Another brief un-hiatus, this one sparked by a question asked by Dave Munger at BPR3:

If you know of a peer-reviewed, open-access journal that does not charge a publication fee, let us know about it in the comments.

Practically every time I talk about OA, online or in meatspace, I hear “I’d like to support OA but I can’t afford it, don’t all those journals charge, like, $2500 per article?”
No. They don’t.
Everyone seems to be thinking of PLoS, never mind that they waive their fees at the drop of a hat; the assumption that most OA journals charge (high) author-side fees is both widespread and completely wrong.
In fact, more than 2/3 of the journals listed in the Directory of Open Access Journals (DOAJ) and more than 80% of OA journals published by scholarly societies charge no author-side fees at all; in contrast, more than 75% of the 247 non-DOAJ journals in a 2005 survey do charge author-side fees (page charges, colour charges, reprint charges, etc) in addition to subscription charges.
Let’s unpack those numbers a little (especially since I generated the first one myself, and so you should take a look at how I did that).
In October 2005, the Kaufman-Wills group published a commissioned survey of journal publishing practices, The Facts about Open Access. The study was initially designed to include only full OA journals (listed in the DOAJ, OA immediately upon publication) and delayed-OA (“embargo”) journals from the HighWire Press stable, but was expanded to include the full range of financial models by inclusion of journals published by the Association of Learned Professional and Scholarly Publishers (ALPSP) and the Association of American Medical Colleges (AAMC). The final report included responses from 248 DOAJ, 85 HighWire, 34 AAMC and 128 ALPSP journals and showed that:

52.8% of DOAJ journals charge no author-side fees at all. The percentage for subscription journals was much lower: ALPSP journals overall (23.4), ALPSP for-profit journals (44.9), ALPSP non-profit journals (10.1), AAMC journals (14.7), Highwire subset (17.6)

These are the figures that Kaufman and Wills summarize as “…more than half of DOAJ journals did not charge author-side fees of any type, whereas more than 75% of ALPSP, AAMC, and HW subset journals did charge author-side fees.”
So — not only do the majority of OA journals charge nothing on the author side, an even larger majority of non-OA journals do charge author-side fees. If the sample is representative, you’re less likely to have to pay to publish if you choose an OA journal than if you don’t.
When I first heard these numbers I thought, as Peter Suber did, that they should “recast the debate” around OA. In January 2006 Peter’s regular yearly predictions included this forecast:

It will start to sink in that fewer than half of OA journals charge author-side fees and that many more subscription-based journals do so than OA journals…. People will stop talking about “the OA business model” for journals as if there were just one. People will talk less about how OA journals might exclude indigent authors and compromise on peer review and talk more about how toll-access journals do so. We’ll start to document the range of models actually in use for OA journals… We’ll get more creative in finding models that suit the range of niches…

He has since called this “the worst prediction I’ve ever made”. I confess myself at something of a loss as to why the Kaufman-Wills study has not come to dominate and reconfigure the OA debate; I can only guess that profit-hungry lowlifes have successfully sidestepped it. In this year’s predictions, Peter expects more of the same:

Because both Hindawi and Medknow have both been profitable for more than year, you’d think that the fact of their success would start to sink in, with corresponding effects on attitudes toward the sustainability of OA journals and interest in their business models. But well-documented truths about OA tend to sink in very, very slowly, because they have to compete with myths, misinformation, and misunderstanding. With regret, I predict more of the same.
In 2005 the Kaufman-Wills Group discovered that the majority of OA journals charged no publication fees at all. In 2006 I predicted that that fact would start to sink in. I was dead wrong. The fact still hasn’t sunk in, and I’ve learned my lesson.
Caroline Sutton and I discovered last month that the OA journals published by learned societies follow same pattern as OA journals overall: most of them charge no publication fees. But while 52.8% of OA journals overall use no-fee business models (from Kaufman-Wills, 2005), we found that 83% of society OA journals use no-fee business models, a significantly greater fraction. However, I’m not predicting that this fact will sink in any time soon. Likewise, we found 425 societies publishing 450 OA journals, a much larger number than the societies known to oppose OA policies. But neither am I predicting that this fact will sink in any time soon. We’ll continue to hear the unargued claim that society publishers are intrinsically vulnerable to OA and predominantly opposed to it.

