17

I've written a statistical software package and published a paper on it. Recently, it appears to have received moderate use, but with at least less than a 50% citation rate. I know this because Google Scholar will tell me that a paper mentions the name of my software package (it's a nonsense word, so this shouldn't be by chance), but the paper will not cite the package itself; in some cases there will be something like "We used package X for statistical programming language R", with a citation for R but not X.

One the one hand, I'm glad people are using my tools and that makes me happy. On the other hand, I would like citations due, if just from a career standpoint. My guess is that many researchers may not be aware that a citation should be included; I estimate that the majority of users of the package probably work outside the field of statistics. R makes it very easy to cite software (R> citation("X") gives you the bibTex)...but I didn't even know that until I started writing software.

It's my view that at least some of the responsibility should lie with the reviewers/editors. Would it be impolite to point this out to the editors of the journal? I really don't want to be in the business of harassing the users of my software, nor trying to point the finger at them either. Or is it just accepted that you should expect a little under a 50% citation rate?

EDIT:

A good question was asked in the comments: what do I hope to achieve? I'm not 100% sure! I suppose I'm wondering if there's a polite way to raise awareness of proper citations for software? Both writing software and continuing to support users of the software is fairly time consuming (I would estimate I've spent over 300 unfunded hours on those two tasks?), but I'm convinced it's one of the ways statistics researchers can be most helpful to outside researchers. As such, I suppose I would like to gently push the system to be more supportive of that effort...without being a grumpy ol' stats guy complaining about anyone who uses their contributions.

Cliff AB
  • 2,523
  • 11
  • 16
  • 3
    What do you hope to achieve by contacting editors? What do the journal's author guidelines say about citations regarding software? If they don't say anything, that's something worth pointing out. –  Mar 20 '18 at 15:13
  • 2
    The way I learned it, only specialized software normally needs citation. E.g. a basic math package that a lot of people use is similar to "common knowledge", but a specialized package for calculating criticality in Soviet-type nuclear reactors would need to be cited, as even math and physics-heads might have never heard of it. – Robert Columbia Mar 20 '18 at 15:21
  • 3
    @RobertColumbia: In this case, I think the authors recognize the software as specialized enough; they did mention that they used it, implying it was specifically targeted at their type of data problem, but perhaps didn't realize they should cite it. It's my thought that an editor should recognize this is a case when a citation is warranted? I do recognize it's not black and white; I've never cited Unix, for example. – Cliff AB Mar 20 '18 at 15:51
  • 3
    For career purposes, it seems like you've already used a good strategy, unintentionally or not: by using a nonsense word that can be easily indexed, you can summarize the number of papers using your package regardless of whether they include it in a reference list. – Bryan Krause Mar 20 '18 at 16:02
  • 2
    Have you published a paper related to your package? My experience has been that authors generally will cite a paper if one exists. – Brian Borchers Mar 20 '18 at 18:28
  • 1
    @BrianBorchers: there has been a paper published. Although the software was released 2 years before the paper was published, so I do recognize that in some cases, missing citations are just a feature of lag time. – Cliff AB Mar 21 '18 at 03:02
  • 1
    Giving the citation of the paper in the user documentation and asking people to cite it can be helpful too. – Brian Borchers Mar 21 '18 at 03:25
  • Look at the bottom of the GMSH web page, there they provide citations for their software, respectively software features. – Dohn Joe Mar 21 '18 at 06:49
  • Does the software tell people how to cite it? –  Mar 27 '18 at 17:43
  • @BrianBorchers: "My experience has been that authors generally will cite a paper if one exists." - while it may help the OP in the concrete case to provide such a paper, it seems questionable to further promote that practice. – O. R. Mapper Mar 28 '18 at 12:11
  • @O.R.Mapper it is increasingly common in my field (ecology/evolution) for software developers to publish a short paper announcing their program, in large part to deal with the issue OP has raised - giving users a way to cite the program. I'm not sure why you suggest not promoting this practice? – Tyler Mar 28 '18 at 13:26
  • @Tyler: Citing software is considered appropriate for a variety of reasons (giving credit, disclaiming parts of the contribution that were not done by the author, pointing out/"advertising" interesting or useful resources to othe researchers). Nothing about these reasons changes depending on whether the software author happened to publish a paper-like document to present the software, so neglecting the citation just because there is no such publication seems unethical. Accordingly, promoting the practice of publishing a paper just to announce the software in order to "giv[e] users a way ... – O. R. Mapper Mar 28 '18 at 21:06
  • ... to cite the program" (as if there were none otherwise) could give rise to the idea that due citation could be skipped if the software author has not provided such a paper. – O. R. Mapper Mar 28 '18 at 21:08
  • See more recent discussion of this, and how to enable citations, at https://academia.stackexchange.com/questions/164177/are-software-citations-indexed – Andy Clifton Mar 21 '21 at 13:18

2 Answers2

3

Reaching out to journal editors might backfire, I'd be careful with that approach.

Ultimately it's going to be up to your users to a) know how to cite your work and b) decide to do so. You have some control over a): you can display a short note with the preferred citation when the program starts. In the case of an R package, you can use the function .onAttach to display a message when your package is loaded. If it's a stand-alone program, put the message in your splash screen, or on the console, wherever a user might see it.

In the short term, you don't have any control over b), so I suggest you focus on maintaining the program and responding to users as best you can. As we use an increasing number of programs, deciding which ones get cited in any given paper is going to be a bit subjective. If my analysis depends on a bug fix or new feature that a developer implemented for me, they will definitely be cited!

Beyond that, your influence is limited to longer-term approaches to changing practices:

  • lead by example, so people become accustomed to seeing software getting cited
  • as an editor or reviewer, request citations be added where you think they should be
  • promote the idea to your students and colleagues
Tyler
  • 892
  • 6
  • 10
1

The RCUDA and RLLVM packages for R that allow the use of GPUs inform the use on the license that the paper should be cited. I think the best way to minimize the problem is to clearly inform the users in the license text of your package and also on the Readme file of the repository (maybe even as the topic 0 of the quick start guide, that we know everybody will read).

However this is no guarantee that the users will follow the instruction.

EDIT
It seems that there is no reference on the RCUDA and RLLVM repositories to which paper should be cited, although there was a pointer to the original paper before.

prmottajr
  • 275
  • 1
  • 2
  • 9