The vexed question of authorship: views of researchers in a British medical facultyBMJ 1997; 314 doi: http://dx.doi.org/10.1136/bmj.314.7086.1009 (Published 05 April 1997) Cite this as: BMJ 1997;314:1009
- Raj Bhopal, professor of epidemiology and public healtha,
- Judith Rankin, senior research associatea,
- Elaine McColl, senior research associateb,
- Lois Thomas, research associate, Department of Primary Health Care, School of Health Sciences, University of Newcastleb,
- Eileen Kaner, research associatec,
- Rosie Stacy, lecturer in medical sociologyc,
- Pauline Pearson, lecturer in primary care nursingc,
- Brian Vernon, lecturer in the ethics of health carec,
- Helen Rodgers, senior lecturer in stroke medicined
- a Department of Epidemiology and Public Health School of Health Sciences University of Newcastle Newcastle Upon Tyne NE2 4HH
- b Centre for Health Services Research School of Health Sciences University of Newcastle
- c Department of Primary Health Care School of Health Sciences University of Newcastle
- d Departments of Medicine and Epidemiology, School of Health Sciences, University of Newcastle
- Correspondence to: Judith Rankin
- Accepted 17 February 1997
Objective: To assess knowledge, views, and behaviour of researchers on criteria for authorship and causes and control of gift authorship.
Design: Interview survey of stratified sample of researchers.
Setting: University medical faculty.
Subjects: 66 staff (94% response rate) comprising several levels of university academic and research appointments.
Main outcome measures: Awareness and use of criteria for authorship, views on which contributions to research merit authorship, perceptions about gift authorship and strategies for reducing it, and experiences of authorship problems.
Results: 50 (76%) respondents supported criteria for authorship, but few knew about or used available criteria. Of the five people who could specify all three criteria of the International Committee of Medical Journal Editors, only one knew that all criteria had to be met. Forty one respondents (62%) disagreed with this stipulation. A range of practical and academic contributions were seen as sufficient for authorship. Gift authorship was perceived as common, promoted by pressure to publish, to motivate research teams, and to maintain working relationships. A signed statement justifying authorship and a published statement of the contribution of each author were perceived as practical ways of tackling gift authorship. Most researchers had experienced problems with authorship, most commonly the perception that authorship had been deserved but not awarded (49%).
Conclusion: There seems to be a gap between editors' criteria for authorship and researchers' practice. Lack of awareness of criteria is only a partial explanation. Researchers give more weight than editors to practical research contributions. Future criteria should be agreed by researchers and not be imposed by editors.
We assessed the views of 66 medical researchers on criteria for authorship and gift authorship
There was lack of awareness of the criteria for authorship stipulated by the International Committee of Medical Journal Editors
While the concept of criteria for authorship was supported, current criteria were not adhered to, partly because they did not accord with researchers' values
Gift authorship was perceived as a common problem encouraged by the systems for evaluating research and developing careers
Future criteria should be agreed by researchers and editors and should give weight to important practical contributions to research
Publication is the key to academic success. In Britain peer reviewed scientific papers are the main determinant of a university's grading in the Higher Education Funding Council's research assessment exercise, which affects income. The pressure to publish is great, as is the temptation to coauthor papers without having made a substantial intellectual contribution (“gift authorship”), factors linked to fraud and misconduct in medical research.1 2 3
In 1985 the International Committee of Medical Journal Editors published criteria for authorship (see box) based on the principle that each author should be able to defend the work publicly.3 These criteria have been accepted by over 300 journals but are often breached,4 5 with Goodman's results suggesting that some authors may not fulfil any of the criteria.5 These findings raise questions about awareness of the criteria and their appropriateness and acceptability to researchers. The idea of assessing researchers' views on authorship was identified and stimulated by the authorship subgroup of LOCKNET, an international network interested in the processes of publishing medical research.6 7
Criteria for authorship of the International Committee of Medical Journal Editors
Authorship should be based only on a substantial contribution to:
Conception and design or analysis and interpretation of data and
Drafting the article or revising it critically for important intellectual content and
Final approval of the version to be published
The objectives of our study were:
To assess whether researchers knew the criteria for authorship of the International Committee of Medical Journal Editors and, if so, whether they thought they were helpful
To compare researchers' views on contributions meriting coauthorship with those of the committee, to understand why the guidelines are breached, and to propose criteria supported by researchers
To ascertain researchers' current practice in relation to coauthorship of academic papers
To seek views on how the teamwork that does not meet the committee's criteria should be credited
To explore researchers' perceptions of good and bad practice in relation to coauthorship and on the prevalence and nature of gift authorship
To assess views on proposed ways of curbing gift authorship.
