The current review identified that only 24% of the 266 R1 and R2 institutions had public, institution-level authorship policies. Given the established importance of authorship (Rennie and Flanigan 1994; Wilcox 1998; Macrina 2011), as well as the potential for ambiguity and conflict over authorship (Smith et al. 2019a; Marušić et al. 2011; Birnholtz 2006; Gómez-Ferri et al. 2019), this appears to be a critical oversight on the part of the most research-productive institutions in the United States. Our work addresses the gap in understanding problematic authorship practices by analyzing the publicly available authorship policies from 64 R1 and R2 institutions, with a particular focus on author dispute resolution. We now discuss the implications of our findings for institutional authorship policies in general and, more specifically, guidance for authorship disputes.
Of the 266 universities identified in the Carnegie R1 and R2 classifications, only 24% (64 universities) had an institution-level authorship policy. This number seems surprisingly low given the primary focus on authorship and publishing as key to academic career success. Established policies can be advantageous for institutions as they guide employee expectations and decision making, keep higher-power individuals accountable, and inform employees on where to turn for help when faced with a problem. As publications are considered the coin of the realm for academics, the advantages that accompany the use of an established policy seem particularly important for authorship practices due to the potential for conflict and abuse surrounding such a high-stakes aspect of the academic job.
The majority of universities with a public-facing authorship policy were classified as R1 institutions (80% at 51 universities) and/or had a medical school (83% at 53 universities); however, grounded in our correlational analyses, medical school status appears to be more strongly related to the likelihood for having an institutional authorship policy. This is consistent with the tradition among medical schools to acknowledge authorship practices via such organizations as ICMJE. While the prevalence of institutional authorship policies may be greater for universities with a medical school and/or high research activity, this does not necessarily speak to the quality of the policy. In fact, institutional authorship policies in our sample demonstrated, on average, a depth of 2.78 out of 4.00, suggesting there is room for improvement. Across all included institutions, the most common policy component was specifying authorship criteria (at 94% of universities), which aligns with published guidelines from ICMJE and other related organizations as discussed above.
The least common policy component was related to collaboration between students and faculty, with only 55% of institutional authorship policies mentioning this issue. Students represent a vulnerable population in terms of authorship ethics, as they do not have the same rights as faculty or the same security in their position at the university. Due to the dependence of most graduate students on their faculty mentor, students may not be comfortable addressing authorship issues when they arise. Further, faculty mentors often control publication and research funding opportunities for students, both of which are vital to student success in school and post-graduation. Institutional guidance for authorship in faculty-student collaborations has the potential to reduce the likelihood for students to be taken advantage of and address any problematic “hidden curriculum” regarding authorship practices. Further, making issues regarding authorship ethics visible to students helps them build awareness of the matter and practice habits of transparency and fairness in the distribution of authorship.
Interestingly, universities without a medical school demonstrated the highest quality institutional authorship policies as they were more likely to include all coded policy components. Indeed, 82% of universities without a medical school included a policy component on authorship practices for student-faculty collaboration (compared to 55% mentioned above for all universities and 49% for just universities with medical schools). This may be due to universities with medical schools more heavily prioritizing the standards put forth by organizations such as ICMJE. For example, while ICMJE provides materials around what constitutes an author and issues such as ghost, gift, and guest authorship, they do not specifically discuss other types of authorship issues, such as dispute resolution or student-faculty collaboration. Thus, universities that emphasize standards from the ICMJE may be less likely to provide authorship guidance stemming from issues that are not discussed by the ICMJE.
Ideally, authorship disputes could be avoided entirely through the use of prevention techniques, thereby ensuring that communication of research findings is not delayed, appropriate credit is assigned, and productive teamwork is not harmed. One effective tool for preventing disputes is clear and early communication between team members regarding authorship expectations, and most policies recommended this practice (63% of included policies). For example, one policy notes that “as early as possible in the research process, Principal Investigators should inform collaborators of the requirements for authorship of any manuscript that will report results of the project.” This step can be easily overlooked; however, an agreement amongst all authors can remind authors of their obligations and prompt revisitation of authorship lists as individual roles evolve throughout the research process. A few policies recommended written agreements, which can be especially beneficial in providing evidence of past agreements for authors in lower-power positions. However, few dispute resolution sections of policies mentioned prior agreements (verbal or written) as a recourse for resolving dispute, though this practice is frequently recommended (Frassl et al. 2018, 3; Vicens and Bourne 2007; COPE 2016). We found no other formal or informal attempts to prevent authorship disputes in the policies surveyed.
