How should we determine author order on papers?
There are lots of models available. Author ordering is important because it's tied up in prestige and credit for work.
In some fields, there are very strong norms about author order models which can clearly communicate whose work is being represented, who to contact about questions, and who is responsible for the work in the paper. In PER, there is no dominant model about how many / which authors to include, how to order them, or what is communicated by author ordering. Generally speaking, our papers have 2-5 authors on them. This is a lot less than the hundreds (thousands) in high energy research, but still enough that we need to think about ordering beyond simply first and last. Naturally, a paper with only one author does not need to think about ordering! But: potential solo-authored papers still need to have a robust conversation about contributors. They're quite rare in PER: I only have one, and many people have none.
I want to emphasize that the absence of strong norms does not mean that any choice is ok (some choices are definitely bad). Different papers -- even within the same lab -- use different choices. It's important to be explicit about which model you're using and when. For some author lists, different models give the same ordering. This degeneracy is ok for the people writing the papers, kinda annoying for the people reading the papers, and downright problematic for people who want to do quantitative analysis of co-authorship.
The work of writing a scientific paper
Lots of people have heard that you need to "make a significant intellectual contribution" to be considered as an author for a paper, but what counts as a significant contribution?
Before we can think about author ordering, we need to think about what kinds of ways people can contribute to a paper. Everyone deserves recognition and credit for their labor, and research labor can take many forms.
In this table (displayed as an image), I broke down the work in writing a paper into two categories, mechanical and intellectual work (the columns) and three tasks, data acquisition/analysis, paper production, and people herding (the rows). We need to consider all of the parts of this table when we think about the work of a paper.
In some labs, only the people who are engaged in the middle row, paper production, can earn authorship on a paper. In other labs, only the people who are engaged in the right hand column, intellectual work, can earn authorship. Conversely, the people herding work of apportioning work or calling meetings is rarely articulated as important to writing a paper, but it is also important and should be acknowledged.
It's pretty common for different humans to do multiple different parts of this work, and for each human to do different things depending on the paper at hand. Because roles in paper writing often get blended together, it can be hard to articulate which roles (or how much of each role) each person performs for each paper.
Who can be an author?
In academic publishing, there are two common ways to credit people: as an author on the paper, and in the acknowledgements section. Different labs have different policies about what counts for authorship, and different publication venues have different rules as well.
Here are some sets of people to consider, ordered by most permissive to most restrictive:
Everyone who looked at or gave feedback on the paper before publication, including beta-readers and anonymous reviewers.
Everyone who worked in the lab
Everyone who worked on either the project or the paper
Everyone who made a substantial intellectual contribution to the project or the paper
Everyone who contributed writing / figures to the paper
Everyone who made a substantial intellectual contribution to the paper
Only senior people
Generally speaking, options 1-3 are used for the acknowledgements section, 4-6 are for authorship on papers, and 6-7 are for (co)PI-ship on grant proposals. But there are always edge cases where another option is appropriate.
In my lab, as in many labs, all authors are required to sign off on a paper before submission, indicating that they agree with the contents of the paper and consent to have their name attached to it. People mentioned in the acknowledgements section are not required to sign off on a paper before submission. It's nice to tell them about it ahead of time, but not required. Sometimes we refer to some people in the acknowledgements by name and others by group membership or role. For example, we might thank "the members of KSUPER, our copyeditor Jeremy Smith, and two anonymous reviewers".
The roles of different authors
We often talk about authorship roles in terms of their order on the paper, but the deeper principle is about communicating what role each author has in the generation of the work.
First author: high prestige, will be the one named in references (e.g. von Korff et al, 2016)
Last author: sometimes high prestige, only named in references if there are few authors, but often communicates senior status (e.g. Nguyen, Chari, & Sayre, 2016)
Middle author(s): ordering is generally less meaningful the more of them there are.
Corresponding author: the person who communicates with the journal during the submission process (paperwork!), and also the person to whom queries after publication should be addressed.
Principle author, also known as Paper Chair: the person who primarily coordinates the intellectual work of the paper.
Senior author, also known as Lab/Paper Manager: the person who primarily coordinates the funding for the paper, and is often responsible for coordinating the logistics of the paper or coordinating how this paper interacts with other papers in a lab.
It's usually true that the first author and the principle author are the same person. Some papers use a convention that the last author and the senior author are the same person, others make first and senior be the same, and others do not correlate seniority with a particular position in the order.
Sometimes the corresponding author and the principle author are the same, but sometimes they are different people. It's rare for the senior author to be the principle author, at least in my lab, but common for the senior author to also act as the corresponding author. My lab usually uses the terms "paper chair" and "senior author". (We don't have separate humans to be the PI and the Lab manager).
Author order models
These models have very clear guidelines. Some research fields and journals have very strong norms and policies about choosing one of these models, but PER generally does not.
Strict Alphabetical: the tyranny of the alphabet! Often used when the number of authors is large. Can be weird for some authors who have two last names, because some electronic systems may sort one name into the middle name field and/or drop one of their names.
