Growth of Open Access

Several studies have looked at the growth of OA in the world of scholarly research, both broadly and by discipline. Bo-Christer Björk, Annikki Roos, and Mari Lauri (2009) made one such early attempt by searching for green OA versions of articles published in 2006 and found they made up 11.3 percent of the research. A follow up of science research articles published in 2008 found a similar share (11.7 percent) for green OA articles (Björk et al. 2010). It went further, though, to discover the share of gold OA articles (8.5 percent) among the study sample. The authors found that a fifth of the articles in the study had been made open whether through green or gold OA. A larger study – this time including 111,312 articles published in 2011 – found a much higher rate of green OA articles, 77.8 percent, by focusing on journals that practice delayed open access (Laakso and Björk 2013). However, a review of such studies estimated that in 2014 green OA articles made up about 12 percent of scholarly research (Björk et al. 2014).

Not all disciplines participate in OA at the same levels. Yassine Gargouri and colleagues (2012) looked at the rates for 14 specific disciplines. In comparing two batches of scholarly articles – those published between 1998 to 2006 and those published from 2005 to 2010 – the authors noted that OA participation grew from 20 percent in 1998 to 2006 to 24 percent in 2005 to 2010. Breaking their results down to specific disciplines, the authors found that mathematics had the highest rate of OA participation in both groups of articles (34 percent in the earlier group, 45 percent in the latter). Clinical medicine had the lowest rate among the earlier group (3 percent), and the arts saw the lowest share among the latter group (10 percent). Overall, most of the disciplines saw an increase in OA participation between the two batches. As Björk and colleagues (2010) found, green OA appeared to be more common for the latter group of articles (21 percent of articles published in paywalled journals) than gold OA (2 percent of all articles).

A study of LIS research on OA found open versions for just 27 percent of the articles in the study sample (Way 2010). Those findings are perhaps lower than expected as librarians have been among the earliest and strongest advocates for OA. Doug Way notes, “In many ways, this involvement likely grew out of the principles of librarianship that align with the OA movement and at the same time out of necessity based on the crisis in scholarly communication that has led to increasing pressures on library budgets” (2010, 302). Holly Mercer (2011) did find higher rates (49 percent) of gold OA participation of academic librarians, discovering that librarians increasingly pursued OA as more OA journals became available.

However, a more recent survey of 1,017 LIS researchers by Wilhelm Peekhaus and Nicholas Proferes (2015) found that although 80 percent had published in a paywalled journal in the past year, only 37 percent reported publishing in an OA journal in the same period. Slightly more than half of the respondents (53 percent) had ever published in an OA journal, and only half had ever deposited their work in an open repository. Reasons librarians gave for not publishing OA included publishing fees, the belief that doing so took longer, concerns over prestige, and concerns of low impact. Almost half of respondents (44 percent) also reported concerns that an OA article would meet with negative views from tenure and promotion committees. In looking at positive factors, Peekhaus and Proferes found that those who had already published OA were more likely to do so than those who had not, suggesting a familiarity with it was likely to encourage OA publishing.

The “do as I say, not as I do” mentality can further be seen in Kristi L. Palmer, Emily Dill, and Charlene Christie’s (2009) survey of academic librarians and their attitudes toward OA. Librarians on average said they agreed with the idea that they should encourage faculty to self-archive their work but then reported only occasionally doing so themselves. The authors noted this dichotomy: “This survey found that agreement with various open access-related concepts does not constitute actual action” (2009, 328). Michelle Dalton (2013) found that OA publishing did not rank high with LIS researchers, who ranked it as one of the lowest factors they considered in deciding where to publish. “Whilst a lack of awareness of the significance of open access publishing could be relevant in other disciplines, this should not affect librarians and information scientists, for whom OA is a fundamental and widely-supported concept” (Dalton 2013, 53).

A broader look at all researchers’ attitudes toward OA found that a discipline’s culture of self-archiving, a feeling of altruism, the promise of academic reward, and strong technical abilities all positively influenced whether researchers took part in OA (Kim 2010). Factors that negatively influenced the decision to publish OA included concern about copyright, time and effort involved, and older age (Kim 2010). The idea of altruism also came out in another, similar study of New Zealand researchers, in which the top reason given to deposit their works was to ensure others had access to it (Cullen and Chawner 2011).

