Abstract
In 2015, the University of California, Berkeley, launched EBSCO Discovery Service (EDS), a web-scale discovery tool, with a goal of improving visibility and usage of collections. This study applies linear regression analysis to usage data for ebooks, ejournals, and abstracts and indexing (A&I) databases before and after implementation of EDS in order to identify correlations between the discovery layer and usage of library electronic resources across platforms. Our findings diverge from conclusions drawn in the previous literature that indicates that resource use generally increases after a discovery tool is implemented. We examine data from a longer period of time than the previous literature had, looking for statistically significant changes in resource use. The discovery layer at UC Berkeley did not lead to equal increases across platforms, but rather to a complex array of increases and decreases in use according to a variety of factors.
Acknowledgments
The authors would like to thank the following individuals and groups at UC Berkeley who helped us to collect and decipher usage statistics in a complex environment: Jason Dezember, Lynne Grigsby, and Sherry Lochhaas from the Library; Lei Zhang, Fan Dong, and the Department of Statistics Statistical Consulting Service.
Notes
1 See, for example, Williams and Foster, Citation2011; Asher, Duke, and Wilson, Citation2013; Bonner and Williams, Citation2016.
2 The precise level of inclusion of content in the EDS indexes varies significantly based on each platform or publisher’s inclusion in EBSCO’s “Complementary,” “Supplemental,” or other form of index, as well as the institution’s subscription status to a particular resource. See EBSCO Help for more information: https://help.ebsco.com/interfaces/EBSCO_Discovery_Service/EDS_Admin_Guide/content_included_in_EDS_profile
3 The steep decrease in use of Highwire was likely also due to a platform shift effective December 2016, when Sage ejournals moved to their own platform. While it was not feasible to track publication and publisher changes across platforms, it may be the case that major changes such as the shift from Highwire to Sage complicate the story that our regression analysis tells.
4 To view the UC Berkeley Library website as of September 21, 2018, see: https://web.archive.org/web/20180921070037/http://lib.berkeley.edu