Walt at Random

Thanks to Dietrich Rordorf’s comment on a post at Scholars Kitchen, I am now aware of MDPI’s sciforum (the link is to the journal reviews/statistics section), which among other things “aims at publishing statistics and rankings of scientific and scholarly journals and their Publishers.” Quoting from the disclaimer:

Statistics are automatically computed from available data, and are not manually curated. While we have made every possible effort to provide meaningful statistics, we can not guarantee the correctness or accuracy of any of the statistics. Statistics might be recomputed anytime without notice. Access to statistics might be disabled anytime without notice.

Quoting the section on Open Access:

Data about papers published under open access licenses are currently collected from two providers: DOAJ (licensing information available on journal-level but only for journals that publish exclusively under open access licenses), and Publisher website metadata (licensing information available on a per paper-level for some Publishers). We will include PubMed Central article-level licensing data in a future update. Because many hybrid journals do not offer metadata or licensing information which are easily machine-readable, the statistics about open access content are likely too low. E.g. JR reports 270’000 open access papers for a total market size of roughly 2.4 million papers for 2013 (which is about 11% open access papers). In reality the share of open access papers might be much higher if all papers published under open licenses in hybrid journals could be easily and properly counted). Green open access, i.e. self-archived pre- or post-prints are currently not included.

The section also specifies where data comes from.

I’m delighted there is such a source. Think of the rest of this as additional data (yes, I’ll be emailing a note to Rordorf, but I’m not sure how he can blend manually-counted and automatically-gathered data).

Note that I don’t include “hybrid” OA in my counts at all, partly because there’s no good way to count it, partly because it’s consistently the most expensive form of OA and, I believe, the wrong way to go about OA. Neither does this site because there’s no good way to count it.

Number of Journals

Sciforum shows 25,064 journals publishing at least one article in 2014, of which 3,693 journals are Gold OA—and the chart shows that as being down from 3,990 with at least one article in 2013.

My study, excluding questionable journals and those not in DOAJ, shows 8,760 gold OA journals publishing at least one article in 2014, but that number is down slightly, from 8,960 in 2013.

That’s an enormous difference, one that I believe speaks to the limitations (at this point) of automated data gathering for OA. Even leaving out the global south can’t really account for omitting more than half of the active journals. (Sciforum does not indicate that it’s limited to STM, so I’m assuming it’s not.)

Number of Articles

Sciforum shows 2,423,122 articles in 2014, up from 2,248,966 in 2014. So “roughly a quarter million” seems like a plausible estimate for the total article production.

But: Sciforum shows 302,339 OA articles in 2014, 279,967 in 2013, 250,237 in 2012 and 196,508 in 2011.

The Gold OA Landscape 2011-2014 shows 482,361 articles in 2014, 440,843 in 2013, 394,374 in 2012 and 321,312 in 2011.

Those are also enormous differences, although slightly smaller percentage-wise. To wit, my actual count (again omitting hybrids, questionable journals and journals not in DOAJ) is 60% higher for 2014, 57% higher for 2013, 58% higher for 2012 and 64% higher for 2011.

If we assume that subscription journals can be measured more accurately through automated means (an assumption I’m a little wary of making), then actual article totals for 2014 are around 2.6 million, of which around 18% are in gold OA journals.

My main takeaway: at this point, automated data gathering severely undercounts the OA field—which, given the ludicrous amount of time spent gathering data manually (but hey! book sales are already up to…well, five copies so far), is at least a trifle reassuring.

I thought it might be interesting to look at Outsell’s Open Access 2015: Market Size, Share, Forecast and Trends (which the CCC seems to have made openly available, or at least it was for a while—if it’s back to $2,500, my apologies)in light of The Gold OA Landscape 2011-2014. Is there anything useful or mildly controversial to say?

Definitional Differences

Outsell estimates OA as “about 1.1% of the total 2014 STM market and 4.3% of the STM journals market”—but remember that this is a dollar share, not magnitude. Also, it’s STM, while a sizable chunk of the OA universe as I measured it is humanities and social sciences.

The methodology is entirely different, of course: Outsell develops estimates where I tried to do a universal count. Outsell also has industry contacts, which I entirely lack.

A third key definitional difference, especially given Outsell’s finding: I explicitly exclude “hybrid” publishing—not only because I think it’s a trap but because it’s essentially impossible to count.

Also, Outsell seems to define “megajournals” based on crossing subject boundaries, which strikes me as odd; I’d define them as journals with very large numbers of articles. There are hundreds of interdisciplinary journals; there are only a few very large OA journals.

