[Rd] Citation for R

Simon Blomberg Simon.Blomberg at anu.edu.au
Wed Jun 15 03:17:02 CEST 2005

As a biologist, I have been following this thread with interest.

I don't understand how someone can truly peer-review any software 
without access to the source. Surely R is the only major environment 
for statistical programming that _can_ be peer-reviewed in that 
sense. (The BUGS project is heading in that direction with OpenBUGS.) 
All the others (SAS, SPSS etc.) rely on their "prestige", since they 
are unwilling to show us exactly how they do their computations.

I agree that the citation problem is not a bug in R, but a bug in the 
publishers. It is up to the users to aggressively tackle resistance 
to citing correctly published software (ie software with an ISBN 
number,as opposed to in-house publishing of User manuals etc.). 
Perhaps the R Development Core Team could provide  a position 
statement on the citation and publication of R that could be used to 
convince editors that the current citation method for R is correct 
and appropriate. If it were signed by all the members, with their 
academic affiliations, it would be hard for editors to disagree.


>On 14-Jun-05 A.J. Rossini wrote:
>>  Fritz -
>>  That's silly.  As someone pointed out, the issue is with the
>>  publisher, not the citation.  If R-Core were a generally well-known
>>  and regarded publishing house such as Springer or Microsoft, it would
>>  not be a problem.  But it's still a nebulous entity to MANY people,
>>  and so many people fail to understand this open source stuff.  It's
>>  seriously discouraged by most journals to cite technical reports, for
>>  example.  And perhaps, R could be considered more of a long-ish
>>  technical report than a book?   Though perhaps Peter D. could be
>>  considered the "editor"?  (these questions are not those that I need
>>  to ask, obviously!)
>>  (just yesterday, I was asked by a reasonably intelligent colleague,
>>  with respect to corporate packaging of R: "So they (corporate
>>  packagers) just pick some version and package it, right?"  and my
>>  flabbergasted response was:
>>  "and so, what the heck do you think they (corporate packagers) do with
>>  SAS, S-PLUS, and SPSS, and why do you think it's different...?").
>>  best,
>>  -tony
>Tony is getting admirably and justifiably tetchy! His comment
>about the packaging of SAS, S-PLUS and SPSS is delightfully
>to the point (and, had I been there at the time, I'd have
>bought him a beverage of his choice in appreciation).
>And, to extend his comment, where for instance does that leave
>WinBUGS (see my previous mail)? Granted, perhaps, that in the
>   Spiegelhalter, D. J., Thomas, A. and Best, N. G. (2000)
>     WinBUGS Version 1.3 User Manual. Cambridge: Medical Research
>     Council Biostatistics Unit.
>     (Available from www.mrc-bsu.cam.ac.uk/bugs.)
>one has a prestigious institute (MRC-BSU) as the publisher.
>But, basically, it's the same self-justifiying ordinance all
>over again: the creators of the software produce their own
>User Manual and "publish" it. The publisher of S-PLUS and its
>documentation is "Insightful Corporation": beyond their
>achievement in developing S-PLUS, whence their prestige?
>(Granted, again, if you look inside the books, you can find
>famous names listed as contributors; but we at R can claim
>many of these same names ... ).
>The issue of "peer review" of software citations has been
>raised. Where is the peer review of the S-PLUS or SAS or
>SPSS User Manual? OK, the thing that needs peer review is
>the software itself. In the "Review" section of journals
>you can find critical reviews of software. But these are
>not referee'd and usually represent the writer's own views;
>furthermore, such reviews are not often cited in articles
>where the software has been used to obtain the results in
>the article. People just use the stuff, and cite the User
>Manual or the "corporate packager".
>Also, from time to time you come across published referee'd
>articles which analyse the performance of software for
>particular purposes, or compare different software packages
>for a given purpose. These could be viewed as peer review,
>but, again, are rarely cited by the simple user.
>Moreover, suppose that citing User Manual or corporate packager
>is considered to justify the use of the software in an article.
>I have seen many articles where such-and-such software was used
>(I mention no names) and "cited" in these terms, where it is
>known to discerning users that the software does not correctly
>address the tasks it claims to work for, or gives incorrect
>results. Yet the "citation box" had been duly ticked, and the
>article has successfully gone through the editorial process.
>This is not peer review either -- and could only come close to
>it if the journal editors' panel of referees could be assumed
>sufficiently knowledgeable to be discriminating about such things.
>But in that case you don't really need a citation: the referee
>says "Used [e.g.] S-PLUS, yes, that's fine for this job and
>I could reproduce the analysis using that software and trust
>the results. Passed." And such a referee could say exactly
>the same for R.
>The true peer review of software is done by discriminating
>users of the software. Peer review of R is done *HERE* (as
>well as other places).
>[I've had the experience of using R for analysis, only to
>have the analysis repeated in S-PLUS purely for the purpose
>of citing S-PLUS rather than R for publication ... ]
>However, the follow-ups in this thread are departing somewhat
>from Gordon Smyth's original wish to pay tribute, in citation,
>to the growing team of people who have, over the last 10+ years,
>contributed centrally to making R what it is.
>Thinking about that question, I can't come up with a better
>idea than his: an up-to-date "Ihaka and Gentleman", which
>does due honour to the greatly enhanced riches of R, and R's
>solidity and quality, at the present time!
>But maybe this might take a book ...
>Best wishes to all,
>E-Mail: (Ted Harding) <Ted.Harding at nessie.mcc.ac.uk>
>Fax-to-email: +44 (0)870 094 0861
>Date: 14-Jun-05                                       Time: 10:45:39
>------------------------------ XFMail ------------------------------
>R-devel at r-project.org mailing list

Simon Blomberg, B.Sc.(Hons.), Ph.D, M.App.Stat.
Visiting Fellow
School of Botany & Zoology
The Australian National University
Canberra ACT 0200

T: +61 2 6125 8057  email: Simon.Blomberg at anu.edu.au
F: +61 2 6125 5573

CRICOS Provider # 00120C

More information about the R-devel mailing list