Practice Collection Assessment and the Library Liaison Program A Practical Focus

In the fall of 2013, the George T. Harrell Health Sciences Library (HHSL) undertook a collection assessment and development project to determine journal and monograph holdings across the Penn State University Libraries in subject areas related to librarian liaison assignments in the health sciences at Penn State Hershey. This project had several aims: a) to assist in collection development activities by showing current journal and monograph holdings in relation to liaison areas in a way related to the strategic plans of the HHSL, the medical center, and University Libraries; b) to inform library users of holdings and improve access; and c) to assist liaison librarians in acquiring a more complete understanding of core resources in their liaison areas. This article presents the process involved in the assessment and development project, connections with relevant areas of the strategic plans, and areas for further improvement.


Introduction
The George T. Harrell Health Sciences Library (HHSL) at the Penn State College of Medicine is the health sciences library for the Pennsylvania State University Libraries.The college has on average 600 medical students in the program at any given time in addition to graduate students, residents, faculty, clinical personnel, and other staff for a campus size of over 10,000.Library faculty serve in many capacities to support both the HHSL and the University Libraries as conduit for communication to and from the approximately 30 departments served (www.libraries.psu.edu/psul/hershey/services/liaisons.html).As part of the widely distributed system of University Libraries, which was described by Crawford & German (2013) as "one library geographically dispersed," the HHSL does not collect all of the possibly relevant materials in the health sciences, particularly in the basic sciences.As a result, print materials of interest to the Penn State Hershey campus may be spread across the state, either in other regional campus libraries or at the central Pattee and Paterno Libraries at University Park.The collections are highly mobile with easy access via the "I Want It" delivery service to any Penn State user at any campus, which reduces the need to duplicate print items.The vast majority of electronic resources are licensed for access by all Penn State users, regardless of location, with a few narrowly clinical titles solely licensed to the Hershey campus.In order to determine if (and demonstrate that) the collections of the HHSL and University Libraries were adequately serving our users' needs, it was necessary to analyze journal and monograph holdings in the health sciences, which had not been done before.
Criteria and methods for this analysis had to be developed, as well as determining which systems and tools were to be used to collect and analyze the data.The University Libraries had a number of management tools available, such as SIRSI ILS, Serials Solutions, and subscription data sources like Ulrich's and Web of Science (Thomson Reuters), which included bibliometric data in the Journal Citation Reports module.Vendor databases like Yankee Book Peddler (YBP) were also used to obtain rough pricing data after purchase priorities were identified.The Assistant Librarian for Collection Development/Digital Resources Management had overall responsibility for the project, including the development of analysis criteria and supporting tools.The remainder of the HHSL liaison librarians were responsible for collecting data in their respective subject areas as well as analyzing and ranking these areas in relation to the strategic priorities of the institution.
The result of this project was a ranking of lacking journal and monograph titles identified as strategically important to the HHSL and Penn State Hershey and an estimate of the expenditure necessary to achieve full coverage in each subject area.

Literature Review
The literature on the role of the liaison librarian in collection assessment and development is quite extensive, coming from a wide variety of perspectives and library environments.existing subject area coverage.Among the relevant citations, several made a connection between the library's collection activities and larger strategic imperatives, particularly Franklin (2009) who emphasized the role of an "assessment culture" in demonstrating value, and Franklin again in 2012, discussing the importance of "demonstrat[ing] the value of the academic research library in providing quality services that advance the institutional mission" (p.94).A number of citations linked budget allocation formulas to strategic goals (e.g., Cochrane, 1995;Kalyan, 2003).Discussion of reconfiguring budget allocations is beyond the scope of the present project.However, the authors present an intriguing possibility to use the data derived from the project as the basis for developing a formula for future allocations.This formula would require additional research as well as additional longitudinal data.