Questionable Statements

Outsell says (p. 5) “Hybrid currently prevails as the Gold model.” If “prevails” means there are more hybrid journals than there are Gold OA journals (by my definition and that of DOAJ, a hybrid journal cannot be gold OA; otherwise it wouldn’t be hybrid), then that may be true, as so many big publishers will happily take big bucks to make an article OA (sort of) in any of their journals. If “prevails” means that there are more OA articles published in hybrid journals than in Gold OA journals, that would be astonishing: it would mean there were close to a million OA articles published in 2014 (not including green OA). I find that hard to believe, and I don’t know how Outsell could determine this number, so I’ll have to assume “prevails” has to do with number of journals.

On pages 5-6, Outsell offers a proliferation of terms and models including “Platinum” OA, “Gold for Gold” and more. As you read this section, it’s ever more clear that the point of Outsell’s report is to inform publishers how they can best maintain and increase revenues—so, for example, institutionally-sponsored OA is not included in the list of Gold models because “it does not produce revenues.” If it doesn’t generate $Gold, it isn’t worth discussing.

Market Size and Forecast

Outsell estimates the 2014 total STM market at a breathtaking $26.2 billion but the serials market at “only” $6.8 billion. Given other estimates of $10 billion, I can only wonder whether this means that the non-STM journals market is $3.2 billion (which seems unlikely) or whether something else is going on.

Outsell’s numbers for APCs are $290.4 million in 2014, $252.3 million in 2013, and $171.9 million in 2012. But those numbers include articles in “hybrid” journals, which Outsell says are the more prevalent model.

My own figures—not allowing for discounts and waivers, but also not including hybrid publications—are $305.4 million in 2014, $241.9 million in 2013, and $195.5 million in 2012. Those numbers do include the humanities and social sciences, but HSS only accounts for $9.5 million in 2014, $7.7 million in2013 and $6.6 million in 2012.

What’s interesting (to me) is that, if you subtract HSS from my figures, they’re not wildly different from Outsell’s numbers (Outsell’s numbers are about 2% lower in 2014, 8% higher in 2013, and 9% lower in 2013). I could conclude that hybrid APCs don’t really amount to much, or that Outsell defines STM more narrowly than my STEM+Biomed, or…

Megajournals

Outsell says that there are approximately 20 megajournals—but also that they published fewer articles in 2014 than in 2013. Looking at the graph, it appears that Outsell is saying that all megajournals put together published about 40,000 articles in 2014, as compared to 43,000 or so in 2013 and 25,000 or so (?) in 2012.

But it’s hard to tell what Outsell considers to be a megajournal. Let’s look at some subsets within the DOAJ landscape (among the 9,512 I report on fully):

• Journals with at least 1,000 articles in some year 2011-2014: There are 40 of these. I count the article totals as 38,830 in 2011; 56,827 in 2012; 70,986 in 2013; and 88,315 in 2014. That’s a much smaller percentage increase from 2012 to 2013 than Outsell’s graph seems to show—but my figures show between 24% and 25% increase in articles from 2012 to 2013 and again from 2013 to 2014. So: I’m showing a healthy increase from 2013 to 2014, not a decrease.
• Cut that down to journals with at least 1,500 articles in one of those years (14 of them), and you still get a healthy increase each year: 27,052 in 2011; 39,683 in 2012; 49,131 in 2013; and 61,844 in 2014.
• Trim it to the eight journals with at least 2,000 articles in one year, and there’s still an increase in each year:23,461 in 2011; 35,145 in 2012; 43,667 in 2013; and 51,646 in 2014.
• Include only “other sciences” (my term for interdisciplinary STM journals) and PLOS One, and use 1,000 as the cutoff, and I get 17,167 articles in 2011; 28,168 in 2012; 38,604 in 2013; and 43,443 in 2014.
• If I include PLOS One and all “Other Sciences” journals (171 journals in all), I get 23,041 articles in 2011; 35,502 in 2012; 47,607 in 2013; and 54,210 in 2014.

I have no doubt that Outsell is able to define some set of “megajournals” that declined in article count from 2013 to 2014. Since PLOS One alone increased (very slightly, from 31,509 to 31,882), that presumably means that all the other megajournals combined went from around 11,500 articles in 2013 to around 8,000 in 2014. That’s a surprisingly large drop (although one very big and very badly run “megajournal” could account for all of it). Certainly possible—again, depending on how you define megajournals.

Competitive Landscape

Table 4 in the Outsell report is interesting if only because of the numbers: the 14 publishers listed publish a total of 11,740 journals in 2014 (which would appear to be about half of the total) but only 1,505 Gold OA journals (and that includes Hindawi’s 438 and the whole set of BioMed Central journals), fewer than one-sixth of the Gold OA total. (Take away Hindawi, which only publishes Gold OA journals, and you’re down to less than one-eighth of the gold OA total). But note that those 14 publishers claim to have 8,404 hybrid journals.