Calls for OA Publishing

Several researchers and even national organizations have heralded the idea that librarians in particular should take active part in OA. Nathan Hall and colleagues point to the American Library Association’s Core Values, noting that they “call for fostering the aspirations of others seeking entrance into the profession. It is difficult to do this with paywalls in place” (2016, 664). They go on to note that the Core Values also promote the idea that library resources should be free to all and it is “thus incongruous not to apply the same essential values to the scholarly communication of librarianship” (Hall et al. 2016, 664). Way argued that LIS researchers need to be more proactive, saying “If publishers do not allow self-archiving or accept addenda, authors then need to refuse to publish in those venues. It would be at that point that the availability of LIS literature would match the rhetoric of many of its practitioners, and the field could point to its own literature as a shining example of free and open access” (Way 2010, 308). The Association of College and Research Libraries (2016) also recently released a position statement on OA, encouraging librarians to not only publish in OA journals but also advocate on behalf of OA with publishers in their roles as editors and peer reviewers. Stuart Lawson, editor of the Journal of Radical Librarianship, began an OA pledge for LIS professionals to sign during his time as administrator of the blog Informed. The pledge, which more than 100 people have signed, calls on researchers to make all of their work OA whenever possible.

Although no broader group exists to represent all researchers of OA, informal calls for OA researchers to practice what they preach abound (see Image 1). This includes the aforementioned proposal for the Open Access Irony Awards, which argued “Sometimes the lack of access is deliberate, other times accidental – but the consequences are serious” (Hull 2012, para. 4). Martin Paul Eve questioned this negative view of OA researchers, however. He argued that to best spread the message of OA, researchers should stop preaching to the choir and thus should publish in paywalled journals to reach the unconverted (Eve 2013). However, it should be noted that a majority of journal publishers do allow authors to openly deposit versions of their articles (SHERPA/RoMEO 2017). Thus, OA researchers who choose to publish in paywalled journals can still take part in the field they study through green OA.

Image 1. Screenshot of a July 28, 2017, tweet by user @AcademicPain.

Research about OA

So far, just one study has attempted to look at whether OA researchers are heeding the call to practice what they preach, and it focused on LIS authors. The study, by Jennifer Grandbois and Jamshid Beheshti (2014), gathered articles published from 2003 to 2011 in English-language journals from a library science–specific database and deleted articles that did not have an LIS author for a total of 203 articles. About 94 percent of them supported OA, but 40 percent of the articles remained behind a paywall, with 36 percent published in an OA journal and 24 percent openly archived. The authors did not attempt to explain the reason for this gap but noted that “while most people recognize the need to save energy and recycle waste, it takes much more than just awareness to get them to change their habits on a large scale. It takes a combination of measures of many different kinds” (Grandbois and Beheshti 2014, 20).

Other studies have started to look at this burgeoning field from other viewpoints. Zao Liu and Gang Wan (2007) found that in the early days of research on OA, LIS journals played a major role, publishing 54 percent of the articles. A more recent study found that LIS journals continued to dominate the field, as Learned Publishing, D-LIB Magazine, and the Journal of the American Society for Information Science and Technology (JASIS/JASIST) were some of the most cited journals in OA articles, although other journals, including Nature, also ranked highly (Düzyol, Taşkın, and Tonta 2010). Sandra Miguel, Ely Tannuri de Oliveira, and Maria Cabrini Grácio (2016) later identified The Serials Librarian as the publisher of the most OA research articles published up until 2015, followed by Learned Publishing and JASIST, tied for second. Although the authors did not track actual OA participation, they did find that a majority of the journals OA researchers publish in allow for article deposit, arguing “thus, it is observed that there is a concern from the authors in reconciling the theme under study and the type of journal chosen for the socialization of new knowledge generated in this subject” (Miguel, de Oliveira, and Grácio 2016).

Güleda Düzyol, Zehra Taşkın, and Yaşar Tonta (2010) found a steady growth in articles about OA, from just two published in 2000 and to a high of 62 articles in 2007 before seeing declines over the next few years. The most common areas of research in those articles included institutional repositories, OA journals, and the broader theme of scientific communication (Düzyol, Taşkın, and Tonta 2010). Another study by Rosângela Schwarz Rodrigues, Vitor Taga, and Mariana Faustino dos Passos (2016) has further expanded on research themes and identified overviews as the most common at 28.2 percent, followed by perceptions of OA second at 21.6 percent, and economics of OA third at 13.3 percent. The study shows that growth of OA was the most common research theme in 2014 and 2015.

