Roger Clarke's Web-Site
© Xamax Consultancy Pty Ltd, 1995-2013
|Identity Matters||Other Topics||Waltzing Matilda||What's New|
Review Draft of 23 April 2009
Published in Commun. AIS 25 Article 19 (July 2009) 155-164, Special Section on Journal Self-Citation
Roger Clarke **
© Xamax Consultancy Pty Ltd, 2009
Available under an AEShareNet licence or a Creative Commons licence.
This document is at http://www.rogerclarke.com/SOS/SCSP-09.html
When is it reasonable to re-use work of your own, particularly if it has already been published? And when is it appropriate to cite your own works, and when is it inappropriate? Rather than being mysteries, this paper suggests that basic principles of academic communication and professional ethics provide a framework for each of us to make our own decisions, and to evaluate actions taken by others.
This Special Issue considers various aspects of a practice among some journals of requesting authors to add citations to previous articles in that journal - a practice referred to [in this Issue? in this paper?] as 'journal self-referencing'.
Consideration of this practice in isolation is problematic. It is one facet of the general problem of how authors select content and references for inclusion in their papers. This contribution focusses on several other facets of the general problem, and demonstrates how basic principles give rise to practical solutions to all of them.
The paper commences by reviewing approaches to plagiarism generally and to authors' decisions about what references they should cite in a paper. It then dissects the notion of 'self-plagiarism', and shows how each of the various issues are amenable to solution by application of some straightforward principles. Some forms of self-plagiarism can be better described as self-citation. The basic principles are applied to distinguish appropriate and inappropriate self-citation behaviour by authors and, finally, by journals.
The context in which this contribution is set is the selection of content for inclusion in an academic paper. Among the vast numbers of guidelines for preparing research papers, most spend much more space on the accuracy of citations than on the more important question of how to decide which references to include and exclude. A small set of norms is postulated in Exhibit 1, based on a scan of a variety of sources.
Although positive guidance about what to cite is in short supply, a great deal has been written on its complement: the negative concept of plagiarism. In Clarke (2006), the following definition of plagiarism was adopted, drawn originally from AVCC (1997): "To plagiarize shall be understood to mean the presentation of the documented words or ideas of another as his or her own, without attribution appropriate for the medium of presentation. ... A researcher or reviewer shall not intentionally or recklessly ... plagiarize". From that definition, the following elements of the notion of plagiarism could be identified (text verbatim from Clarke 2006):
"The wrong in plagiarism lies in misrepresenting that a text originated from the person claiming to be its author when that person knows very well that it was derived from another source. ... The plagiarist tries to take undeserved credit for an accomplishment that belongs to another person" (Samuelson 1994, p. 24). Plagiarism represents failure to cite when a citation should have been provided. Many sources imply that all such failures are worthy of the academic equivalent of capital punishment. The contrary position was argued in Clarke 2006): the seriousness of an act of plagiarism varies a great deal, and along multiple dimensions.
This paper's focus is not on plagiarism per se, but on 'self-plagiarism'. Uses of the term are generally intended to invoke the pejorative tone of the root-word. However, as the following sections will make clear, the relationship between the two ideas is not simple.
The term 'self-plagiarism' is often used without sufficient care about its meaning. This section identifies multiple variants of the practice, within two broad categories.
On the basis of the definition of plagiarism discussed in the previous section, 'self-plagiarism' can be defined as 'the re-presentation of the documented words or ideas of oneself, without appropriate attribution'.
The following variants can be usefully distinguished, based on the scale of the material that is re-presented.
Occasional examples come to light of a paper being published, in identical form, in two journals. This is prima facie a grievous breach of ethics, because it represents fraudulent claims of originality. The breach is all the more serious if there is evidence of active intent to defraud, such as the provision of a different title or abstract, but an identical body.
In Hexham (1992), this is referred to this as 'recycling fraud', because "the argument, examples, evidence, and conclusions remain the same without the development of new ideas or presentation of additional evidence. In other words it is recycling fraud when two works only differ in their appearance but are presented as separate and distinct works". For Roig (2006, pp. 17-18), it is 'redundant publication' or 'duplicate publication', which he notes has become a problem in the biomedical literature. In some contexts, "multiple publication of the same scientific work in more than one journal [is designated as] a serious deviation from accepted practices and as actionable misconduct" (Samuelson 1996, p. 25, referring to a law journal article that ascribed this to the U.S. National Science Foundation).
