Open Access Does Not Equal More Citations, Study Finds

A new study suggests that while open access appears to increase the readership of scholarly articles, it doesn’t increase how often they’re cited.

The study stands in contrast with earlier research that suggested open-access articles were referenced by other scholars more frequently.

Philip M. Davis, a postdoctoral associate in the department of communication at Cornell University, was given access to 36 subscription-based journals produced by seven different publishers. In 2007 and early 2008, he randomly made approximately 20 percent of their articles free.

He tracked the number of abstract views, full-text downloads, PDF downloads, and citations within the next year for the 3,245 articles in the study. The findings were published Wednesday in the Federation of American Societies for Experimental Biology Journal.

Free articles were downloaded more than twice as frequently as the paid-access articles, and PDFs of the free articles were downloaded 1.6 times more frequently.

But despite appearing to be more widely read, the free articles were no more likely to be cited than their counterparts.

Mr. Davis suggests that this may be because access is not a problem for most researchers who would cite the articles in their own work.

“For the most part, authors are located at institutions with very good access to the literature,” he says. “If authors don’t have formal access through their library, they seem to get informal access,” such as by asking colleagues at other universities who do have access to send a copy, or getting a copy from the authors.

In a post on Nature’s The Great Beyond blog, open-access advocate Stevan Harnad, a research chair in cognitive science at the University of Quebec at Montreal and professor of cognitive science at the University of Southampton, questions Mr. Davis’s methodology and motivation.

Mr. Davis says he doesn’t see his study as a blow to open access—if anything, he thinks it calls into question the wisdom of looking only at citation counts to measure the impact of a journal article, particularly given the ease of tracking article downloads online. “Twenty years ago, there was no way of measuring readership,” he says.

He says there should be some attempt to take into account the impact that journal articles have on all the readers who don’t conduct their own research on a similar topic.

“When we really think about what citations measure,” he says, “they’re really only looking at access to literature for a very narrow group.”

Return to Top