Even with all of the enthusiastic support for open access among librarians, relatively little of the LIS literature is openly available. Perhaps there is some delicious irony in the fact that a self-described "open access heretic" such as myself was the editor of the Journal of the Medical Library Association when it became fully open access back in 2000 -- the first major LIS journal to do so.
Despite the manifest problems with Jason Griffey's 2004 master's paper, it was effective in highlighting the fact that virtually all of the ALA-associated journals were closed. Now, Doug Way has released a paper on "The Open Access Availability of Library and Information Science Literature" that indicates that the situation has improved little in the five years since Griffey posted his paper.
Of course I scanned the paper immediately to be sure that the JMLA was included and to find out what he had to say about it. My interest was immediately piqued by "Table 2: Percentage of OA articles by journal" which says that 94% of JMLA articles are OA. What? All of the JMLA is OA. All of it. Every page. Back to 1898. What gives?
Turns out to be a methodological issue. Way is not actually measuring OA availability; he's measuring the degree to which Google Scholar is effective in identifying OA versions of articles from his set of twenty LIS journals. If I'm calculating the math back correctly, he identified 68 research articles published in the JMLA during 2007, and was only able to identify 64 of them using Google Scholar. I'm dying to know which four didn't show up!
But that percentage was better for JMLA than for the two other OA journals that Way identifies. Google Scholar found only 70% of the articles in Law Library Journal and only 68% of those in College and Research Libraries.
This methodological confusion pervades the discussion section, and one has to read carefully to keep things straight. For example, Way says that, "No journal, including the three OA journals included in this study, had OA versions for 100% of their articles." In fact, what he means is that Google Scholar was not successful in finding all of the OA versions actually available. I know that Google Scholar failed to identify 6% of the OA articles in the JMLA. And it appears that it failed to identify about 30% of those in the other two OA journals (unless those journals are not, in fact, completely OA).
If I had been editing this article, I probably would have tried to put the brakes on his conclusions somewhat. (This was, by the way, the number one editing issue that I had when I was working with the JMLA -- authors claiming more than their results actually showed). His results show that Google Scholar is not a completely reliable means for identifying OA versions of articles. It is probably also safe to conclude that fewer than 50% of the articles published in the non-OA journals have an openly available version, but the specific numbers that he presents should be approached sceptically.
One other point about the JMLA. Way mentions that all of the OA versions that Google Scholar identified were "found in PubMed Central and not on the journal's Web site." This is completely unsurprising, since the JMLA does not have a separate journal website. Except for a very few early release articles, the electronic versions exist only on PMC.
Despite these quibbles, the larger point that Way makes is indisputable -- for all of the advocacy work that librarians have done in support of various OA initiatives, they have not done a very good job of making their own research output widely available and easily discoverable. Our advocacy efforts would be more persuasive if we were taking a more aggressive leadership role within our own field.