Methodology
A two-phased assessment and development project was initiated in August 2013.Phase 1 consisted of a) developing a set of tools to inform collection assessment at the HHSL, and b) completing a collection assessment related to current liaison coverage by the end of December 2013.
At the close of this assessment, Phase 2 involved taking the Phase 1 data and developing a model to prioritize purchases from the materials identified as not being held by the HHSL or the University Libraries.Given budgetary constraints, it was understood that not all of these purchases would be possible to acquire, nor those possible achieved in the current fiscal year.A target cap for both journals and monographs was set at $60,000, so there would be a cutoff point even with materials that were ranked as priorities.Funding for initial purchases of monographs and ongoing subscriptions for journals was factored into the regular budgeting process, with the bulk funded through using carryover funds from the previous fiscal year.This model was implemented by the HHSL liaison librarians in order to a) rank priorities and identify specific monograph and journal titles for purchase.These would be purchased in electronic format, if possible, and licensed for all Penn State users in keeping with the  Monographic selection was less difficult, since the most clearly recognized core title list is Doody's Core Titles (DCT), which supplanted and was seen as "the likely heir to" the Brandon-Hill list in 2004 (Shedlock and Walton, 2006).
For journals, finding a method for ascertaining core titles was more problematic.There has been significant discussion in the literature concerning the evaluation of journals, including extensive analysis of a number of ranking To rely on such conflicting data is a discussion beyond the scope of this paper.For the purposes of this project, a factor had to be chosen.Buela-Casal and Zych (2012) cite a number of studies showing the popularity and durability of the IF, indicating that it is a well-known measure, highly promoted by journal publishers, and accepted (or at least tolerated) by researchers.A decision also had to be made as to the IF chosen: single-year or 5-year.It was decided that in order to account for single-year spikes in impact of any one journal title, journal holdings were to be analyzed using the 5-year IF.
For both monographic and journal data, it was clearly recognized that any attempt to assign subject categories to a specific liaison department was necessarily a fluid construct that allowed for a great deal of discretion by those doing the mapping.The initial data sets were significant in size: The 2013 DCT list contained some 2,400 titles with 117 subject categories, and the journal data contained over 8,000 titles in 226 subject categories.Often there was no one-toone relationship between a journal or monographic subject heading and a liaison department assignment (Figure 2).
For example, while Anesthesiology is consistent, there is no liaison assignment for Biochemistry (DCT) or Biochemistry and Molecular Biology (JCR).These would fall under the Basic Sciences liaison.It was also recognized that there would be overlap in these selections.For example, titles in Case Management in the DCT list might be selected as appropriate for both Nursing and Family & Community Medicine.
Thus, for the data collection phase of the project, data from DCT and JCR had to be mapped to the data collection templates, and the University Libraries' holdings had to be determined.To facilitate this mapping, master spreadsheets for journals and monographs were created.These spreadsheets included a complete DCT and JCR title list with title, control, and subject data.Another discretionary decision was the determination to analyze monographs and journals differently: Liaison librarians could choose appropriate subjects from the DCT list and assess all the titles in those subjects, while journal data was limited to the top 20 titles in any given JCR subject area.This construct was developed after some discussion (in no particular order of priority):  It was considered easier and more consistent to communicate to liaison departments that the library held x% of the top 20 journals in their subject areas rather than a series of widely variable numbers.


It facilitated coverage comparisons among departments for the purpose of prioritizing purchases.


It recognized that purchase models for journals and monographs were markedly different in terms of cost, in that more monographs could be purchased for a given dollar amount than journals.Thus, a wider choice of monographs was available to improve subject coverage in areas that might be lacking, and this diversity was an advantage and a reason to not limit as with journals.
 It recognized that many journals crossed a number of disciplines, whereas a monograph (unless it is a general textbook) is more narrowly targeted.Equally, the crossover would again lead to duplication in journal titles that would be addressed by de-duping the ranked lists at the close of the analysis.
The construct was an attempt to reduce the number of titles analyzed due to the volume of data (8,000 journals and 2,400 monographs) and number of subjects assigned (226 journal subjects and 117 monographic subjects) within the time frame of the project.
The liaison librarian's role was to select items from these lists according to the methodologies outlined above and analyze the library's holdings of journals and monographs.This process was the most labor-intensive part of the project.While those titles with exact ISBN or ISSN matches were obvious, those without control number matches had to be manually searched to verify holdings.The additional work involved in verifying titles without exact control number matches pointed to the need to develop tools to more fully automate the process, which could reduce the need for the artificial limitations that were placed on this phase of the project and allow for a more comprehensive (and more frequently recurring) assessment.
Data collection work began in October 2013.Selections by the reviewing liaison librarians resulted in 62 monographic subject areas and 52 journal subject areas to be analyzed for holdings.Several reviewers made significant changes to the reporting template, which necessitated additional labor in reformatting.There were also many variations in format of the data reported, which necessitated labor in normalizing responses.The major issue in assessing holdings of monographic titles was the source of the catalog record.If there was not an exact ISBN match, the librarian had to conduct a manual search to determine holdings of that title.A further complication was that the title data in the DCT list did not always match the cataloged title of the monograph (e.g., in the DCT list, Drug Eruption Reference Manual: DERM is listed as Litt's D.E.R.M.: Drug Eruptions and Reactions Manual for the earlier edition, which the University Libraries own).Not all reviewers exhibited the same level of thoroughness in searching for variations in title, ISBN mismatches, or title variations in earlier editions.To extend or repeat this analysis model, additional effort would be recommended to streamline this part of the process to improve the consistency of the data.
Journal holdings assessment was more straightforward as ISSN matching for journals was much more consistent than ISBN matching was for monographs.However, analysis of actual holdings required a librarian to make a judgment call to determine if there was current full access to a given title."Partial open access" did not count as a current subscription.The second phase of the project consisted of ranking priorities, identifying specific monograph and journal titles for purchase, and increasing the liaison librarians' awareness of institutional strategic priorities in relation to collections and collection development.It required developing a model that used strategic planning documents to rank potential purchases of journals and monographs.This ranking process was intended to diverge from the "wish list" model and eliminate favoritism or bias as much as possible from developing collection priorities.It was also intended to look at collection development in a more high-level fashion, reflecting the strategic direction of the institution rather than a single department, clinician, or researcher.
An initial meeting was held in January 2014 to determine the ranking process.Similar to the issues in the Phase 1 assessment (e.g., variations in the taxonomy between the JCR and DCT subject categories), rankings were grouped in like categories as closely aligned as possible (e.g., Pediatric Surgery and Pediatrics were grouped together, as were Geriatrics and Long-Term Care).These groups were at the discretion of the liaison librarian doing the rankings.For categories where the taxonomy was identical for both journals and monographs (e.g., Pathology or Medical Education), they were equally ranked.
The results of this process were a ranking of purchase priorities in subject areas of strategic importance to the HHSL and, by extension, the University Libraries as a whole, as well as an estimate of the collections expenditure necessary to achieve full coverage in these areas.Realizing that even this prioritized subset would require more funding than the $60,000 target, IF and DCT ranking scores were retained to allow for additional cutoff points within each subject set (i.e., only purchase monographs with a DCT score over 2.25 across the board).

