Since the SKC team is studying article processing charges, it’s important to emphasize that the majority of fully open access journals do not use APCs; our studies of this method are not intended as an endorsement of this business model. The following data is from a follow-up analysis based on the May 2014 survey of a sample of the 2,567 journals listed in DOAJ at that time that had article processing charges, 26% of the 9,709 titles in DOAJ at that time. For details on the sample, see our article in MDPI’s Publications. César’s dataset and documentation explaining the preparatory analysis of subjects in DOAJ will be posted shortly. The following table illustrates the difference in the tendencies of journals from different subjects to charge APCs. Medicine is the discipline most likely to charge APCs; 47% of the journals in our APC sample were from medicine compared to 23% of journals listed in DOAJ at that time. The “subject tendency to charge APCs” is calculated by subtracting the percentage of journals in DOAJ from the % of journals from sample with APC. A positive result means a greater tendency to appear in DOAJ than to charge APCs, while a negative result means a greater presence in DOAJ as a whole than in the APC sample. For example, humanities includes 14% of the journals in DOAJ but only 1% of the APC journals.
| Subject | Number of journals in DOAJ | Percentage of journals in DOAJ | % of journals from sample with APC | Subject tendency to charge APCs |
| Medicine | 2,235 | 23% | 47% | 24% |
| Biology and Life Sciences | 664 | 7% | 14% | 7% |
| Technology and Engineering | 970 | 10% | 13% | 3% |
| Physical Sciences and Math | 666 | 7% | 9% | 2% |
| Science General | 134 | 1% | 1% | -1% |
| Agriculture | 414 | 4% | 3% | -1% |
| Law | 183 | 2% | 0% | -2% |
| General Works | 502 | 5% | 2% | -3% |
| Education | 596 | 6% | 2% | -5% |
| Social Sciences | 1,984 | 20% | 8% | -13% |
| Humanities | 1,360 | 14% | 1% | -13% |
Reference
Morrison, H.; Salhab, J.; Calvé-Genest, A.; Horava, T. Open Access Article Processing Charges: DOAJ Survey May 2014. Publications 2015, 3, 1-16. http://www.mdpi.com/2304-6775/3/1/1
Cite as:
Morrison, H. (2015). Which subjects are most likely to charge article processing charges? Sustaining the Knowledge Commons / Soutenir Les Savoirs Communs. Retrieved from https://sustainingknowledgecommons.org/2015/05/05/which-subjects-are-most-likely-to-charge-article-processing-charges/
Sustaining the Knowledge Commons / Soutenir les savoirs communs 
For what it’s worth, here’s a quick post with two tables (one for 2013 data, one for 2014 but missing seven topics) based on percentage of *articles* that appeared in journals clearly charging APCs. It may offer interesting comparisons. http://walt.lishost.org/2015/05/percentage-of-oa-articles-involving-apcs/
Thanks Walt – many different angles and no lack of work to do, looking forward to your article. Did you use DOAJ content metadata? If so do you find disparities similar to the ones I’ve noted here? https://sustainingknowledgecommons.org/2015/05/06/tech-tip-for-doaj-journals-contributing-article-level-metadata/ This could confound results – one or two really large OA publishers in STM with more advanced metadata exporting capacity could have a real impact. Publishers like BioMedCentral and Hindawi have been around for years and working with PubMedCentral on metadata issues for years, for example.
(The article is actually an issue of Library Technology Reports.) No, I don’t use DOAJ metadata–I used a download of DOAJ’s spreadsheet (on May 7, 2014) as a source of journal titles, starting dates, URLs and subjects. Everything else is based on thousands of direct visits to journal sites. (Documented extensively in Cites & Insights last year; I also looked at the Beall sideshow. You’ll find it all at http://citesandinsights.info )
Walt – I tried to leave this comment on your blogpost here http://walt.lishost.org/2015/05/percentage-of-oa-articles-involving-apcs/ but got a spam message so posting it here.
Walt, this is very helpful. Do you have non-anonymized data? There is no reason to anonymize data gathered from websites that are deliberately and consciously made publicly available, as is the case with open access journals. The reason I ask is because my research involves exploring correlations between this kind of information and other data either available through DOAJ or gathered by my team (e.g. detailed APC information as of May 2014). Our data and documentation can be downloaded from the OA APCs dataverse: http://dataverse.scholarsportal.info/dvn/dv/oaapc/
Note that all of the more than 10 thousand journals listed in DOAJ have articles available, not just the 6,490 (as of last year) with content in DOAJ. Maintaining a minimum number of publications is a criteria for inclusion in DOAJ, and new journals generally have to establish a track record before they can be included. As an aside, there may be good reasons for DOAJ to reconsider; inactive journals and articles published in them that met DOAJ criteria at the time of publication should not be removed, IMHO.
In case others are interested in this type of research, note that it is not always necessary to be able to read articles in another language to figure out the numbers of articles published. For example, I don’t speak a word of Portuguese, but I can look at the Revista Lusófona de Educação http://revistas.ulusofona.pt/index.php/rleducacao/index from DOAJ. From the 8 links at the top of the screen, “actual” for current issue and “anteriores” for previous issues seem like reasonable guesses (one could also try popping these terms into Google Translate). I’m willing to hazard a guess that “Artigos” means articles and the count for the current issue is 10. This won’t work for me for languages based on non-roman characters. From our sample of 1,432 journals last year I only found one publisher (Hans Publishers) where the team was not able to figure out the information on article processing charges (in Chinese). Fortunately my friend and colleague Ian Song volunteered to translate.
Partial responses:
1. The comment was trapped as spam because it had more than one link. 99% of the “comments at Walt at Random are spam, so I have the controls set fairly tight.
2. I explained my reasons for not releasing unanonymized data in a separate post (having to do with approximate article counts for journals that are really difficult to count, and the likelihood that one anti-OA troll would move to discredit what I’m doing), and I think they stand, esp. given the total lack of funding for the work I’m doing.
3. I wasn’t looking for articles *in* DOAJ, I was looking for articles at the URLs listed in DOAJ, and hundreds of journals had already gone bad–noting that this was under DOAJ’s old rules. I’ve talked to DOAJ about a couple of desirable tweaks in their new rules.
4. I agree regarding many non-English sites, and if I was doing this again (which might happen in 2016), I’d use Google Translate in Chrome to see how many more I could handle. It’s not just the article count: I was also looking for actual APC size and clarity, or explicit “no-fee” statement, which is harder to gauge if you can’t read the language.
At this point, I can’t see changing the decision to anonymize freely-provided datasets; that it makes this data less valuable as a wholly free pool of sources for paid researchers to use for other projects is, perhaps, unfortunate. I’m explicitly avoiding journal-naming issues, and plan to continue that.
Thanks for this very interesting post and just as interesting discussion. I don’t see business as a separate entry in your table. Could you tell me under which category it was processed? Thank you.
Good question Marie-France…we’ll look into this..
Hello Marie-France, thanks for your comment. In the data, now available in http://dataverse.scholarsportal.info/dvn/dv/oaapc you will find three columns for the subject classification: Main Category, Sub Category 1 and Sub Category 2. For our research, business was included under Social Sciences (Sub Category 1), as “Business and Commerce.”