Monitoring the transition to open access through its mode of implementation: A principal component analysis of two surveys

• Published in: PloS one Vol. 17; no. 7; p. e0271215
• Main Authors: Keiko Kurata, Keiko Yokoi, Tomoko Morioka, Yukiko Minami, Masashi Kawai
• Format: Journal Article
• Language: English
• Published: Public Library of Science (PLoS) 01-01-2022
• ISSN: 1932-6203
• DOI: 10.1371/journal.pone.0271215

Open access (OA) is transforming scholarly communication. Various modes of OA implementation have emerged, which reflect the complexity surrounding OA development. This study aimed to examine this development from the perspective of how OA is implemented. The sample comprised 2,368 randomly selected articles published in 2013 and 2,999 published in 2018 indexed in the Web of Science. We also conducted searches in Google and Google Scholar in 2015 for articles published in 2013 and in 2020 for articles published in 2018. Selected articles were categorized as either an "OA article," "electronic subscription journal article," or "not available online." OA articles were classified into 10 implementation modes: Gold, Hybrid, Delayed, Bronze, Subject Repositories, Institutional Repositories, Personal/Institutional Websites, Academic Social Networks (ASNs), Others, and Web Aggregator. Overall, 56.5% of all sampled articles in 2013 were available for free on at least one website in 2015, while 61.7% of all sampled articles in 2018 were freely available on at least one website in 2020. Concerning implementation mode, ASNs had the highest frequency (44.4% in 2015 and 56.0% in 2020), followed by Subject Repositories (35.0% in 2015 and 39.6% in 2020) and Gold (24.1% in 2015 and 37.4% in 2020). To obtain an overview of OA implementation, we conducted principal component analysis with OA implementation mode as the variable for both 2015 and 2020. The first principal component was the axis indicating the number of overlapping OA implementations for each article in 2015 and 2020, while the second principal component was the axis orthogonal to the first, which was difficult to interpret. We identified three groups of OA implementation in each plot of the principal component scores for articles in 2015 and 2020; however, the OA implementation of each group differed in 2015 and 2020. This diversity reflects the respective positions of various stakeholders regarding OA.