How Much Space Does a Library Need? Justifying Collections Space in an Electronic Age

Nancy J. Butkovich
Head, Physical and Mathematical Sciences Library
The Pennsylvania State University
University Park, Pennsylvania
njb2@psu.edu

Copyright 2010, Nancy J. Butkovich. Used with permission.

---

Abstract

In 2002, plans to merge Penn State's Physical Sciences Library and Mathematics Library provoked a controversy in the Eberly College of Science over the size of the library needed to support its departments. The College contended that a physical collection no more than 5 years old was adequate. A study of astronomy, chemistry, mathematics, physics, and statistics faculty publications showed that a much older collection was required to include 90% of their cited references. How much older varied by discipline. These data were then used to determine collection space allocations in the new library. A follow-up study in 2007 found that the patterns were generally still valid. Although the results are specific to Penn State, the data may be similar to other U.S. institutions with similar degree programs. The method is easily adaptable to other institutions.

Background

Prior to 2003, Penn State's University Park campus had separate mathematics and the physical sciences libraries, which served the Eberly College of Science (ECOS). Physical Sciences included chemistry, physics, and astronomy, while the Mathematics Library served mathematics, computer science, and statistics. Both libraries had zero-growth collections with 50-60% of each collection stored off-site.

In 2002 structural problems in the building housing the Mathematics Library resulted in reduced occupancy limits and additional materials moving off-site. With 75% of the math collection off-site, the library could no longer meet users' needs, so the University Libraries decided to merge the two facilities. Because the Physical Sciences Library was already at zero-growth, that facility would need significant renovations in order to accommodate the mathematics and statistics collections. The computer science collection was moved to another library.

Administrators in the Eberly College of Science (ECOS) and the Departments of Chemistry, Physics, and Astronomy and Astrophysics objected. Although Chemistry was moving to a new building, the extra space needed for the combined library was already earmarked for use by the other two departments. The ECOS administration believed that, with all the resources needed by faculty being available online, a large library facility was not necessary and a five-year collection was adequate. The only dissent from within ECOS came from the Mathematics Department faculty, who were strong advocates of not only maintaining but also increasing the size of on-site collections.

Problem Statement

The ECOS position is not unique to Penn State. Many articles have discussed the growth and development of electronic resources or the predicted demise of the "print" library (for example, Butler 1999; Kurtz, et al. 2006; Mullin 2008). An administrator at the California Institute of Technology claimed that "everything ... was available on the web, and that the main library ... was just a warehouse for books and serials that nobody used anymore" (O'Donnell 2001). In response Caltech librarians conducted a citation study of faculty publications and determined that less than 40% of publications cited by Caltech faculty in general, and about 24% in the geosciences, were available electronically.

At Penn State format was not an issue; if electronic access was available, the print equivalents were not retained on-site. A concern was whether material over five years old needed to be retained on-site. To answer this, citation analysis studies of the tenured/tenure-track faculty in the Departments of Astronomy and Astrophysics, Chemistry, Mathematics, Physics, and Statistics were conducted. Although citation data are "dirty" (Buchanan 2006), they provide a measure of use that is familiar to faculty.

Methods

In 1965 Bourne posed the question, "What does the library have to do to satisfy 90% of this population's needs?" Many studies have used some variation of this concept, such as an analysis of photocopy requests (Basile and Smith 1970), collection use data (Smethurst 1977), document delivery requests (Chrzastowski and Anthes 1995), and current awareness search results (Youngen 1999). Ackerson (2001) examined the physical chemistry literature and concluded that while median ages of cited physics and chemistry references were relatively recent (generally about six years old), a much larger date range was needed to fill 90% of the cited references.

A study conducted at Caltech (O'Donnell 2001) was used as the model for gathering data and included all assistant, associate, and full professors identified in the faculty directories on the Astronomy and Astrophysics, Chemistry, Mathematics, Physics, and Statistics web sites. Because the ECOS administration was interested only in what its faculty published, citation data for faculty from other departments and colleges or resource use by undergraduates, graduate students, or visitors were not included in the study. This unpublished study was repeated in 2007 to determine whether the results were consistent with the original data.

Table 1 summarizes the demographics of the departments during the two sample years. The Physics and Astronomy and Astrophysics departments added faculty with corresponding increases in the number of published articles and cited references; the other departments were stable.

Table 1. Demographics of faculty and their publications.

