Here are some guidelines HEGIS has developed on authorship and data ownership. Ideally, authorship is discussed among researchers from day one of a project but reality is usually messier than that. Barring an explicit agreement or "prenup," paper drafts should note a preliminary authorship order that is then refined over time. Similarly, data agreements should be in place before collection or analysis begins.

• Authorship
  • What does author number and order mean?
  • Who should be an author?
  • Mentoring and authorship
  • Authorships to avoid
• Data
  • General issues
  • Sharing and archiving

Authorship

What does author number and order mean?

The significance of the order and number of authors varies by field. Per the discussion below, authorship should be accorded to anyone making a significant contribution, but there are nuances to author number and order that bear on the questions of significance.

• Single authorship means you did the work, but allows for the possibility that other people helped out in minor ways (see below). These people should be recognized in the acknowledgements.
• Dual authorship usually means a pretty even split in the work but often the first author has taken greater responsibility for the write up and almost certainly has handled correspondence with the publisher unless it makes sense for the second author to do it (e.g., sometimes an advisor will handle the details when working with a student who is moving). Dual authorship with a mentor/advisor, especially when the mentor is listed second, can be used to recognize the overarching role played by the mentor in helping to develop thesis research.
• Multi-authorship is harder to assign and assess. Multi-authored papers are common in some fields and increasingly common in many more, especially with the rise of large funded projects and interdisciplinary research. Ideally, authorship is assigned in order of significance of each person's contribution. In some fields, first and last authors are accorded special significance, where the project leader goes first (this seems more common on large projects in engineering or the hard sciences) or last (more common in medicine). In a lab environment in particular, the lab leader may expect to be an author on all papers reporting on research conducted in the lab, but this practice can be contentious depending on disciplinary and institutional environment. On large papers where it is difficult to ascertain or reconcile significance for each individual's contributions, it sometimes makes sense to make the author list alphabetical (forward or reverse, so Zoltan Zarec does not always lose out) or alphabetical after the first author.

Who should be an author?

Authorship is usually accorded to anyone who has made a significant intellectual contribution to the research, but the definition of 'significant' can be tricky. When in doubt, it is usually better to include than exclude a coauthor, especially once you have more than two coauthors.

Significance usually connotes that the paper required the participation of a particular person. Did he or she bring to the paper a unique skill, method, or perspective? If so, then that person should be an author. On the other hand, if the role was limited to performing a short, standard operation such as georectification or database manipulation, the person may not necessarily merit authorship because the contribution was not sufficiently special.

Significance can be measured in many ways and it helps to break it down. As a rule of thumb, if you meet several of the following criteria then you should probably be an author, although some criteria (e.g., Research) are more important than others:

• Idea: did you help formulate the initial idea or provide a key insight?
• Planning: did you get into the nitty-gritty of coordinating data collection, methods, literature review, or theoretical conceptualization?
• Facilitation: did you provide space, people, or other resources? Did you help write the grant that funded the research?
• Research: did you help do the research, such as by collecting data, developing a model, do the statistics?
  
• Analysis: did you help analyze data, do key database operations, or develop visualizations that further the analysis?
  
• Write up: did you help write up the results or provide substantive/substantial comments or revisions?

Mentoring and Authorship

Mentoring/advising and coauthorship is usually rewarding but can also be trying. In most fields, students are generally encouraged to follow criteria set out by their advisor or lab as long as these guidelines have been established ahead of time and are generally consistent with the norms of the discipline. A good mentor has typically met several of the authorship criteria noted above and will therefore often be a good candidate for coauthor. Students earlier in their careers typically include their advisor as a coauthor, but with the passing of time and greater research independence, students may publish a subset of papers without their advisor or mentor. At the other extreme, particularly in the humanities, the importance of sole authorship may preclude inclusion of advisors and mentors as listed authors, especially on the first significant publication, but they are often added to a subsequent work in recognition of their contribution. Most of the onus is on mentors to consider their role in papers produced in their labs, especially given the role of sole- and first-authorship in furthering their advisees' careers, and develop a fair plan of publishing.

Authorships to Avoid

It is important to develop a sense of where potential coauthors are coming from with respect to their expectations on publishing. When all else fails, trust your gut and look to your long term reputation over short-term gain.