OA Finding Tools

In the studies that searched for OA versions of articles, scholars tended to rely on Google and Google Scholar to find open versions (Lyons and Booth 2011; Antelman 2004; Wohlrabe and Birkmeier 2014; Björk et al. 2010; Matsubayashi et al. 2009; Norris, Oppenheim, and Rowland 2008; Way 2010). Although researchers would sometimes use other tools as well, such as PubMed Central, OAIster, and OpenDOAR, the better results of Google and Google Scholar persuaded researchers to rely on them. For instance, Michael Norris, Charles Oppenheim, and Fytton Rowland(2008) reported that Google and Google Scholar found 86 percent of the open versions, whereas OAIster and OpenDOAR found just 14 percent. However, Way (2010) voiced concerns with Google Scholar after finding it failed to return articles published in gold OA journals. He did note that he continued to use it because so many researchers use it themselves.

Several new tools have appeared in the past few years that offer alternatives specifically for finding free or open versions of articles. These include Lazy Scholar (LS), a Chrome extension that searches Google Scholar to let readers on a paywalled page for an article know if a free version is available; the Open Access Button (OAB), which uses both a website and a Chrome extension to let readers search for open articles; and Unpaywall, the newest offering, which also acts through a Chrome extension to search for open articles via the DOI.

Collecting Articles

The first part of the study sample collection focused on finding articles published in paywalled journals. The study searched the term “open access” in the topic field and then filtered the results to remove articles that had been published as OA. This returned 10,265 articles. A quick examination showed that this was likely because the WOS OA filter failed to catch hybrid OA articles on other topics. Because of time constraints, the search was further filtered to include only articles published in LIS journals, as prior studies show they have historically published a large amount of research about OA (Liu and Wan 2007; Miguel, de Oliveira, and Grácio 2016), which returned 521 articles.

To ensure that the study sample did not focus too much on LIS research, a second search was performed limiting the search for “open access” to titles only and using the same parameters as the first search except for limiting to any journal subject. After deduplication, this search returned 525 articles not found in the first search. These 1,046 articles constituted the initial study sample of articles about OA published in a paywalled journal.

Metadata for the articles from both searches were exported into a spreadsheet. Titles and abstracts for the articles were checked to see if the articles were actually related to the study of OA. The study deleted articles if they referred to OA in another concept (such as open access fisheries), if OA was only a tangential part of the study, or if it was not clear. After this filtering, 526 articles remained in the study sample.

The second part of the search focused on finding articles published in gold OA journals. It followed a similar method as the previous steps, although this time filtering the WOS results to include only articles published as OA. This resulted in 164 articles. The total study sample size was 609 articles.

As stated before, the WOS OA filter is not perfect, grouping hybrid OA articles in with paywalled articles. Whenever these errors were found, the study moved the articles to the appropriate category.

Finding Open Versions

The next step of the study used the Chrome extensions for Google Scholar, Lazy Scholar (LS), Unpaywall, and the Open Access Button (OAB) to look for green OA versions of paywalled articles. All four were used on each article.

Searches for open versions of paywalled articles were performed away from the author’s university to ensure the institution’s subscriptions to various journals did not impact results.

The study also tracked the discipline of the first author of every article in the study sample and did so first by checking author information in the article and then, if needed, searching the web for more information. If the author’s discipline could not be found or was unclear, it was marked as such.

Finally, the study also tracked the copyright owner for each article and whether an article was gratis or libre (for articles published in gold and hybrid journals). This information was collected from the article and, if not available there, from the journal’s website. If the copyright owner was not clear, it was marked as such. If it was not clear if an article was licensed for reuse, the article was marked as being gratis.

All data collection for this study took place in April 2017.

Issues with Methodology

Although WOS includes a wide coverage in terms of disciplines, it is limited in that it is selective of what journals it includes, aiming for what it considers the best (Testa 2016). Using only this database to collect articles therefore excluded any research articles about OA that were not published in these journals. However, time constraints prevented a wider search and consideration of more articles, and as WOS has a reputation for carrying most if not all the traditional prestige journals, which are often a big draw for those seeking tenure and promotion, it was decided that the database would suit the purposes of this study. Other databases, such as Academic Search Premiere or even Google Scholar, could be used to find more articles.