Ah, but Outsell’s only looking at STM. That does make a difference, because there are so many HSS gold OA journals (more than 4,000 in my study): If you remove those from my count, I’m left with just under5,500 journals. The 14 biggies still account for less than one-third of the STM Gold OA total (around one-fifth without Hindawi), but that’s an improvement.

Overall

I question the supposed decline in megajournal publishing activity, but that’s a matter of definition.

I don’t particularly question the estimated APC totals—unless Outsell is seriously claiming that hybrid publications account for most APCs, in which case I’d question them a lot: I’d believe that waivers and discounts might reduce my numbers by, say, 15%-20%, but that would still leave a huge gap.

Since Outsell doesn’t estimate overall article counts at all, my primary focus, I have no comments.

In general, given differences in definition, the only fault I might find with Outsell (I can’t fault them for a laser focus on $$$) is the decline in megajournal publishing from 2013 to 2014–and, again, I’m sure they managed to define a group of journals that comes out that way.

First, the update: As of early this morning (September 25, 2015), there have been 1,198 downloads of Cites & Insights 15:9 (1,061 of them the single-column version). Also as of now, sales of the book have more than doubled, to a total of four paperback copies and one PDF ebook. For those of you hoping to see an anonymized version of the data available on figshare, we’re now one-seventh of the way there; one more sale would put us one-sixth of the way there.

Now, a quick note about agriculture (which, for this study, includes aquaculture, fisheries and other aspects of raising and processing plants and animals, including food and some aspects of nutrition). It’s one field that gets only a stub chapter in the excerpted version—but it’s also a field where including the whole world of DOAJ-listed journals makes a big difference.

To wit: where the interim report and the Walt at Random post (which omits one graph) show 286 journals (excluding C, as almost all of the new book does) publishing 14,879 articles in 2014, the complete study includes 418 journals (excluding C) publishing 19,861 articles in 2014. That’s a full third more articles from 46% more journals (as expected, the added journals tend to be smaller). There’s also a shift toward no-fee journals: of those actually publishing articles in 2014, the no-fee percentage was 62% for the smaller group and 66% for the more complete set—and while it’s still true that (as with most STEM fields) a majority of articles are in APC-charging journals, the percentage of articles in no-fee journals went from 44% for the smaller group to 48% in the larger (and for that larger group, articles in no-fee journals were at least half of all articles in 2013, 2012 and 2011).

In the smaller report, the average cost per article for articles involving APCs was $734.07, or $397.05 for all articles. Those figures don’t change enormously: for the fuller report, the average is now $729.95 for articles in APC-charging journals or $349.89 across all articles.

Agriculture OA journals in 55 countries published articles in 2014; 19 countries published 300 articles or more. It’s an interesting list from top (Brazil, with more than the next three countries put together) to bottom (Egypt—among the 19 top countries, that is, a list that also includes Bangladesh, Romania and the Czech Republic).

For more details on this area and on 27 others (together with lots of other stuff), buy the book.

Oh, and for graphics fans, here are the two graphs from Chapter 12, Agriculture:

Figure 12.1. Agriculture articles per year

Figure 12.2. Agriculture OA journals by starting date

The first in an occasional series noting interesting items in The Gold OA Landscape 2011-2014 that aren’t in the excerpted version in Cites & Insights 15:9—or observations drawn from the book.

Chapter 6, Country of Publication, is mostly tables showing countries in which gold OA journals are published arranged in different ways with additional facts. The C&I version includes Table 6.2, countries ranged by percentage of OA journals that don’t charge APCs (from the 100% free of Cuba, Venezuela, Denmark, Costa Rica, Estonia, Philippines, Sri Lanka and Ecuador down to the only four where at least 70% of gold OA journals do charge author-side fees: the United Kingdom, United Arab Emirates, New Zealand and Nigeria). It also includes all or part of Table 6.3, articles by country in 2014 for countries with at least 1,000 (and percentage in no-fee journals) from 89,485 in the U.S. (17% in no-fee journals, although 62% of U.S. gold OA journals don’t charge fees) down to 1,001 in Peru (94% in no-fee journals: 96% of the OA journals in Peru don’t charge fees).

The issue doesn’t include Tables 6.1, 6.4, 6.5 or 6.6, illustrating different aspects. For those, you need to buy the book.

Here’s one I could have included in the book but didn’t, thrown in as an online bonus: potential revenue in 2014 by country. There are 75 countries where at least one APC-charging journal published articles in 2014, so it’s a long table. Here goes:

If you haven’t purchased the book (paperback or site-licensed PDF) yet, please do. (Yes, it will eventually be available—paperback only—from Ingram, Amazon or Barnes & Noble, but that will be something like two months from now.)

What the heck: as more copies sell, I may try to find more interesting new stuff that’s not in the book (but might be next year, if the project continues).