Duh?! Anyone or any statistical society can launch “Open Spatial Statistics” or “CSDA w/o Els***” as an electronic open-source journal with the same (a) editorial board, (b) workload, (c) requirements. The typical example is Bayesian Statistics. Zero fees, a minimal cost to ISBA, and a terrific outcome/impact factor. So I do not see how and where Elsevier is contributing!

Your argument does not work well for new journals as their fame is yet to build…

]]>But that’s the point isn’t it. Where does this campaign end: either with nothing; with a price reduction and more flexibility for libraries (which is obviously the best outcome); or with small journals (like Spatial Statistics) closing or becoming a laughing stock due to a lack of quality papers/reviewers/editors. The point that I was (inelegantly) making was that without the Elsevier journals, the journal landscape in statistics would be noticeably poorer. I mean, journals like CD&SA are fairly uneven, but they do contain fantastic, interesting, or just plain useful things that it would be difficult to publish elsewhere!

—————-

Random thoughts:

– When I talk about comp. stats papers, I tend to think specifically of papers that are about computations that are fundamentally of use to statistics, but not strictly about statistical issues. I would take as an example, Håvard Rue’s fast sampling from GMRFs paper, which appeared in JRSSSB. Of those sorts of papers, very very few appear in the “traditional” (aka top four) journals. Of course, this sends us veering into the awful “yes, but is it statistics” territory. These are the sorts of papers that are far easier to publish in commercial journals, or journals in other subject areas.

– I honestly thought S&C was also an Elsevier journal! That’s nice!! I would say it’s one of the few places where you can publish the sorts of papers that I’m talking about.

– JCGS is one of those Journals that I don’t tend to send stuff to because I am completely unable to parse the second sentence of their Aims & Scopes and don’t want to waste their (or my) time. (the first sentence makes it look promising, but it’s really hard to tell what the editorial policy actually is)

– SIAM JSC is interesting. It used to have Statistical in the title, but it was dropped (allegedly) due to lack of interest. It’s certainly one of the more flexible numerics journals, but it’s a long way from a statistics journal.

]]>I believe the only way Elsevier is going to reduce its prices is when the universities stop paying those fees. Simple “Supply and demand” effect.

An example of this is Harvard University

http://www.guardian.co.uk/science/2012/apr/24/harvard-university-journal-publishers-prices

]]>Hey Dan, I am not going to begrudge you the opportunity to publish! And I have myself published a lot of papers in Elsevier journals in the past.

Now, there exists a certain number of computational statistics journals, from JCGS and Statistics and Computing (although too expensive as well!) to Computational Statistics and Siam Journal on Scientific Computing (where I never managed to publish). Major “traditional” journals also take computing topics into consideration: think of the influential computing papers published in JASA, JRSS Series B, Annals of Statistics (admittedly less) and Biometrika.

]]>(the “exceptional article” referred to above is http://dx.doi.org/10.1137/110834469, which is related to the very important practical topic of “optimal preconditioning”)

]]>(Also, before my high horse explodes completely due to over-use, I feel I should point out that I’m a postdoc living from visa to visa and so if the hiring committee [which may not actually include any actual statisticians] doesn’t care about where the paper is published, I find it hard to. And this is regardless of my open source leanings {I sate those with arXiv and a personal website})

]]>To further elaborate (because, why not), take this exceptional article by Michael Stein, Jie Chen, and Mihai Anitescu which essentially gives a way to massively speed up maximum likelihood computations for certain Gaussian random field models. Unquestionably, the interested audience for this paper consists mostly of computational statisticians who, at a guess, are unlikely to read the SIAM Journal on Matrix Analysis. But I’d be hugely surprised if any prestigious statistics journal would deign to publish such an article.

]]>Beyond that, I have another one (or possibly two) paper(s) that is almost finished and they will probably go into an Elsevier journal. Mainly because the topic (a complex, specialist, but definitely statistical useful computational procedure) has no home in any reputable statistics journal and I don’t desperately want to go 10 rounds with a (hopefully sympathetic) A/E. I also don’t want to publish it in a numerics journal as the paper is of actual interest to statisticians.

I really do not think that “traditional” statistics journals (or society journals) do nearly a good enough job with computational statistics to make a boycott of journals that actually would consider publishing my papers (some of which are probably ok or better).

]]>“Voluntary posting of AAMs in the arXiv subject repository is permitted.” Isn’t it what it means ?

]]>