Contribution (first = most, last = least): this works great when the number of authors is small and their contributions are clearly different in scope (not just kind). This model can lead to strife if authors believe that their contributions of are equal scope (if different kind), or if the list of authors is pretty long. Fighting over who is author 4 of 6 vs. 5 of 6 is painful and unproductive. Sayre's Law applies (no relation).
Seniority (first = most, last = least): Like strict alphabetical, this is technically easy to figure out, at least in typical PER papers with 2-5 authors. However, because of the conflation of first author prestige and senior prestige, this system tends to perpetuate inequities in who gets credit for papers in citation. Do not recommend.
Reverse seniority (first = least, last = most): This is still easy to figure out, and considered "nicer" than strict seniority for junior people. However, for people who believe that ordering should be related to contribution -- and for journals which require that -- this can inappropriately elevate undergraduate students (for example) who worked only on a small part of a project that was primarily chaired by a graduate student.
To help address some of the problems with simple models, there are a number of hybrid models that try to balance considerations of contribution and role (which are often hard to order strictly) with considerations of the alphabet and seniority (which are easier to order).
Contribution + alpha (first = most contributed, remainder: alpha): This model balances the prestige of being the first author in citations with the lower-strife tyranny of the alphabet. It's great for long author lists.
Contribution + reverse senior (most authors by contribution, last is PI): This balances the prestige of author order with signalling who is the most senior person on a project. This is a good model for medium-length author lists where the PI doesn't want to fight about where they belong in the list, and/or where the PI wants to signal their seniority by using the last slot. It still inherits many of the benefits and drawbacks of strict contribution models.
Senior + contribution (first is PI, rest are by contribution): This is the worst of seniority combined with contribution, because it implies that the senior person is also the highest contributor. Do not use this model to determine authorship. Unacceptable.
Contribution + alpha + reverse senior (first = most contributed, middle by alpha, last is PI): Ok, this model looks complicated, but it's actually pretty common for papers with 4-5 authors. (Of course, it's degenerate with other models for papers of 3 or fewer authors). The reasoning behind it usually goes like this. We care about the first author, because that's prestigious in citations. We care about the last author, because people often look there for the senior person. But everyone else in the list is just kinda in the middle, so to avoid fighting over who goes where we're just going to use the alphabet.
Ad hoc models
It's possible for authors to sit around and discuss author order based on whatever matters to them. There are a few pitfalls with this approach, and generally speaking I do not recommend that you use it unless all the authors in the group are an established collaboration with prior papers together.
One major pitfall is that ad hoc negotiations tend to favor people with more experience and more power. If you're not explicit about what model you're using and why -- if you don't have robust conversations about what models are available -- you might not accurately reflect the needs and contributions of the different authors in your group.
Another major pitfall is that "whatever matters to them" often means "whoever needs it more". This is antithetical to contribution-based models. Fights over who needs (first) authorship more are painful for all involved. They don't reflect who contributes or in what way they contribute, and it's hard to agree on how we should rank neediness once a disagreement starts. Does the postdoc need it more because she wants a faculty job? The grad student need it more because she wants it in her dissertation? The junior faculty need it more because his institution only counts first-author papers?
That said, ad hoc planning works pretty well for established groups, papers with short author lists where multiple schemes are degenerate, and papers where one author cares a lot and the other authors do not.
It may have occurred to you that these different schemes may produce wildly different author orders. While there are no strict norms in PER, it is very common to consider contribution in ordering. Because this is common, we generally ascribe importance to the first author position on a paper.
As a reader of papers, strict contribution requires that the last author is the least contributory, and therefore probably the least important to contact for followup information on the project. In contrast, contribution + reverse senior tells you that the last author is actually quite important to consider when integrating ideas across multiple papers or contacting for followup information. Unless you know (of) the people involved, it's not clear which of these two schemes is being used, so the existence of both schemes in the field acts as a way to discriminate against emerging researchers and field switchers.
Efforts at some journals to more clearly signal which authors did which work can help ameliorate this problem, but these norms are still emerging. Explicitly noting the corresponding author is a more common thing, but not all journals even do that.
As a writer of papers, it's very important that all people on your research team are in agreement about (a) what work counts for authorship and (b) what author ordering scheme you are using. As a research lab, you should have regular, lab-wide conversations about these two points in general. As potential authors on each paper, you should have an early conversation about which model that paper is using to manage expectations. Return to the conversation several times over the course of the paper as people's contributions change.
This is cool! Can you share more?
These materials are drawn from a workshop on authorship developed in collaboration with Scott Franklin and Mary Bridget Kustusch as part of PEER. The workshop also includes role plays for potential co-authors to discuss thorny situations in author ordering, as well as guidance on how much and what kinds of material belong in a paper for different journals.
Do you want to experience the workshop? You can join a workshop that's already happening, such as via a PEER field school, or I can run this workshop for you and your lab. Contact me for possibilities.
It's ok to use these materials in your own lab or teaching to help facilitate a discussion around author order. Please direct your lab members to this website directly, and credit it appropriately. It's not ok for you to use these materials as part of a paid workshop. If you're confused about how or when to use these materials, contact Eleanor Sayre for help.