The need to present our findings at an international conference on authorship7 imposed strict deadlines for interviewing. We drew up a sample of 70 academic and research staff from the Faculty of Medicine as listed in the University of Newcastle's handbook for 1995-6. As heads of department may influence policy on authorship, we approached all 17 departmental heads designated by the university for interview. To ensure representation of staff with varying experience in publication, we stratified our sampling frame by seniority. A candidate's probability of selection, by random numbers, was proportional to the size of the stratum. We also identified reserves for each candidate.
We invited first choice candidates, by a letter from the dean of medicine and RB, for semistructured interviews addressing the above objectives. Interviewers, all authors of this paper, were allocated to interviewees of similar status. We conducted the interviews, scheduled for 30-40 minutes, between April and early June 1996.
JR and RS read the responses to open ended questions. They identified themes, developed a coding frame, and applied it to quantify these responses. For questions where respondents could make more than one comment, the number of comments is reported. Respondents were categorised according to grade at interview. We analysed the quantitative data with spss.8 To examine the effect of seniority, we used the χ2 test for categorical variables and Kruskall-Wallis analysis of variance for continuous variables.
Nine of the first choice candidates were unavailable: five were no longer in post so reserves were interviewed. In the other four cases, there were difficulties in arranging interviews and no time to approach reserves. We therefore completed 66 interviews (94% response rate). Table 1) shows that the sample achieved a mix of staff and that most respondents had published papers, which was associated with seniority (Kruskall-Wallis χ2=44.8, df=3, P<0.01). The six subjects with no publications were junior researchers.
Knowledge of criteria for authorship
Table 2) shows that, while most respondents thought that there should be criteria for authorship, only 16 were aware of the guidelines of the International Committee of Medical Journal Editors. Of these 16, four could not specify any of the criteria, one wrongly believing that they involved identifying the percentage input of authors, four specified one criterion, two specified two, and five specified all three. Only one person knew that all three criteria were necessary for authorship.
Views on the criteria
Most respondents agreed with each criterion individually but disagreed that all three should have to be met (table 3). A substantial minority misunderstood the first criterion, taking it to require contribution to all four aspects of a study. This was seen as restrictive and unfair, especially to junior staff. The second criterion was thought to be poorly defined and intellectually unsound. Practical barriers were mentioned in relation to the second and third criteria. Twenty comments indicated that the criteria were too restrictive and that greater flexibility was required. Specific reservations included concerns that they could exclude key players in a research team (nine comments) and that they were out of touch with the realities of modern research (four comments). Some respondents felt that in multidisciplinary research teams it was unreasonable to expect each author to defend every aspect of a paper.
Over half of the respondents thought the committee's criteria were not usually adhered to; 21 comments indicated no surprise. Ten comments indicated that breaching the criteria was wrong or unethical, but 15 comments suggested that non-compliance might reflect flaws in the guidelines and that researchers did not accept their authority. Bending the rules rather than deliberate breaching was said to be more likely (five comments). Difficulties in meeting all three of the criteria and the inevitability of breaching them were identified (11 comments), as were ignorance of the criteria and a perception of their irrelevance (four comments). In discussing why the guidelines might be breached, issues of power, status, and “nepotism” were recognised (22 comments). The pressure to publish was also cited (16 comments), and adding senior colleagues was a strategy for increasing the likelihood of publication (four comments). One respondent likened criteria for authorship to the speed limit on roads: “Good that it's there, but not necessarily adhered to.” Others simply did not agree with them: “It's appropriate that they are [breached],” and, “I don't recognise their authority.”
Views on grounds for authorship
Almost everyone thought that both major practical and intellectual contributions were definite or possible grounds for authorship (table 4). One research associate explained: “There needs to be a difference between intellectual contribution and the execution of work. You need to have made a contribution to one or the other but it may not be realistic to have both. The criteria seem to overlook the practical, doing the experiments. They seem to say if you have ideas you become an author, if you work in the laboratory 12 hours a day you don't get on. It should be teamwork.” However, being head of a department, bringing in additional money, providing access to research subjects, or preparing diagrams were rarely thought to deserve automatic authorship. Published acknowledgement was the appropriate credit for contributions not meriting authorship.
Views on gift authorship
Table 5) shows that most respondents thought that gift authorship, defined in the questionnaire as when people included as coauthors have not contributed significantly to the research, should be banned. The commonest reasons given for gift authorship were pressure to publish (25 comments), enhancing chances of publication (16 comments), to repay favours, to motivate a team and encourage collaboration (14 comments), and to maintain good relationships (14 comments). Gift authorship was seen as unethical, dishonest, and unacceptable (11 comments) reducing the paper's validity and diluting the input of those who had made major contributions (11 comments). Those who thought that gift authorship should not be banned believed that it aided collaboration (five comments) and that prohibition would be difficult to achieve (six comments). The respondents thought that requiring researchers to sign a statement justifying authorship and to specify the actual contribution of each author were practical and effective strategies for reducing gift authorship (table 5). Limiting the number of publications listed in a curriculum vitae or a system of fixed credits per publication9 found less favour.