Even with this fairly extreme form of breach, however, some mitigating factors may exist. One example is a multi-authored paper, with different authors submitting separately to different journals. Another, more reasonable scenario is submission to the editors of two journals in distinctly different areas (e.g. one journal tightly focussed on a discipline, and another that adopts a multi-disciplinary approach to a research domain). This demands declaration at the time, by the author, to both editors. If both editors in due course accepted the paper for publication, each needs to declare the parallel publication to the journal's readers. Similarly, publication in a second journal of an article directly translated from an original, or translated with only limited enhancements, requires declaration to the editor and to the readership (Roig 2006, p. 18).
A special case in this category is the presentation of a conference paper, together with submission to a journal, commonly after the conference, but possibly in parallel with it. The norms vary among disciplines and among journals. As noted in Collberg & Kobourov (2005, p. 91), ACM and IEEE have formal policies on the matter, in the ACM's case indicating 25% new material as a norm. The policies of JAIS and CAIS, on the other hand, are merely that "the authors must certify that the manuscript ... is not currently under review in any other journal or conference". On the surface, JAIS and CAIS accept post-conference submissions, but preclude parallel submissions (although the policy does not preclude an author putting forward a case that a particular paper justifies an exception being made).
Moreover, conferences with high standing frequently have arrangements in place with journals for post-conference publication, including fast-tracking. A common expectation is that the paper submitted to the journal will have been further developed, to reflect feedback received in the interim. It remains to be seen whether the maturation of electronic publishing will result in major changes to both journal papers and journals, and perhaps a gradual merger of conference and journal venues (Clarke & Kingsley 2008).
A further special case might be referred to as 'formal re-publication', e.g. in books that gather together previously-published papers on a theme. Re-presentation in itself is no sin, provided that the fact that it is a re-presentation is declared at the outset to the (re-)publisher, and to the reader. (If the author is not the copyright-owner, a licence is also needed, of course). Declaration in such cases cannot be achieved through a citation and an entry in the Reference List. A separate Declaration is needed, preferably both in the Editor's Introduction and before or after the text of the article.
A new paper may use a 'substantial portion' of a prior work. Examples include:
The term 'substantial portion' has been adopted here in full knowledge that it has a meaning in copyright law. Copyright law is distinct from the primarily ethical issue of plagiarism (Clarke 2006). This paper focusses on ethical issues, and generally leaves the separate legal questions to one side. In this case, if the author is not also the copyright-holder (e.g. if the copyright was assigned to the publisher of the journal in which the original paper appeared), then re-publication of a 'substantial portion' will likely be in breach of the publisher's copyright unless a licence to do so is sought and obtained.
Where a substantial portion of a previous work is re-used, it is unequivocally necessary for reference to be provided to the previously-published work. It should be clear to the reviewer, editor, publisher and reader, from the text, which portions have previously been published.
This is very common in several circumstances. One is the common phenomenon of 'one research program begets multiple papers', because a proportion of each (such as the background to the study) is common to all of them. Another circumstance is publication of closely related material in two venues for two very different audience, e.g. in academe and in professional practice; in two distinct disciplines; and in a discipline and a multi-disciplinary research domain. Samuelson (1994) addresses this in some depth.
Publishers, editors and reviewers may express concerns about a variety of forms of re-publication of a substantial portion of a work. A vignette is presented in Appendix 1, drawn from one specific incident that confronted the author. Over an extended period of time, the author has encountered nervousness in relation to all of the following factors on various occasions:
It can be reasonably anticipated that journals will evidence quite varied behaviour in relation to such questions during the coming years, de-stabilised as they are by rapid changes in electronic publishing and diversity among journal-publishing business models. As discussed in Clarke & Kingsley (2009), there is a very real risk that the promise of open and early accessibility of research enabled by the digital and Internet revolutions may not be fulfilled. Desirable change is being undermined by conservative attitudes among many journal publishing organisations. These arise from their desire to sustain business processes and a scale of resourcing that are still attuned to an earlier era of high revenues from captive library subscriptions.