Only 16% of R1 and R2 universities (n = 42) have dispute resolution policies. We noted both substantial overlap and distinctions between policies from different institutions. For overlap, text comparison revealed that some policies were almost identical as noted through acknowledgements of other source policies. Other policies were fairly idiosyncratic, which suggests that these policies arose in response to particular events at each institution. The small number of dispute resolution policies is surprising given (a) the prevalence of disputes previously discussed and (b) the fact that journals often avoid resolving disputes. In cases where prevention fails, it helps to have an institutional dispute resolution process for several reasons. First, public policies offer transparency regarding processes, which benefits all parties by establishing shared expectations. Second, the process of building a policy, provided it incorporates a diverse and inclusive set of perspectives, is better able to accommodate the contexts in which such disputes arise. Third, a formalized process can often allow for reasonable resolution without disputants resorting to damaging public or social media complaints, or silencing the complaints of those in lower-power positions. These factors strongly point to the importance of institutional policies to resolve authorship disputes.
Aside from the overlap in policies using almost identical verbiage, a range in dispute resolution provisions emerged when examining the 42 dispute resolution policies identified in our study. The majority of policies with dispute resolution provisions (n = 36 out of 42, 86%) suggested that authors first try to resolve disputes among themselves, or involve the department chair (n = 32, 76%) or Dean of the college (n = 31, 74%). These are reasonable first steps with smaller authorship teams within departments or colleges, but when institutional teams reach across colleges, this guidance may not be helpful without specifying how Deans can work together on authorship disputes. Other potential approaches were anticipated in some policies; for example, some recommended involving a Provost (n = 9, 21%), Ombuds (n = 7, 17%) or RIO (n = 9, 21%). Some policies (n = 11, 26%) reference the formation of a dispute resolution board or committee, but for the most part, these policies do not offer many details concerning the composition of the committee or its specific role.
In addition, there seems to be no best-practice consensus on inter-institutional authorship disputes. Authorship is increasingly multi-disciplinary, multi-institutional, and trans-national (Adams and Gurnsey 2018; Hsiehchen et al. 2015). Very few policies explicitly addressed this potential issue (n = 5), leaving multi-institutional author teams without guidance on how to resolve disputes. One policy suggested that a “Publication Subcommittee representing all Investigators should be established at the beginning of any multi-center study for the purposes of expediting, coordinating, and monitoring the paper-writing processes.” While this recommendation may prove useful for larger project teams that span across multiple universities, there are many circumstances where small authorship teams span multiple universities and disciplines. There were no recommendations for resolving authorship disputes in the latter circumstance within our sample.
A few policies specifically indicated that if a dispute is sent to a dispute committee with the permission of all parties, then any party has the right to make the committee’s (non-binding) recommendations public. In such cases, some parties may, rightly or wrongly, attempt to adjudicate the dispute in the court of public opinion. Policies often indicated that authorship disputes do not qualify as research misconduct, but many institutions also include the Research Integrity Officer or the department of research integrity in the hierarchy of potential recourse, which may send mixed messages to those using the policies. If journals routinely remand authorship disputes to institutions for resolution, universities with no policies or vague policies may lack the resources to resolve these disputes, delaying publications and thus hampering the dissemination of research findings and publication credit for their faculty and students. Seen with this lens, it is ironic that one university’s policy notes the potential for involving a journal editor in cases of disputed authorship.
