II. COLLECTION ASSESSMENT

A significant portion of the planning grant focused on understanding the development and use of the Tri-Colleges’ collections and their impact on the growth and space planning needs of the three libraries, both individually and as a consortium.

Data-Gathering Process

A Statistics Task Force was designated to gather data on the collections: their size, strength, duplication, growth rate, and use.² Members of this task force were Scott Silverman, Norm Medeiros, Barbara Weir, and Linda Bills.

As a preliminary step, the colleges worked with Electronic Scriptorium to identify and merge remaining duplicate bibliographic records in the database. This process eliminated about 60,000 duplicates that had survived the original record merger, done in 1990. Additional database massaging was done to facilitate the creation of reports based on call numbers.

The task force consulted bibliographers and determined that statistics should be gathered for 215 subject areas.³ Their goal was to obtain as much information as possible to cover four areas of inquiry:

1. Collection size, growth, and duplication (quantity)
2. Collection strength (quality)
3. Collection use
4. Patron needs

The following statistics were accordingly gathered. Unless otherwise stated, statistics came from report functions in Tripod. Tripod was fully implemented in 1991, and all such statistics date from that time or later.

Collection Size, Growth, and Duplication.
Number of monographic titles and volumes, and annual growth from 1991 to the present
Monographic title and volume overlap among the libraries

Collection Strength⁴
Number of foreign language titles
Collection age by publication year
Interlibrary loan lending activity

Collection Use
Number of circulation transactions
Circulation rates (e.g., number of volumes with 0, 1, 2-5 or 6+ circulations) in 11 years
Circulation rates to faculty
Circulation rates to non-Tri-College libraries (ILL loans)
Circulation distribution to faculty and students based on department or major
Cross-library borrowing within the Tri-Colleges

Patron Needs (non-Tripod)
Analysis of courses offered in the last four semesters
Borrowing from outside of the three colleges (that is, ILL borrows)

Data were prepared for the bibliographers in a FileMaker program created by Linda Bills. Organized by subject area, the program displays both summary and detailed data for each of the criteria listed above. It also produces comparative reports specific to each type of data, for instance, a report listing each of the 214 subject areas showing the number and percent of uncirculated titles in each college and for the consortium as a whole. The comparative reports are intended to help bibliographers see patterns of growth and use and quickly spot exceptional cases. Breakdowns are available for the consortium as a whole and for each library. Examples of the reports are provided in Appendix 3 (http://www.brynmawr.edu/consortium/MellonPlanningGrant).

As the recommendations of this study are implemented, feedback from the subject specialists will aid in determining which data are useful, what additional data are needed, and how the data should be presented. The aim is to develop an ongoing data collection and analysis tool.

Data-Gathering Results

The results reported here reflect the broadest perspective on the data. The three discipline divisions cover roughly the following Library of Congress classes: Humanities A-BD, BH-DU, E-F, M-PZ, TR, and Y-Z; Social Science BF, DX, G-GC, and GN-LG; and Science GE, Q-TP, TS-TX, and U-V.

Collection Size and Growth

Discipline size and growth patterns reflect the stable distribution of monographs in the disciplines, with the preponderance of books in the humanities. Thirty percent of titles and 32 percent of volumes currently in the collections were added in the last 10 years (Table 1).

Table 1. Subject Distribution of the Tri-College Collections
Discipline	1991 # Titles	%	1991 # Volumes	%	2001 # Titles	%	2001 # Volumes	%
Humanities	494,071		714,620		628,492		928,329
Social Science	153,973		210,984		213,606		295,601
Science	90,367		114,819		116,390		153,822
Total			1,040,423		958,488		1,377,752
Percentage Purchased 1991?2001

Collection Overlap

“Collection overlap” refers to the total of all titles and volumes that are held at more than one campus. “Overlap” is not the same as “duplication,” which the Planning Group defined as copies of a title beyond the first one located on the same campus.

Monographic overlap rates were tracked both by the number of titles held in more than one library and by the number of volumes represented by that overlap (Table 2). The number of volumes was more than twice the number of titles, reflecting not only multivolume sets but also instances of “triplication.” Rates were measured against the total collection. The data suggest that the introduction of the shared catalog in 1991 led to reduced duplication in all disciplines.

Serial subscription overlap rates vary by library (Table 3). There are 7,259 current print serial subscriptions⁵ on the three campuses. These represent 5,216 separate titles, and 2,026 second or third subscriptions, with 530 titles subscribed to on all three campuses.