Analysis of Ranking Recommendations
Data analysis began after collecting reports from the liaison librarians.The initial ranking was sorted by two factors: first, by the number of reviewers who chose a given subject ("weight") and, second, by the average of their rankings ("avg rank") (Figure 5).As with the initial analysis, there were subjects that were similar but not identical across the JCR and DCT data, and they were grouped together as much as possible in the ranking exercise.
The ranking results were then combined with the holdings and cost data from the previous report to produce cost estimates across subject categories.Not all categories had matches for both journals, and monographs or duplicates crossed other subject areas (Figure 6, sorted by subject area).Collection development is a classic example of a "moving target," in that during course of the assessment, holdings changed as journal and e-book packages were added and adjusted, and priority titles became available or changed format.Spanning budget years with purchases also caused changes in cost or edition for an identified item, which impacted the number of titles that could be acquired while still keeping within our target budget.The major impediment to keeping the process going is the lack of tools to reduce the labor involved in verifying holdings.Further development of these tools would make it possible to repeat the process more frequently and improve the library's responsiveness in collection development.As a result, the HHSL development team has been working on tools to improve and expedite this process and plans to repeat this project in late 2015.
HHSL collection development policy (bit.ly/1EqSpyc), and b) increase the liaison librarians' awareness of institutional strategic priorities in relation to collections and collection development.Pennsylvania Libraries: Research & Practice Collection Assessment and the Library Liaison Program palrap.orgPhase 1-Tools and AssessmentThe project began with developing a set of tools and resources to inform collection analysis and communication at the HHSL.These tools were created and maintained by the Assistant Librarian for Collection Development/Digital Resources Management.A wiki page for HHSL liaison librarians was constructed (Figure1), which included the major data sets to use in the analysis as well as a rolling ledger of internal and external collection requests with their current status.This ledger was an Excel spreadsheet that contained title, cost, and status information as requests were received, and it was updated as purchases or decisions were made in order to facilitate communication between liaison librarians and their departments.At the close of the assessment and development project, all resulting documents were posted to this wiki page for use in continuing liaison activities.The ledger has been maintained since the project's conclusion and is posted to the collection development wiki page on a monthly basis.

Figure 1 HHSL
Figure 1 HHSL collection development wiki Using a similar technique to that shown inBrennan (2011), which embeds call number or subject data in a URL to search the library catalog, the spreadsheets also contained links to allow one-click searching of both monograph titles in the library catalog by ISBN and journal titles in the Serials Solutions journals list by ISSN.These URL links sped up data collection for those titles that had exact control number matches.An instruction session was conducted for liaison librarians to demonstrate how to use the Excel filtering function to take data from the master lists and move it to their templates, so they could complete the holdings assessment.

Figure 2
Figure 2Sample comparison of liaison assignments to JCR and DCT subjects Figure 5Ranking sample Pennsylvania Libraries: Research & Practice Collection Assessment and the Library Liaison Program palrap.orgschemes such as the H-factor, Eigenfactor, and Journal Citation Reports (JCR) Impact Factor (IF).Key problems cited by Glänzel and Moed (2013) include the "tend[ency] to condense … statistics into one single indicator at the end, but … demand solutions for inference at the individual level at the same time" (p.387).
Each liaison librarian was given a duplicate-identified list of all of the monograph and journal subject categories used in the initial coverage analysis and tasked with the ranking of purchase priorities in terms of strategic importance to Penn State Hershey's services and programs.This ranking was to be derived from a number of factors, including