Journal, PubMed citationsa for years 1994–2003
First authorb
Year published
Year retracted
Reasonc
BMJ, 26627 (0.004% fraud, 0.004% error)
Williams [22,102] Hawthorne [103]
1993 2000
Fraud Error
Vigilant [104] Altamiranod [105] Kawasaki [106] Schon (7 articles) [107–113]
1997 2000 2003 2000–2001
Gowene [87] Van der Zee [114] Schon (2 articles) [115,116] Trotochaud [117] Nitta [118] Ricaurte [119] Carelli [120]
1998 1996 2002 2000 1998 2002 1997
1998 2003 None None None None 2001 2002 2003 2003 None None 2003 1999 2003 2003 1999 2003 1998 None None
Envir. Mol. Mutag., 912 Fund. Appl. Toxicol., 614 Mutagenesis, 820 Mutat. Res., 5339 Nature, 17651 (0.02% error, 0.04% fraud)
J. Appl. Toxicol., 725 J. Toxicol. Environ. Health, 424 Science, 17772 (0.02% fraud. 0.03% error)
Toxicology, 2168 Toxicol. Lett., 2278 a b c d e
Error Error Error Fraud
Fraud Error Fraud Error Error Error Error
Number of journal citations in PubMed for years 1994–2003. First author of paper and not necessarily one implicated for fraud and/or error. Error includes all reasons other than public recognition of fraud. First author disagreed with retraction. Last author (not first author) implicated in fraud [total citations in none journals = 13280,