Notably, policies seemed to lack discussion of enforcement. While some indicated, for example, that a dean’s decision about authorship would be “final” or “decisive,” there was no indication of consequences should authors not abide by that decision. Several institutions mention possible “sanctions,” but rarely specify the exact sanctions and who might impose them. In sum, explicit discussion of enforcement of authorship policy was rare. Without enforcement or consequences, it is difficult to determine what impetus institutions will have to publicize their policies or what incentives faculty and staff may have to learn about and follow a policy. Moreover, individuals in lower-power positions have the most to lose by reporting authorship abuse, yet only one policy surveyed specifically prohibited retaliation for authorship disagreements and complaints. Presumably, most institutions already have a non-retaliation policy. If so, authorship dispute resolution policies should explicitly invoke it in order to ensure that potential disputants feel protected from reprisal. Overarchingly, the majority of dispute resolution policies lacked sufficient specificity for the policy to be of much use. The wide diversity of recommendations and lack of evidence for consensus about best approaches to resolving authorship disputes indicate that there is as yet no consensus on best practices for intra-institutional disputes. Thus, more work on productive approaches to authorship dispute resolution policies is needed.
While we believe this work takes an important step forward in examining the prevalence and content of institutional authorship policies, there are several limitations to our study. One limitation is our reliance on university websites to obtain authorship policies. Here, access to authorship policies may be subject to technical issues. For example, any potential search hits for an authorship policy had to be accompanied by a functional web link to the actual policy. Any potential authorship policies with broken links (i.e., it was not possible to access the policy via the university’s website) were not included in our study. Additionally, some universities may have institution-level authorship policies, but not provide them publicly on their websites. Any such policies were not included in our study. Thus, interpretation of our findings should be from the perspective of universities providing public-facing authorship policies. However, we suggest that having public-facing authorship policies (rather than policies merely accessible within the institution) are important for stakeholders. For example, inter-institutional collaborations are already common, so collaborators from other institutions could access policies for all members of a research team if needed.
Recommendations for Future Policies
Our survey found many policy components to be both common and supported by the literature in this area. However, it did not reveal broad consensus on the fine-grained details of institutional authorship policies (such as dispute resolution practices). Below, we share our holistic interpretation of good practices developed from the survey and literature review that will guide our own institution’s authorship policy process.
Reasons for Having a Policy
An institutional authorship policy is important for creating and supporting an ethical culture. Given the increasing potential for authorship disputes, and the lack of alternative processes for addressing them, a clear policy will provide predictable mechanisms that may help to ease tensions between collaborators. Moreover, normalizing authorship discussions early and often in collaborative processes may have additional benefits for students and junior faculty, who may lack the ability to raise these issues themselves. By making authorship criteria transparent, graduate students may learn the lessons we want them to learn about good authorship decisions, rather than learning—and then later reproducing—a “hidden curriculum” whereby the power to make authorship decisions is abused. Thus, it is clear to us that a well-publicized institutional authorship policy would be a benefit for all institutions.
Elements of a Policy: Justification, Applicability, Criteria, Prohibitions
Our survey found several common elements in extant policies that we intend to include in our own policy development:
Justification: Most policies include an introductory section communicating the reasons for an authorship policy, including protecting the integrity of research and careers, and supporting and promoting healthy communication. Empowering those potentially affected by an authorship policy with knowledge about why the policy is important (and communicating those reasons in publicity efforts) is another good way to promote ethical culture.
Applicability: A policy should also communicate the scope or applicability of the policy. For example, will faculty, students, and certain staff be subject to the policy, and under what conditions? Because authorship disputes between faculty and graduate students can occur, policies such as faculty or graduate student grievance processes would likely not be able to address a faculty/student dispute.
Criteria: Although policies frequently presented general or commonly accepted criteria for authorship, they were almost universally extremely careful to acknowledge and defer to wide disciplinary diversity in authorship norms (including author ordering). Many mentioned (without endorsing) international standards such as those from the International Council of Medical Journal Editors (ICMJE) or the Committee on Publication Ethics (COPE). By communicating both general criteria for authorship (e.g., making a significant intellectual contribution) as well as referring to a variety of existing norms, this element of a policy can uphold ethical culture and inform institutional members of resources to which they can turn for more information. For this reason, it also seems wise to include, as an appendix to a policy or in some other form, a set of references about authorship issues.