The Kaufman-Wills study is not the only one of its kind, either. As discussed in the quote above, just last month Peter Suber and Caroline Sutton of Co-Action Publishing released preliminary findings from their ongoing study of OA journals published by scholarly societies. They identified 468 societies which publish, between them, 450 full OA journals and 73 hybrid (“pay-for-OA”) journals. Of the full OA journals, only 75 charge author-side fees — meaning that more than 80% of society journals do not charge such fees.
Finally, there’s me. All of the above got me to wondering what proportion of journals in the entire DOAJ database charge author-side fees (since Suber and Sutton showed that when the dataset was expanded, at least among society publishers, the no-fee percentage went up considerably).
Fortunately, the DOAJ now includes a metadata field indicating whether or not a particular journal charges author-side publication fees. Unfortunately, that field is not included in the downloadable comma-delimited metadata file they make available. Fortunately, it’s not a whole lot of work to make a replacement file by copy-and-pasting from the “browse by title” page. Unfortunately, you have to do this from the new “for authors” section, because the front-page browsing interface doesn’t include the “fee/no fee” field. What’s unfortunate about that, for my purposes (though it’s a wonderful thing overall), is that the “for authors” browse does include hybrid journals, in which regular articles are subscription-only but authors can pay extra to have their work made OA. (In fact, even the logo at the top is different; on the front page you are seeing the Directory of Open Access Journals, but in the “for authors” section it becomes the Directory of Open Access and Hybrid Journals.) The front page says 2971 journals are indexed, but if you browse by title from the “for authors” page, the totals add up to 4638, the database having apparently added 1667 hybrid journals.
There’s probably a smarter way to do this using the OAI-PMH, but that syntax is as impenetrable to me as Ancient High Martian, so I simply pasted the browse-by-title pages into a text document and imported that (colon-delimited) into Excel. Then I filtered on “publication fees”, sorted by yes/no/missing and read off the totals from the row numbers. Horrible hack, but it worked.
Including hybrid journals, we get:

charge publication fees: 2185 (47%)
don’t charge pub fees: 1998 (43%)
fee information missing: 455 (10%)
total no. of journals: 4638

Given the DOAJ definition of hybrid journal, those should obviously be excluded and the data reworked. This is where a smart person would have stopped and waited for the DOAJ to autogenerate the numbers, but I went ahead and deleted the hybrid entries by hand. (Shut up. I wanted to know, OK?) That yields:

charge publication fees: 534 (18%)
don’t charge pub fees: 1980 (67%)
fee information missing: 453 (15%)
total no. of journals: 2967

The second total should presumably be 2971 and it would make sense for the “missing” totals to be the same in both sets, so either there are some errors in the database or I made a couple myself. In either case the errors appear small and make no difference to the percentages, and anyway did I mention this kept me up to 4 am? Actually I suspect some inconsistencies in the database, because the front-page total does not update as quickly as the actual entries, and because there are in fact hybrid journals which don’t charge fees (e.g. Emerald Engineering’s model).
So now we have three studies (OK, two studies and one ungainly hack) showing that a (strong) majority of OA journals do not charge author-side fees, and one of those studies further showing that a strong majority of non-full-OA journals do in fact charge author-side in addition to subscription fees.
Now, can we please put to rest the myth/FUD/whatever that there is only one OA model, the author-side fees/PLoS model? While we’re at it, let’s have a few more closely related ideas go the way of the dodo: that OA journals discriminate against indigent authors (because they charge publication fees — except that most of them don’t); that OA journals will compromise on quality (in order to collect payment for manuscripts — except that most of them don’t); that if most journals went OA, universities would have to pay more in author-side fees (which, remember, most OA journals don’t, but most non-OA journals do, charge) than they do now in subscription fees.
I swiped that list of candidates for memetic extinction from Peter Suber, and you should go read his full discussion, which offers a lot more detail, especially on that last point. Me, I’m going to take a nap and go back to my blog hiatus. But the next time you hear someone talk about the “cost” of publishing in OA journals, please point ’em here.