The term 'substantial' is vague, but implies both quantity and substance. In some circumstances, a small quantity may nonetheless be of sufficient substance that attribution is warranted, if only as a matter of self-defence. Such 'small but significant portions' include critical passages of text, a critical concept and even a key phrase or neologism, but particularly key diagrams.
Attribution in the form of a citation is unnecessary in the case of other short passages, simple diagrams, segments of diagrams, phrases, or unremarkable neologisms. Moreover, unless there is a reason for a citation to be provided (as defined in Exhibit 1), it is unethical to do so, because it would be of the nature of patronage, or more precisely 'self-patronage'.
An important inference from the above discussion is that there is a fine line between, on the one hand, an inappropriate failure to attribute to one's own prior work and, on the other, inappropriate attribution to one's own prior work. Professional jealousies play an important role in preventing individuals from in effect recommending themselves into a respected position within a discipline. Reputation is earned, not promulgated; and standing is granted by one's seniors and peers, not asserted.
A further form of inappropriate behaviour that is sometimes referred to as 'self-plagiarism' is a little different from that described in the previous sub-section. This can be defined as 'the excessive or unjustified re-presentation of the documented words or ideas of oneself, with appropriate attribution'.
This differs significantly from the previous category, in the following ways:
As with plagiarism generally, care is needed before a conclusion is reached that a particular instance is appropriately classified as this category of self-plagiarism. There are many circumstances in which re-use of material is justified, on the grounds that appropriate, precise and clear expression has been previously prepared, and is as relevant in the new paper as it was in the paper in which it was originally published. An important example of such a circumstance is exposition of a particular aspect of existing theory, particularly an aspect of theory that was an original contribution by that author in a previous work. Another example is the description of the research method used in a project which has given rise to multiple articles.
It is not uncommon in the IS discipline for a single body of empirical data to be used as a basis for multiple papers. Care is needed, however. Roig (2006, pp. 19-20) refers to this disparagingly as 'salami-slicing'. Referring to the biomedical literature, he describes 'data fragmentation' as a practice whereby the quality of research reporting is reduced because a body of knowledge that would have been valuably communicated in a single paper has been separated into two or more papers. He also criticises 'data augmentation', where a second article reports on data from a follow-up study, in combination with data from an earlier study. In each case, the sin is the failure to make the facts clear to the reader (and, from the outset, to the editor and reviewers). If, however, a paper makes a sufficient contribution, and refers to and cites prior papers in the series, Roig's criticism of 'salami-slicing' must be at least moderated, and possibly entirely withdrawn.
A particular challenge arises where an author wishes to re-use material that is in an earlier working paper or PrePrint, but that paper has yet to be formally published. On the one hand, it lacks the authority of a previously-published refereed work. On the other hand, to omit it would be to implicitly claim originality. Where an author decides it is appropriate to re-use material in such circumstances, a citation to the working paper or PrePrint should be provided, and a note included in the Reference List to the effect of 'currently before the referees of <journal-name>/<conference-name>'.
As with the previous category of self-plagiarism, application of the norms identified in Exhibit 1 is sufficient to enable authors to judge whether or not it is reasonable to incorporate material from their own earlier publications.
The various malpractices discussed above under the heading of 'self-plagiarism' blend across from 'failure to provide attribution when it was necessary to do so' to 'provision of attribution when it was not justified'. At the extremity, it therefore becomes a question of the appropriateness of self-citation. This section considers the ethics firstly of authors citing their own prior works, and then returns to the Special Issue theme of 'journal self-referencing'.
Authors seek to create and sustain their 'brand'. This is achieved by being noticed. An author can be noticed because of what they publish, how often they publish, the locations in which they publish, the specific citations that their publications garner in later papers, and overall citation-counts.
Self-citation has some attractions as part of the process of brand management. It draws previous papers to the attention of reviewers, editors and (if the new paper is accepted for publication) further readers. Depending on the rules applied by the analyst, they may be included in citation-counts. (They are, for example, in Google Scholar counts). Moreover, as discussed above, there are circumstances in which self-citation is necessary, and others in which it ought to be at least permissible.
On the other hand, self-citation is generally regarded with distaste by reviewers and editors, because it smacks of self-promotion, and carries with it risks similar to the less acceptable forms of self-plagiarism discussed earlier. The considerations involved were canvassed in a reflective Editorial in a specialist journal (Readence & Barone 1996).