Experiences of authorship
Forty two (64%) respondents had experienced difficulties with authorship, the most common being exclusion from authorship when it was apparently deserved (table 6). Senior staff were more likely to have experienced problems (χ2=15.0, df=3, P=0.001 for any problems), reflecting that they had coauthored more papers (Kruskall-Wallis χ2=44.8, df=3, P<0.01). Almost a third of respondents had assigned inappropriate coauthorship. Problems with authorship were memorable and upsetting.
While these results should be extrapolated cautiously, our medical school is not atypical and our findings are in accord with quantitative work in other centres and with much comment and analysis.4 5 10 Our sample included researchers with widely varying seniority and experience of authorship and reflected the range of disciplines in a modern medical school. The high response rate and the frank responses testify that the respondents welcomed debate on authorship. The semistructured format of the interviews allowed for discussion of the responses.
While the researchers supported the idea of criteria for authorship, many did not adhere to those of the International Committee of Medical Journal Editors, partly because these did not fit with their values. Most respondents disagreed that all three of the committee's criteria should be met since this excluded researchers with important but limited roles. Gift authorship was perceived as an unacceptable and unethical practice that was common and difficult to prohibit. It was seen to be encouraged by academic reward systems.
There has long been concern about multiple and unearned authorship. In 1957 Hewitt noted increasing multiple authorship and argued that bestowing unearned authorship was no favour. The second of his 10 commandments for writing is, “Thou shalt not allow thy name to appear as a co-author unless thou hast some authoritative knowledge of the subject concerned, hast participated in the underlying investigation, and hast laboured on the report to the extent of weighing every word and quantity therein.”11 This principle is echoed by other writers in the phrases “significant contribution,” “intellectual input,” and “public responsibility.” In 1982 Burman observed no correlation between the number of authors and citation and called for criteria for authorship.12 He recommended that order be based on contribution, with the head of the laboratory as the last author. The practice of heads of laboratories and departments being coauthors of all papers from their units is now unacceptable.
The criteria of the International Committee of Medical Journal Editors, therefore, were developed on the basis of debated principles, in response to increasing concerns about gift authorship and fraud in science. They have been widely published, but in our study many researchers were unaware of the criteria, tended to misunderstand them, and perceived them as open to misinterpretation. This is partially understandable because, while the criteria are unequivocal, the discussions interpreting them are not. For example, Huth develops the argument for the criteria by emphasising that public responsibility for the content, including intellectual content, is the key to authorship.13 While technical work, referral of patients, and collection of data alone do not justify authorship, development of the hypothesis, design of the study, and analysis of results may. The detailed discussion contradicts the criteria also given in his text (pages 229 and 44), showing that the criteria alone do not cover adequately a complex matter. Huth's discussion is closer to the views of researchers than the criteria.
Criteria for authorship need to accord more with researchers' values, particularly giving more weight to important practical contributions, which nearly all guidelines consider unworthy of authorship. Researchers views were similar to those of Staheli, who noted that substantial contribution is the key, that excluding key contributors is unethical, and that coauthors include those whose ideas started the study and those who gathered data.14
Problems not addressed by the current guidelines of the International Committee of Medical Journal Editors include ordering of authors and failure to gain authorship when deserved. Most writers on authorship agree that the first author is the one most closely associated with the work. The special status of the last author is recognised by several writers, but the general view is that the order of authors should reflect the rank order of contribution. The Council of Biology Editors recommends each journal specify the criteria for order of authorship.
Our data should help researchers and editors develop a shared understanding. The strategy for communicating and implementing the criteria of the International Committee of Medical Journal Editors has largely failed. New initiatives should engage researchers and meet their legitimate needs. Future guidelines should be developed collaboratively and not be imposed on researchers by editors.
We thank the respondents for being so frank; Professor George Alberti, Dean of Medicine, for support; Dr Richard Smith for calling for the study; Drs Nigel Unwin, Neville Goodman, and other colleagues in the authorship subgroup of LOCKNET for ideas; an anonymous referee whose comments reshaped the paper; and Pat Barkes and Suzanne Young for secretarial support. The order of the authors reflects peer group views on overall academic and practical contribution to this paper and is therefore rank order. The questionnaire is available from JR.