• Asymmetrical authorship: one too-common pattern is the asymmetric coauthor, who demands coauthorship on any article he or she has done any work on regardless of significance (e.g., the work is limited to suggesting minor edits on a final draft or adding a paragraph) but not including coauthors on their "own" pieces despite significant inputs of other people. While some researchers have turned asymmetric coauthorship into a successful strategy (and we all know a few), they the run the risk of getting a bad rep.
• Honorary authorship: although it varies by discipline as to whether it is a warning sign, some authorships are "honorary" in that they are given out as a recognition of past work or importance and not for work done on the publication as such.
• Hands-off authorship: if your name is on the paper, you own it for good or bad, so make sure you know what the paper is about. There have been a few high-profile cases of papers being found to be based on incorrect data or analysis where one of the putative authors of the paper later denies any substantive involvement.
• Recycled authorship: the seriousness of this kind authorship tends to vary by discipline, but some authors become well-known for recycling papers, or taking an existing paper, mixing in some new material, and the publishing it as a new paper. Editors and reviewers are the traditional gatekeepers in this instance, but they can only be so effective when dealing with authors who feel pressure to publish more in a world with an expanding number of journals.

Data

Data want to be free, don't they? Well, not necessarily.

General issues

Federal mandates. Although guidelines vary by program, one rule of thumb is that data collected under federal funding should be made available after three years of collection and/or analysis. Funding agencies tend to not enforce this expectation in explicit terms, but if someone asks for your data and you do not give it to them in a reasonable time frame, the decision can come back to bite you. On informal terms, you do not want to be a person who becomes known for not sharing tax-payer funded research, or more formally, the asker is within their rights to complain to the funding agency, which can lead to your proposals being rejected (e.g., there have been a few cases where people got burned at one funding agency because they were seen as not sharing their data after several years). Other funders sometimes have their own rules, so take a look at their guidelines or speak with your program manager. More broadly, researchers have an legal responsibility to fulfil the terms of their funding, and an ethical expectation to eventually share their research, data included.

Project/lab agreements. There seem to be two models for sharing data that were collected as part of a larger funded project, regardless of source. The first model holds that, where within reason, anything made in the lab (project) belongs to the lab (project). The second model gives collectors of data a good period of time (around three to five years) to publish and then the data are opened to others in the project/lab. There are legitimate disagreements about either model, in that while some argue that the data should be completely free for use by anyone in the lab after the initial waiting period, others contend that the data collector is due recognition and ownership. At the very least, acknowledgments are in order to recognize who collected the data and under what circumstances. More broadly, in cases where the data are not in the public domain, the person who wants the data should make a good faith effort to identify ways of coauthoring with the person who collected the data (more on this below).

IRB/Confidentiality concerns. Human subjects concerns are paramount, and provisions established by the IRB follow the data for their lifetime (e.g., confidentiality practices). These cases are usually negotiated on a case by case basis.

Your own data. Rules for sharing data seem to be evolving for data collection funded by agencies with no sharing rules or collected on your own time. Overall, barring general issues of copyright, the peer review and publication process generally holds that anything published with data that you collect is fair game in a scientific sense (e.g., peer review, verification, or replication) but the applicable norms are evolving by discipline and outlet.  Some journals, like PNAS, hold that the author must share data and coding from the time of publication, with the sanction for non-compliance being that PNAS can refuse subsequent publication. Otherwise, the university and/or the lab may have claims on your data in the sense that if you collected them while in residence or having used any University property or resources, these data are not absolutely yours.

Sharing and archiving

Broadly speaking, there are several different ways to approach archiving and sharing in the lab.

1. Data collected as part of a research assistantship. These data belong to the PI and a copy must be stored in the lab. Barring any disagreements or specific arrangements, these data fall under the provisions of the above noted policies on data sharing. Anyone in the lab is welcome to use these data as long as there is discussion/effort put into investigating some form of collaboration first or acknowledgment second.
2. Your own data that you may want to share. It is often a good idea to share your own data, especially if offers cross-over with larger funded projects or topics of interest to the lab. These data are 'yours' in the broad sense and only subject to the considerations above, although it is nice idea to share them as possible with other lab members when the situation warrants. Consider the time saved and expertise garnered over the years because you used lab data.
3. Your own data that you want under wraps. The lab will keep archival copies of your data as a form of disaster insurance. There have been several instances in the last ten years when a student left the lab and needed to dig up data after a few years of being at their new institution because of data loss and corruption.

In terms of use, data falling under categories 1 and 2 above will, over time, move towards standard data-sharing practices in the sense that there is pressure to share (formal, as in funding agency limits, or informal, as in scientific publishing practices). That said, it is a good idea to explore collaboration or some sort of credit-sharing arrangement. In practice, many kinds of data require a degree of shared research in order to make sense of the data, regardless of the quality of metadata or seeming simplicity of the data.