« International Council on Nanotechnology and Rice University Launche Environmental Health and Safety Nano Database | Main | Brief Review: Information Sources in Engineering - 4th Ed, Edited by Roderick A MacLeod and Jim Corlett »

PNAS, Open Access, and Levels of Interest - Commentary by George Porter

.: George Porter, contributor to STLQ and Librarian at Sherman Fairchild Library, CalTech, offers a thoughtful analysis on open access and the impact on readership, using PNAS as a test case. - Randy.

There has been a great deal of speculation about the willingness of authors to pay fees to provide greater access to their articles. Contributors to the discussion on all sides of the Open Access (OA) debate (and there are certainly more than two schools of thought on the subject) have been able to find numbers to support their theses. However, it has proven exceedingly difficult to find a virtually unassailable test case from which all interested parties could concede that valid lessons might be drawn.

An opportunity presents itself with the Proceedings of the National Academy of Sciences of the United States of America (PNAS) is a respected, established scientific journal.

PNAS is a signatory of the DC Principles and has chosen a 6 month window of exclusivity for subscribers. In addition, PNAS instituted a Walker/Prosser model of OA by the article in late 2004. Under this model, authors may pay an additional fee to provide immediate global access to their published article.

I decided to do a narrowly drawn experiment: established journal, no charitable grants or personal fortunes underwriting the publishing enterprise. No likelihood of the journal altering its editorial standards to enhance revenue. This seems like a very reasonable test case -- the most easily tested hypothesis being: Will authors, all other factors being equal, pay an extra sum to provide readers with earlier (in this case -- six months) access to their findings? Is reducing the amount of time that an article is only available through a subscription worth the additional expense to authors?

Using the PNAS Open Access articles section of PubMed Central, I observed 268 articles for which the OA fee ($750 or $1000, reduced rate for authors at subscribing institutions) has been paid over the last 6 months, ~12% of the published articles. Of those, 24 articles have been the cover feature article, >90% of the cover articles in the period.

Obviously PNAS has identified a service for which a segment of their authors are willing to pay. The proportion of authors willing to pay increases significantly with the self-perceived significance of the work being published.

Does Open Access have any impact upon readership? By examining the 50 Most Frequently-Read Articles in PNAS, which is based only on the article views from the Highwire home of PNAS, it may be possible to draw some additional conclusions. [Article views during July 2005]

I examined the first twenty pieces from the top 50.

  • Open Access articles: 8
  • Open Access articles, prepagination: 2 (including #1)
  • Cover articles: 6 (4 Open Access)
  • In this issue (Free): 3
  • DC Principles free: 5 (1 Open Access)
  • Remainder: 3
Some of the entries fit in more than one category. Four of the Open Access articles, out of a total of 8 for which an OA fee was paid, were featured on the cover of the journal. One of the Open Access articles would have been freely available in July 2005 under the PNAS commitment to the DC Principles, which they have chosen to implement as a 6 month embargo.

Three of the top 20 most-read articles in July 2005 required a current subscription and were not featured on the cover of a June or July issue. A preliminary conclusion might be made that ease of access enhances the likelihood of a large readership. The notoriety of cover articles is not to be underestimated in garnering readers. There is a caveat, though. Five of the six cover articles were in the top 8; the 6th was #18 and not Open Access.

-- George Porter


George - this is a very helpful piece of research.
One suggestion:

It would be helpful to point out that the PNAS charge for open access is actually a surcharge, on top of the $70 per page charge, plus charges for images, tables, etc. That is, the OA charge for a 12-page article, with graphs and tables, may be less than half the total author payment.

George, you say both:
"The proportion of authors willing to pay increases significantly with the self-perceived significance of the work being published."
"A preliminary conclusion might be made that ease of access enhances the likelihood of a large readership."

Is any way of deciding which of these two alternative hypotheses is correct?

In response to Heather Morrison:

The link provided to the Open Access surcharge also shows PNAS's standard page charges. Since I did not review any data regarding the mean, mdeian, or modal length of articles published in PNAS, I chose not to speculate on the apportionment of an author's publication fees.

My primary point, with respect to authors, was whether or not a meaningful fraction of authors are electing to pay an OA surcharge in PNAS. My back of the envelope calculations indicate approximately 1/8 of the articles published in PNAS over the last 6 months were subsidised by their authors or their authors grant funding agencies to permit immediate access.

A secondary point, the Open Access rate for articles featured on the cover of the issues (90+%), was not a something which I anticipated nor was specifically looking to assess it. It was a observation which leaped out of the data, thanks in large part to the color coding of the PubMed Central OA content and the rich metadata provided by PNAS and their publishing partner, Highwire Press.

The breakdown of author fees is totally unrelated to the readership analysis.

In response to David Goodman:

I think that to a large extent the authors' self-assessment of the significance of a to-be-published article is independent of the potentially enhanced readership conferred by Open Access status for the article. The author(s) must decide, before publication, whether or not this specific article is noteworthy enough to pay a surcharge. [Let me stress, this analysis is strictly based on the PNAS implementation of the Waler-Prosser model of Open Access by the article.]

The readership of newly published Open Access article appears to be larger than the readership of the majority of the non-OA articles, based on this analysis of a snapshot (July 2005) of use data from PNAS. Other noteworthy articles did rise to the top as well -- feature articles from the covers fof the last 6 issues all ranked in the top 20. Four of those were Open Access articles, all of which were in the top 10, but one of the non-OA cover articles was also in the top 10.

Timeliness, topicality and ease of access, as Goodman inquires, are ot completely independent aspects of this situation. The author is making an assessment of value vs. expense in advance of publication. Readers are making their decisions in a broader context, and usually 2-4 months after the submission of the article.

I did not review all of the available data, nor did I have access to all of the necessary data to assess the readership of all OA and non-OA PNAS articles published within a 6 month window or viewed in the month of July. My proposition, based on a brief review of the data, is that these are not alternative hypotheses, but in fact, are complementary hypotheses.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)