The norms identified in Exhibit 1 can again be applied to provide guidance, as follows:
These principles are as applicable to the author's own works as to those of other people, with the exception that 'argument from authority' needs to be conducted somewhat more cautiously. It is incumbent upon the reviewers to scrutinise such citations a little more carefully, in order to identify gratuitous mentions, misleading suggestions of intellectual dependency, and invocation of unrefereed sources invested with spurious authority. Reviewers can be reasonably expected to be especially harsh on an author who commits more than minor transgressions of these kinds. Rather than building brand, such practices seriously undermine it.
This has an important implication for so-called 'blind reviews'. These have been adopted by many journals, with the expectation that reviews will be of higher quality if the reviewer can be prevented from knowing the identity and affiliation of the author. However, one of the inadequacies that reviewers need to filter out is inappropriate self-citation. It is therefore important that reviewers see not only which citations are to the author's own papers, but also the venue in which those papers were published, and in some circumstances the title and even the content of the cited papers. Yet the 'blind review' norm actually prevents reviewers from performing this function, because it demands replacement of all details of self-citations with an empty 'AUTHOR (Year)' entry. In short, the term 'blind review' is apt in a manner that its proponents never intended: it ensures that the reviewer is blind to inappropriate self-citation - and hence reduces the quality of reviews in the important area of self-citation.
As a reviewer for venues that prefer 'blind review', I endeavour to respect the express or implied request that I not consider the author's identity as a factor in the review. However, if after the first reading of the paper I judge that the author's prior work is important to the new paper, I actively seek out the prior literature (by means of searches on keywords, informed by my knowledge of the field). I can usually identify the relevant works and in the process inevitably become aware of the author's identity. I complete my review, taking the relevant aspects of the prior work into account. I then declare the facts to the editor/program chair, and leave it to them to use, de-value or reject the review, as they see fit. I acknowledge that this practice raises ethical issues, but I contend that the sanctity of the nominal blindness to author identity is a lower-order value than the quality of the review.
Journal editors also need to build and sustain brand. They therefore have an incentive to ensure that new papers cite previous papers in the same journal. That has direct benefits (because the journal's use as an authority will be noticed by readers) and indirect benefits (because it will increase citation-counts for previous papers and for the journal as a whole).
Applying the norms expressed in Exhibit 1 to this context leads to the conclusion that there are forms of behaviour motivated by that incentive that are ethical, in particular:
Equally, the norms lead to the conclusion that it is unethical behaviour:
A separate contribution in this Special Issue considers the question of the application of the AIS Code of Research Conduct to misbehaviour of those kinds.
A range of possible approaches can be taken to dealing with the various forms of misbehaviour discussed in this paper. Formal legal approaches are largely unavailable, unlikely to become available, and in any case undesirable. Progress is more likely to be achieved through refinements to existing, semi-formal processes within the discipline.
The primary approach is through the quality assurance process. Reviewers and editors need to be sensitive to the issues and norms, apply the principles, and ensure that authors apply the principles.
A key means of ensuring ongoing visibility of the principles is through encapsulation in the AIS Code of Research Conduct. Further, the terms of the Code need to be binding on all AIS members, processes need to exist whereby accusations of inappropriate behaviour can be investigated, and mechanisms are needed whereby members who commit serious breaches are sanctioned.
The current version of the AIS Code expressly addresses self-plagiarism of works, saying:
Do not submit for publication or presentation articles or papers you have already published elsewhere. ... you should not submit a manuscript which is identical or very similar to work you have published previously (or which has been accepted elsewhere for publication). ... There are naturally exceptions to the above guidelines for reprints of an article in an edited collection or book. ... Presenting a paper at a conference to obtain comment and discussion, and then later revising the paper for submission to a journal is another legitimate exception. However, in such cases, prudence suggests that you alert the editor in your submission letter and in the article draw the reader's attention to the conference paper ... (AIS 2003).
The 2009 revision to the Code, currently in process, broadens the expression somewhat, to "Do not make misrepresentations to editors and conference program chairs about the originality of papers you submit to them" (AISTF 2008b). It retains the substance of the Code and expands the advice about exceptions.