Electronic serials overlap extensively with print journals. Purchasing decisions for electronic journals and journal collections have been made primarily to provide access to titles known to be used in print. Collectively, the Tri-College libraries have 5,216 print journal titles and 2,200 e-journal subscriptions, not including full-text coverage in aggregator databases such as Lexis-Nexis. The total number of journal titles available in print or electronic form has increased by 10 percent, or 500 titles, since the Tri-Colleges began jointly licensing content for the consortia member libraries. Seventy-three percent of the e-journals, compared with 50 percent of print journals, are available in all three schools.

The growth of electronic resources in the colleges is illustrated in Table 4.

BMC= Bryn Mawr College; HC=Haverford College; SC=Swarthmore College. E-Journals = single title purchases; EJ Coll. = Collections of e-journals such as JSTOR, Ideal; Databases = A&I or similar resources; DB w FT = A&I databases with some full text such as Lexis-Nexis or Expanded Academic.

Collection Age

Thirty-two percent of the circulating monographs in the collection were published before 1950, and 45 percent were published between 1950 and 1990. The remaining 20 percent were published since 1990 (Table 5).

Interlibrary Lending and Borrowing Activity

ILL activity may be an indirect measure of collection strength. Data indicate that the Tri-Colleges loan nearly twice as many items as they borrow. Further data gathering on ILL could indicate the value that the research community places on specific parts of the collections. A snapshot of ILL patterns showed that approximately 50 percent of Tri-College loans were in the humanities, 40 percent in the social sciences, and 10 percent in the sciences.

Circulation

Examinations of usage levels in the consortium revealed that 57 percent of the 1.39 million volumes in the circulating collection had not been borrowed since Tripod was implemented in 1991 (Table 6). Approximately 175,000 of these volumes with no circulations are overlaps (held by more than one library).

A further examination of circulation levels for duplicated items shows that the percentage of these items with zero circulations is slightly lower than that for volumes as a whole, indicating that some of the overlap is warranted by higher usage (Table 7). However, it is still clear that over half the overlap volumes have little or no current use.

An analysis by publication date for volumes in all disciplines shows a direct correlation between circulation activity and the age of an item (Table 8).

Cross-Library Borrowing

The rate of cross-library borrowing within the Tri-Colleges has been tracked since the introduction of the shared online system. Table 9 indicates that, after an initial jump, cross-campus borrowing leveled off for many years and has recently begun to rise again. The higher levels for Bryn Mawr and Haverford are largely due to the existence of Bi-College programs and the Bi-College approval plan.

We undertook a more detailed analysis of cross-library borrowing by faculty departments and student major. In the process, we uncovered data-tracking inconsistencies that, although now corrected, meant that faculty data could not be gathered retrospectively. Student data were available; Tables 10 and 11 show the type of information obtained.

 

Patron Needs Based on Curriculum

The Planning Group created a system to categorize the colleges’ courses and enrollments into the same subject divisions that were applied to the collections and to rate courses in terms of degree of dependence on library resources. Although the measures of need obtained seem useful, assigning the subjects proved difficult and has not yet been completed. Further work is needed to determine whether this measure can be efficiently applied to patron needs analysis.

Borrowing from Beyond the Tri-Colleges

Another potential measure of unmet patron needs would be interlibrary borrowing activity from outside the three colleges. Although we have rough numbers of requests, subject classification information is not available for PALCI and difficult to compile from Online Computer Library Center (OCLC) requests.

The interlibrary borrowing patterns of the three schools reveal an emphasis on books (60 percent books to 40 percent journals). Half of the books borrowed came from PALCI, a statewide system with patron-initiated borrowing and fast response time. Currently PALCI handles only monographs; its introduction has increased the interlibrary activity for books. Nearly half (47 percent) of interlibrary borrowing from outside the Tri-Colleges was done by undergraduates.

Key Findings

• Approximately 40 percent of the circulating titles in the three college libraries have copies in another library.
• Fifty-seven percent of the circulating volumes have not circulated in the last 11 years; an additional 17 percent have circulated only once. Older materials circulate less frequently than more recent materials do. Fifty percent of overlapped titles have had no circulation in 11 years.
• Overlap in the current approval plans between Swarthmore and the Bryn Mawr/Haverford shared plan is 80 percent. Overlap within the Bi-College plan itself is 15 percent.
• Patrons are comfortable borrowing materials from the other libraries. Cross-library borrowing rates are steady or rising.
• Humanities materials account for 67 percent of the monographic collection. Social science materials account for 21 percent and have the highest overlap rate. Science materials account for 12 percent and have the lowest overlap rate.

