Purpose & Goals

Marquette University’s Raynor Library completed a large-scale weeding project in 2023 to accommodate a renovation that significantly reduced available collection space. With shrinking stacks and more weeding on the horizon, an opportunity arose to refine the approval plan with the goal of reducing the ongoing print footprint of the collection. This assessment project targeted various profiles of the approval plan, which is how the library receives most of its print, to identify areas that can be switched from automatic print shipments (or autoshipments) to slips.

Design & Methodology

The approach for this project was a two-stage review of approval profiles, using circulation data at one stage, and citation data at the other. GreenGlass data was used to pull circulation rates for every autoship profile of the approval plan, and a heat map of the data established a baseline for performance evaluation, using the principle of relative efficiency. For example, 39% of the entire collection has never circulated overall, so if 22% of a smaller area of the collection has never circulated, it is deemed a strong performer. For the second stage, an R-based text mining program was used to simplify the gathering of citation information. A list of print titles sent on approval over an eight-year span was run against a corpus of recently published discipline-appropriate faculty and graduate student publications from Marquette’s institutional repository. Citation counts were then pulled and tallied after data cleanup and the elimination of false positives. The citation counts and circulation rates generated by this process were used to compare subdisciplines within an autoship profile, along with the size and cost of each subdiscipline’s annual autoshipments. For example, since approval profiles are based on Library of Congress Classification (LCC) ranges, when researchers looked at the JC’s, they investigated autoshipped print titles acquired over the life of the GOBI approval plan in each subdiscipline of the JC’s, comparing JC 1-50 (Theories of the state) to JC571-605 (Rights of the individual), and every subdiscipline in between. The data for each range was scrutinized for citation counts and circulation rates, and problem areas were identified using both sets of metrics. The results were further scrutinized for limitations, including a survey of faculty and department pages to see if research on subjects relevant to the profiles under consideration were not uploaded to the repository.

Findings

The researchers initially targeted a print-preferred profile of the collection that delivered the highest volume of titles, the B’s (Philosophy). They identified several subdisciplines with low circulation and citation rates. A low circulation rate was gauged as 50% or more of a collection area containing zero checkouts, and a low citation rate was gauged as one or fewer unique cited titles per area, or subdiscipline. Due to a much smaller amount of citation data, however, performance was mostly gauged by the relative counts in each subdiscipline, as a baseline metric like the 39% figure in circulation was not established in citation counts. Researchers also assessed e-preferred areas, as Raynor Library still receives a large volume of print titles that meet subject-based criteria, but are not published electronically, in its e-preferred approval profiles. The citation rates for print titles in these e-preferred areas performed much worse than their print-preferred counterparts.

Action & Impact

The best approval areas to target for revision in this assessment project were profiles with low circulation and citation counts, and a high volume of print autoshipments, since revising these would have the greatest impact on the collection footprint. Additionally, given the results for e-preferred areas, one sustainable measure for the collection footprint may be to slip all print in these e-preferred areas, while maintaining print autoshipments in the few high-performing subdisciplines. All of these findings will be shared with the appropriate subject specialists to generate a conversation about possible approval profile revisions.

Practical Implications & Value

This assessment project’s main impact is the use of a process that lowers the largest barrier to citation analysis: time. Citation counting for bibliometric analysis has traditionally been carried out manually, and this R-based text mining tool allows for relatively efficient, high-volume citation analysis. While researchers in this context used it to assess print autoshipments for collection space purposes, one can imagine this approach being used in several other collection assessment contexts where citation data could augment other forms of usage data to help inform decision-making processes.