R citation which year to cite?

Question

Most of the time my R version and packages are reasonably up to date. Now I wonder what year to cite when using citation("somepackage"). Basically all packages are 2012 then. How can I find out the real date when they first came up? Or should I really cite the most recent version?

Smitty

I know this is not exactly a programming question, but it's not for crossvalidated neither. At least it's related to a programming language and here's where the R folks are at. – Smitty

If you use the newest version, then I think you should cite this version. – Refractory

Hmm, it's that you cite packages like papers. Let's say that you write about development in reproducible research and how knitr followed sweave it's a bit strange to say that one is (Leisch,2012) and the other is (Xie,2012). Just look a journal of statistical software's most recent articles. I bet that the author has not used the 2010 version of superlearner cause it was updated in 2012. I mean you don't set versions back usually. – Smitty

Whichever year comes up for the version of the package you have when you call the function, which would correspond to the date for the version you're using, one assumes. – Thereunder

Answer 1

The norm in the medical field seems to be to cite something like this in the text:

We used Stata version 9.2 (Stata Corporation, East Nowhere, Texas, 2006).

That norm argues for citing the year of the version you actually used. And, in my view it makes sense if you want subsequent authors to be able to replicate your results.

There's a whole emerging field that's trying to figure out how to cite things other than papers (see, for example Heather Piwowar's work). If there's a Journal of Statistical Software article on the R version you want (more likely for packages than for base R), you could cite that. Or cite the R Journal's updates for the relevant version.

Pinprick

+1 like your suggestion to cite the advertisement in the JSS or R journal. – Smitty

Package authors can specify what they want to show up in the citation() command. – Premier

Actually, it is the other way around: citation() defaults to providing a citation for R, but can also supply one for packages if the given package(s) have a corresponding file. – Manet

Answer 2

Why doesn't anybody mention the obvious? There is a citation() command in R itself:

edd@max:~$ R --silent -e 'citation()'
R> citation()

To cite R in publications use:

  R Core Team (2012). R: A language and environment for statistical computing. 
  R Foundation for Statistical Computing, Vienna, Austria.
  ISBN 3-900051-07-0, URL http://www.R-project.org/.

A BibTeX entry for LaTeX users is

  @Manual{,
    title = {R: A Language and Environment for Statistical Computing},
    author = {{R Core Team}},
    organization = {R Foundation for Statistical Computing},
    address = {Vienna, Austria},
    year = {2012},
    note = {{ISBN} 3-900051-07-0},
    url = {http://www.R-project.org/},
  }

We have invested a lot of time and effort in creating R, please cite it 
when using it for data analysis. See also ‘citation("pkgname")’ for citing 
R packages.

R> 
R> 
edd@max:~$ 

Manet

"Why doesn't anybody mention the obvious?" Because you hadn't posted yet, obviously! – Pinprick

eh @Dirk, I did mention that in my initial question. Because I update quite regularly I just end up with all my citations being (package,2012). But if I look at recent JSS or R Journal articles I realize that not all of their R citations are of 2012. – Smitty

Because there is a) lag between submission and publication and b) not everybody uses most current version. Now ask yourself: whose interest do you serve by citing old versions? – Manet

@Dirk, it's not that I like citing old version, it's more convenient to just use citation. It just that I was surprised that in some journal articles citations very quite outdated, while I thought the author must have had a newer package when writing it. However, I seem to get point here... slowly. – Smitty