Comparing conventional and alternative mechanisms of discovering and accessing the scientific literature.

This study compares the bibliographic and full-text coverage of 15 conventional and alternative discovery/access mechanisms: two multidisciplinary library databases (Scopus and the Web of Science Core Collection), five single-subject databases, the integrated library search (ILS) mechanism of Manhattan University, a scholarly search engine (Google Scholar), two web-based scholarly databases (Dimensions and OpenAlex), two academic social networks (Academia.edu and ResearchGate), and two pirate sites (Anna's Archive and Sci-Hub). The analysis is based on known-item searches for 875 target documents in chemistry, materials science, cardiology, public health, economics, education, and psychology. Overall, Google Scholar, OpenAlex, and the ILS are the most comprehensive sources of bibliographic records. Google Scholar's coverage rate is higher than that of all the Manhattan University databases combined, and Scopus-the most comprehensive multidisciplinary library database-has a lower bibliographic coverage rate than Google Scholar, both of the web-based scholarly databases, one of the two ASNs, and one of the two pirate sites. In terms of full-text coverage, the best multidisciplinary options are the ILS, Google Scholar, and the two pirate sites. Although several of the alternative discovery/access mechanisms are deficient in terms of their user interfaces, search capabilities, and metadata, they nonetheless provide excellent bibliographic and full-text coverage of the scholarly literature. In contrast, many single-subject library databases provide very incomplete coverage of their own subject areas. These findings have implications for scholars and students as well as system-wide implications for the use, development, and evaluation of information resources.