The Code also addresses self-plagiarism of substantial portions, saying:
... you should not attempt to build a new article largely from a re-working of your previous publications. Even this advice is subject to exception - as when a scholar re-weaves the threads of previous thought to reveal new patterns, perspectives or insights, or seeks to provide a comprehensive summary or 'state of the art' report on a particular research stream" (AIS 2003).
The 2009 revision clarifies the context within which the exceptions arise, by adding the qualification "unless there is a sufficient new contribution" (AISTF 2008b).
The major contribution of the revised Code is the Recommendation to Council and the membership that all members be bound by the Code, in all of the roles that they play (AISTF 2008a). This ensures that not only authors, but also reviewers and editors, should shortly become subject to institutionalised controls over the inappropriate aspects of self-plagiarism and self-citation discussed in this paper.
The practice of 'journal self-referencing' can be better appreciated by viewing it as a particular instance of a broad notion of self-promotional activities that authors, as well as editors, are tempted to engage in.
The analysis conducted in this paper demonstrates that these practices do not exist in a theory-free vacuum. Existing knowledge can be readily applied to the range of ethical challenges that confront authors and editors alike. With the maturation of the discipline, norms are increasingly clear, they are being codified into a quasi-legal instrument, the AIS Code of Research Conduct, and that Code provides a basis on which each of us can consider our own possible courses of action in advance, and evaluate the actions of others.
AIS (2003) 'Code of Research Conduct' Association for Information Systems, September 2003, at http://home.aisnet.org/displaycommon.cfm?an=1&subarticlenbr=15
AISTF (2008a) 'Report to Council by the Task Group to Review the Code of Research Conduct' Association for Information Systems, November 2008
AISTF (2008b) 'Code of Research Conduct' Task Force to Review the Code of Research Conduct, Association for Information Systems, November 2008
AVCC (1997) 'Joint NHMRC / AVCC Statement and Guidelines on Research Practice' Australian Vice-Chancellors' Committee, May 1997, at http://www7.health.gov.au/nhmrc/research/general/nhmrcavc.htm and http://www.nhmrc.gov.au/funding/policy/researchprac.htm, accessed 27 June 2005, no longer accessible in April 2009. The relevant text was omitted form the 2003 revision at http://www.nhmrc.gov.au/publications/synopses/_files/r39.pdf
Collberg C. & Kobourov S. (2005) 'Self-Plagiarism in Computer Science' Commun. ACM 48, 4 (April 2005) 88-94
Hexham I. (1992) 'On Plagiarism and Integrity in Scholarly Activity' Humanist: Humanities Computing 5, 4 (3 April 1992), at http://www.ucalgary.ca/~hexham/study/plag.html
Readence J.E. & Barone D.M. (1996) 'Communicating with readers and reviewers' Editorial, Reading Research Quarterly 31, 3 (September 1996) 240-242, at http://dx.doi.org/10.1598/RRQ.31.3.1
Roig M. (2006) 'Avoiding plagiarism, self-plagiarism, and other questionable writing practices: A guide to ethical writing' St John's University NY, 2006, at http://facpub.stjohns.edu/~roigm/plagiarism/
Samuelson P. (1994) 'Self-Plagiarism or Fair Use?' Commun. ACM 37, 8 (August 1994) 21-25
Given the topic addressed in this paper, it appeared to the author to be particularly appropriate to ensure that the instances of self-plagiarism were readily apparent, and to segregate the self-citations from the other references.
Clarke R. (2006) 'Plagiarism by Academics: More Complex Than It Seems' J. Assoc. Infor. Syst. 7, 2 (February 2006), at http://jais.isworld.org/articles/default.asp?vol=7&art=5, PrePrint at http://www.rogerclarke.com/SOS/Plag0602.html
Clarke R. & Kingsley D. (2008) 'e-Publishing's Impacts on Journals and Journal Articles' Journal of Internet Commerce 7,1 (March 2008) 120-151, PrePrint at http://www.rogerclarke.com/EC/ePublAc.html
Clarke R. & Kingsley D.A. (2009) 'Open Access to Journal Content as a Case Study in Unlocking IP' Proc. Unlocking IP Conf., Sydney, 16-17 April 2009, PrePrint at http://www.rogerclarke.com/II/OAJC-0904.html
An accusation was made against this author, by a reviewer of a paper I had submitted to a refereed conference. The reviewer's concern was that "an almost identical version of the paper [had been] submitted at another workshop".