Policies also frequently encouraged collaborators to articulate together which standards they plan to follow through open, two-way conversation resulting in oral or written agreements, with the latter being preferred. Other forms of documentation, such as creating a project on the Open Science Framework and listing author order at the beginning of a project, can serve a similar purpose (https://osf.io/). Such steps facilitate an initial conversation even if authorship order ultimately needs to be changed based on contributions. Several policies in our survey suggested referring back to such agreements to resolve disputes, and published literature similarly suggests that early discussion and formalized authorship agreements can be helpful (Marušić et al., 2014). Policies should also note that authorship can change in the course of research and invite collaborators to revisit authorship determinations as warranted throughout the lifecycle of research production and dissemination. These practices benefit students in particular by educating them in good publishing practices and offering a more transparent approach to assigning authorship credit. Finally, the practice in several policies of explicitly assigning to the communicating author the responsibility to discuss authorship collegially seems to be a beneficial policy element.
Prohibitions: One way in which several policies were quite specific was in explicitly denouncing practices like ghost, guest, or gift authorship. Because there is more agreement about prohibited practices, and because explicit denunciation of these practices in a policy potentially arms a wronged party, this seems to be a beneficial policy element as well.
Authorship policies implicitly acknowledge that disputes are predictable and likely. We think that no authorship policy will be complete without addressing possible disputes, especially given the lack of alternative resolution mechanisms we outlined above. Given the high stakes attached to authorship decisions, a transparent, predictable process based in policy may help both to avoid disputes in the first place and address them if they arise.
Policies we surveyed overwhelmingly recommended trying to resolve disputes initially at the lowest hierarchical level possible, e.g., between the collaborators, with the help of the chair of a department if the dispute concerns only members of that department, or with the dean of a college if the dispute stretches across units. Because this is likely to resolve at least some disputes, we think these are good initial steps. Common next steps among policies were to consult a “neutral third party,” including an Ombuds, or occasionally a Grievance Committee, Dean, Vice-Provost, Provost, Research Integrity Officer, General Counsel, or Vice-President of Research. Some policies also included bodies unique to that institution. A few policies indicated that if a student is involved, the Chair of the Dissertation Committee, the Director of Graduate Studies, or the Graduate Student Grievance Committee might be included in an informal resolution process. While it may be helpful to include an interim step after early attempts at resolution and before invoking a more formal process, we found no consensus regarding whom to invoke as a neutral third party.
In our survey, only eight institutions with dispute policies mentioned a final, formal step of convening a committee or panel to consider disputes, and only five of them (Clemson, Duke, Emory, George Mason, and Louisiana State) outlined specific processes for this. The others indicated that a committee was one option among others, and in one case only provided for a student committee. Although a very small proportion of the 266 surveyed institutions include this step, we think that policies should acknowledge the possibility of protracted and challenging disputes that cannot be resolved by the steps outlined above, and provide a more formal resolution process that encompasses disputes between any combination of student, faculty, and staff.
Faulkes (2018) has recommended alternative dispute resolution mechanisms, either at the level of a journal editorial office or through independent third parties who may serve as mediators on a fee-for-service basis. However, we think there are benefits to a process that remains at the institution. Our tentative plan is to provide for an Authorship Dispute Committee, with membership drawn from various institutional areas such as the Ombuds for faculty and students, administrators, and content area experts as needed. A few policies include very detailed steps regarding written notice, time frame, and appeals. This has the benefit of providing a very clear process, but also makes for a much more cumbersome policy that is consequently less likely to be read, so we have not yet reached a decision on this step.
The most challenging aspect of an institutional authorship policy may be enforcement. Most policies did not make any reference to enforcement, and those that did were not explicit about how this would be achieved, instead simply stipulating that a certain decision would be “final.” This is an area needing significant additional analysis (including legal analysis) because it involves balancing faculty rights to academic freedom and due process against an institution’s obligation to resolve authorship disputes.
As a result, our current plan is not to include any provision for enforcement.
Once established, an institutional authorship policy should be publicized across campus, and training on good authorship practices should be offered to both faculty and graduate students. Faculty would benefit from help in initiating difficult conversations with graduate students, and graduate students could use the policy to raise what might otherwise be a difficult topic with faculty supervisors. It is important to acknowledge that publicizing such policies can predictably flush already-existing disputes and tensions into the open, so it will be critical to have mechanisms in place for that eventuality prior to publicity efforts.
We hope that as more institutions develop authorship policies, additional research will bring to light evidence for best practices. In the meantime, we offer these ideas to others as our considered opinion developed by months of immersion in existing policies.