By focusing part of this search on LIS journals, the study might also overemphasize articles by LIS researchers and underrepresent research by non-LIS researchers.

Another issue with this methodology is it might have missed green OA versions of articles, even after using the four search tools. Although studies have found Google Scholar to be one of the most effective ways of searching for open versions, Way (2010) has shown it is not perfect. Therefore, it is possible that this study undercounted the number of green OA articles.

Proportion of Open Availability

When looking at the entire study sample (N=609), articles published in gold OA journals made up the largest portion with 31.7 percent (193 articles), followed by green OA articles at 27.6 percent (168 articles). Paywalled articles made up 26.3 percent of the sample (160 articles), and hybrid OA articles made up 14.4 percent (88 articles) (Figure 1). A total of 73.7 percent of the articles were made OA in some way.

Dividing the articles by those published as open and those published behind a paywall, 53.9 percent were published in a paywalled journal, and 46.1 percent were published as open. Of the articles published behind a paywall, 51.2 percent were later deposited and made open. Of the articles published as open, 68.7 percent were published in a gold OA journal.

Although not constant, the number of articles about OA has increased over time, whether published in a paywalled, hybrid, or open journal (Figure 2).

Figure 1. All articles broken down by type.

Figure 2. Articles by year published broken down by type. This does not include articles published in 2017.Title: Longitudinal participation in OA - Description: Breaks down the change in each type of article (gold OA, green OA, paywalled, hybrid OA) from 2010 to 2016), with each one showing an increase over time.

Author Disciplines

Although the study tracked each first author’s field of study, there were so many (ranging from various fields in the sciences, social sciences, and humanities) that most fields had less than a handful of authors. The study was also concerned with incorrectly labeling some authors, as it was clear they came from an interdisciplinary background. Instead of dividing each author up by their respective disciplines, then, this study opted to compare LIS researchers to non-LIS researchers.

LIS researchers made up the largest contingent of any discipline, serving as the first author for 225 articles in the study sample (36.9 percent). Within articles with a LIS first author, 29.3 percent were paywalled with no open option available, compared to 25.6 percent of articles with a non-LIS first author (Figures 3, 4). A greater share of non-LIS authors also published in open journals (35.5 percent of articles published by non-LIS first authors) than did LIS first authors (26.7 percent of articles by LIS researchers) . A chi-square test found a p value of .032 when comparing the differences in OA rates between LIS and non-LIS researchers, meaning the evidence strongly supports a statistical significance in their differences.

Figure 3. Articles with a LIS researcher as first author.

Figure 4. Articles with a non-LIS researcher as first author.

Copyright

Authors were more likely to retain copyright when they published in gold OA journals, with almost two thirds (64.2 percent) doing so, compared to 58 percent of authors publishing in hybrid journals (Figure 5). Publishers owned the copyright in 18.1 percent of articles published in gold OA journals and in 19.3 percent of hybrid articles. There was one instance of the author and publisher sharing ownership to copyright in an open journal.

Breaking down copyright ownership by author discipline, LIS researchers retained ownership in 66.7 percent of their articles, and non-LIS researchers retained ownership in 63.2 percent of their articles. A chi-square test comparing these variables returned a p value of .122, so it does not appear to be a statistically significant finding.

Of hybrid articles, 70.5 percent appeared to be only gratis, lacking any indication of reuse. Just 29.5 percent appeared to be libre, whether through some notice of allowed reuse by the journal or a Creative Commons license (Figure 6). In contrast, 80.8 percent of articles published in gold OA journals were libre, usually by incorporating a Creative Commons license. Just 17.1 percent of these articles appeared to be gratis. Four of the articles published in College & Research Libraries, now a gold OA journal, were published before the journal began using Creative Commons licenses and thus are only gratis although published in a now-libre journal (shown as Other in Fig. 6).

Figure 5. The number of gold and hybrid OA articles in which the author owns the copyright, publisher does, it's unclear, or both own.

Figure 6. The number of hybrid and gold OA articles that were published for free with no reuse licensing (gratis) and those published as both free and licensed for reuse (libre).

Acknowledgments

The author would like to thank Jon Bull at Valparaiso University and Eleni Castro at Boston University for their thoughtful comments and editing of this article.