As a member of editorial advisory boards and conference program committees (and as an occasional author of papers on topics in the area of research ethics), it was clear that I needed to respond promptly and precisely.
The paper in question was one of a number I had prepared in the area of online authentication. One of the series had been an invited paper for a symposium of the (U.S.) National Academy of Science, and another had been presented at an international refereed conference, and published in the proceedings.
I had prepared two further working papers in the series, which I had published on my web-site (as I have every paper that I've written since mid-1994). I had notified the URL to colleagues and specialist lists, and the web-pages had attracted moderate numbers of hits. Variants of the material had been presented at seminars in ten Australian cities and in four other countries, and elements of it had appeared in various reports to my consultancy clients.
I prepared a shorter version of the second working paper, and submitted it to a highly specialised international workshop. It was rejected, with little comment from the referees. (It was an engineering event, and limited comment is not uncommon in that community). Some months later, I submitted to the conference in question a paper that was only editorially different from that which had been rejected by the workshop.
The referee was therefore justified in saying that "an almost identical version of the paper [had been] submitted at another workshop". But a number of further facts were relevant:
I submitted to the program chair that, in my experience, exposure of a work-in-process in such forms as seminars and Departmental Working Papers (and their modern equivalent, PrePrints) is not normally regarded as pre-publication for the purpose of publication in conference proceedings or a journal. Further, I had never heard of the rejection of a submission to a journal, conference or workshop precluding re-submission somewhere else. The author should of course reflect new information that becomes available from commentators, especially from the referees who recommended its rejection. But, particularly in the absence of substantive and constructive criticism, nothing precludes re-submission of an as-yet unloved paper in the same form as that in which it was previously rejected.
The Program Committee accepted my explanation. (They also accepted the paper for the conference, subject to changes recommended by the referees).
This paper is a distillation of ideas arising during professional practice in the IS discipline over three decades. I've benefitted greatly from discussions with colleagues who have been active in this and related areas, from multiple interactions with Robert Davison and his hand-chosen discussion-partners on several ICIS panels, and recently from working with Robert and Cynthia Beath on revisions to the AIS Code of Research Conduct. I was stimulated by the posting by the Editor that gave rise to this Special Issue. The large number of thoughtful responses to it confirmed my longstanding impression that, although abstract discussions about ethics interest only a small proportion of IS researchers, 'people know unethical behaviour when they see it'.
The author is a 40-year veteran of the IT industry and a 35-year veteran of the IS discipline, with a publications list dating from the 1970s. He has worked intensively in a number of topic-areas and has on multiple occasions re-visited topic-areas many years later. He publishes across multiple disciplines, and in both the academic and professional literatures. For all of these reasons, he has long been an inveterate self-citer and an inveterate re-user of his own material. Further spice is added to the mix by the fact that one of the topic-areas in which he has long published is professional ethics. As a result, he has been frequently challenged to 'do the right thing', most commonly by his own conscience and his preference to 'build his brand' rather than damage it, but on occasions also by colleagues, reviewers, conference chairs and journal editors.
Roger Clarke is Principal of Xamax Consultancy Pty Ltd, Canberra. He is also a Visiting Professor in the Cyberspace Law & Policy Centre at the University of N.S.W., a Visiting Professor in the E-Commerce Programme at the University of Hong Kong, and a Visiting Professor in the Department of Computer Science at the Australian National University.
The content and infrastructure for these community service pages are provided by Roger Clarke through his consultancy company, Xamax.
From the site's beginnings in August 1994 until February 2009, the infrastructure was provided by the Australian National University. During that time, the site accumulated close to 30 million hits. It passed 40 million by the end of 2012.
Sponsored by Bunhybee Grasslands, the extended Clarke Family, Knights of the Spatchcock and their drummer
Xamax Consultancy Pty Ltd
ACN: 002 360 456
78 Sidaway St, Chapman ACT 2611 AUSTRALIA
Tel: +61 2 6288 1472, 6288 6916
Created: 9 April 2009 - Last Amended: 23 April 2009 by Roger Clarke - Site Last Verified: 15 February 2009
This document is at www.rogerclarke.com/SOS/SCSP-09.html