Modeling Journal Bibliometrics to Predict Downloads and Inform Purchase Decisions at University Research Libraries

University libraries provide access to thousands of online journals and other content, spending millions of dollars annually on these electronic resources. Provid-ing access to these online resources is costly, and it is difficult both to analyze the value of this content to the institution and to discern those journals that compara-tively provide more value. In this research, we examine 1,510 journals from a large research university library, representing more than 40 % of the university’s annual subscription cost for electronic resources at the time of the study. We utilize a web analytics approach for the creation of a linear regression model to predict usage among these journals. We categorize metrics into two classes: global (journal focused) and local (institution dependent). Using 275 journals for our training set, our analysis shows that a combination of global and local metrics creates the strongest model for predicting full-text downloads. Our linear regres-sion model has an accuracy of more than 80 % in pre-dicting downloads for the 1,235 journals in our test set. The implications of the findings are that university libraries that use local metrics have better insight into the value of a journal and therefore more efficient cost content management