Data Collection Issues

Data collection accuracy and usefulness could be improved in several ways, a few of which are not under the Tri-Colleges’ direct control. For example:

• An ongoing data collection system should be instituted, particularly for “snapshot” statistics that are not available in the historical system reports.
• The bibliographers should update the current subject categories, which now number 215, to bring them in line with collection concerns.
• Swarthmore and Haverford currently do rolling inventories; the Bryn Mawr collection was last inventoried in 1990. Coordinating inventory schedules with collection evaluation schedules would be valuable.
• Each library has a different program to capture in-library and noncirculating use. In-library use in particular is of concern to the faculty, and tools for measuring this would be helpful. Likewise, the statistics programs used for this study could not count reserve use.
• A few collections restrict circulation to their own campus; use statistics for these materials will not be a true reflection of demand.
• Interlibrary loans from outside the Tri-Colleges could be more closely analyzed if more data were available, especially from PALCI.
• A better way needs to be devised to analyze patron needs as reflected in courses and enrollment.

---

FOOTNOTES

²The work done under the planning grant followed well-established practices and methods such as those outlined in the Guide to the Evaluation of Library Collections (Subcommittee on Guidelines for Collection Development, Collection Management and Development Committee, Resources Section, Resources and Technical Services Division, American Library Association (ALA). 1989. Chicago: American Library Association); and in the Guide to Review of Library Collections: Preservation, Storage, and Withdrawal (Lambert, Dennis K., et al., compilers. 2nd ed. 2002. Chicago, Ill., and Lanham, Md.: Association for Library Collections and Technical Services, ALA, in cooperation with Scarecrow Press). Studies of collection overlap, use, and age and composition go back at least 100 years. Thomas E. Nisonger’s 1992 bibliographic study Collection Evaluation in Academic Libraries: A Literature Guide and Annotated Bibliography (Englewood, Colo.: Libraries Unlimited) and the sources it cites document this literature. Nisonger is updated by the more selective Collection Evaluation Techniques: A Short, Selective, Practical, Current, Annotated Bibliography 1990-1998 (Strohl, Bonnie, compiler. 1999. Collection Evaluation Techniques Committee, Collection Development and Evaluation Section, Reference and User Services Association. Chicago: ALA).

We chose to collect circulation figures in spite of the controversy surrounding them, and we will be looking for ways to use these numbers, along with other measures of the value of titles, in order to make judgments about weeding. Early results of conversations with faculty members showed that their views on the value of circulation figures for collection management align with the famously negative reaction of faculty at the University of Pittsburgh to their library’s 1978 research, as reported in the so-called Pittsburgh Study (Kent, Allen, et al. 1979. Use of Library Materials: The University of Pittsburgh Study. New York: Marcel Dekker). See Nisonger, op. cit., pp. 55-60, for citations to articles relevant to the controversy. Perhaps because our circulation counts included the entire collection, as opposed to only recently acquired materials, our overall noncirculation rate was much higher than Pittsburgh’s or those in other follow-up studies (for example, Hardesty, Larry L. 1981. Use of Library Materials at a Small Liberal Arts College. Library Research 3[Fall]: 261-82, and Hardesty, Larry L. 1988. Use of Library Materials at a Small Liberal Arts College: A Replication at Eckerd College of the 1978 University of Pittsburgh Study. Collection Management 10[3-4]: 61-80). Our rate of noncirculation for materials acquired in the last 10 years more closely aligns with Kent’s and Hardesty’s findings

Our motives and methods for counting titles by subject follow the tradition of the Association of Research Libraries’ North American Collections Inventory project (Farrell, David, and Jutta Reed-Scott. 1990. The North American Collections Inventory Project: Implications for the Future of Coordinated Management of Research Collections. Library Resources and Technical Services 33[January]: 15-28 and the National Shelflist Count studies of the 1970s and 1980s. See, for example, the last iteration of the count, Titles Classified by the Library of Congress Classification: National Shelflist Count, 1985. 1986. Chicago: Resources and Technical Services Division, ALA). Our integrated library system allowed us to count titles more precisely and by more parameters than the National Shelflist Count could.

Our figures also align with those of Hardesty’s overlap study (Hardesty, Larry L., and Collette Mak. 1994. Searching for the Holy Grail: A Core Collection for Undergraduate Libraries. Journal of Academic Librarianship 19 [January]: 362-71), which provides both a concise review of the overlap literature and reports results that reflect our own in terms of percentage of titles uniquely held.

³Only Library of Congress and Dewey print monographs were examined. Special collections and government documents were excluded.

⁴Collection strength was the most difficult measure to obtain. Subjective expert evaluation was not possible. At the time of the study, the OCLC Automated Collection Analysis Service (ACAS) reports were not affordable. With the recent reduction in ACAS prices, the colleges hope to revisit this option and to explore other methods for comparison with peer collections.

⁵“Serials” include print periodicals, annuals, monographic serials, and continuations. Excluded are government documents and online and microfilm subscriptions.