	Departments
	Astro		Chem		Math		Phys		Stat
	2002	2007	2002	2007	2002	2007	2002	2007	2002	2007
No. faculty	19	24	31	34	56	55	39	52	24	24
No. articles	57	71	93	89	162	142	117	153	69	60
No. cited refs.	1971	3407	3556	3715	3124	3432	3559	5654	1807	1667

Each faculty member's name was searched in Web of Science (Science Citation Index section) and if necessary, the Social Science Citation Index section), and three publications, ideally the most recent ones, were identified. Personal web sites for a small number of new faculty were also examined because the affiliation given in Web of Science indicated their previous institutions. Papers jointly authored by faculty within the same department were included under one of the authors and different papers were selected for the other. Dual appointment faculty were counted in both departments; since each department was examined separately, this did not bias the results.

Once the articles were identified, the cited references were exported into Excel. Publication dates were converted to ages (2002 = 0 years old, 2001 = 1 year old, etc. for the first and 2007 = 0 years old, 2006 = 1 year old, etc. for the second) to facilitate comparison between the datasets. Cumulative percentages of cited references in each date range were totaled until targets representing 70%, 80%, and 90% of the total number of cited references for each department were reached.

Discussion

Table 2 shows the percentage of cited reference and show that the data are highly skewed toward the newest material. Mathematics and Statistics faculty cited references two years old or less at a much lower rate than other faculty; however, their peak age was the same as the others: 3-7 years. This may be due to differences in journal publishing frequency, assuming that mathematics has a higher percentages of journals published monthly or quarterly and has longer citation time lags than disciplines with many weekly or bi-weekly journals. The latter would have more articles available to be cited at an earlier date that would the former. Mathematics also cited older material more often in both sample years, and Statistics showed a similar trend in its 2007 data. This is possibly due to improved collection access resulting from having more of the mathematics collection available on-site.

Table 2. Number of cited references as a percentage, by department and age. (Columns may not total 100% due to round-off).

Age Range of Cited References	Departments
	Astro		Chem		Math		Phys		Stat
	2002	2007	2002	2007	2002	2007	2002	2007	2002	2007
0-2 yrs	26.8	26.1	16.7	16.5	5.2	4.3	16.6	14.7	3.0	4.3
3-7 yrs	39.6	37.0	40.6	36.4	25.7	21.3	36.9	34.8	27.6	32.4
8-12 yrs	14.7	15.8	19.2	20.5	19.1	18.3	16.2	18.3	22.5	22.0
13-17 yrs	6.8	8.4	9.7	11.1	13.9	12.5	10.0	11.4	16.8	12.5
18-22 yrs	4.0	4.7	4.4	5.6	9.8	9.2	5.2	4.6	9.6	6.6
23-27 yrs	2.6	2.6	3.1	3.0	7.6	8.1	3.8	3.3	7.0	6.5
28-32 yrs	2.1	1.3	1.9	1.9	4.3	6.6	2.7	2.5	4.9	5.0
33-37 yrs	0.6	0.9	1.3	1.5	3.9	5.1	1.6	2.0	2.4	3.2
38-42 yrs	0.4	0.4	0.6	1.1	2.2	3.6	1.2	1.2	1.8	2.0
Over 42 yrs	0.8	0.8	1.2	1.6	4.7	6.4	2.2	2.7	3.7	4.1
No date given	1.6	2.0	1.2	0.8	3.6	4.5	3.6	4.7	0.7	1.3

The purpose was to determine how far back in time one needed to go to include 70%, 80%, or 90% of the references cited by the faculty in each department during the two surveys. Astronomy reached the target percentages first at all three levels, followed by Chemistry, Physics, Statistics, and Mathematics (Table 3).

Table 3. Ages at which the target percentages of cited references were reached.

Target Percentage of Cited References	Departments
	Astro		Chem		Math		Phys		Stat
	2002	2007	2002	2007	2002	2007	2002	2007	2002	2007
70%	8 yrs	9 yrs	10 yrs	11 yrs	20 yrs	25 yrs	13 yrs	13 yrs	18 yrs	17 yrs
80%	12 yrs	13 yrs	14 yrs	15 yrs	26 yrs	32 yrs	18 yrs	19 yrs	23 yrs	24 yrs
90%	20 yrs	20 yrs	21 yrs	22 yrs	39 yrs	45 yrs	29 yrs	33 yrs	31 yrs	33 yrs

Ages at the target levels were generally consistent between the surveys, although two anomalies were noted. Mathematics showed a five- to six-year increase at each target. This is likely due to improved on-site access to older material. Table 2 seems to verify this belief, showing a higher percentage of older material cited in 2007 than in 2002. On the other hand the Physics "bump" occurred at only the 90% level and may not be significant. In the case of Chemistry, these values are generally comparable to those noted by Ortega (2008) in her study of University of Oklahoma faculty. However, the 90% numbers are considerably smaller than those noted by Ackerson (2001). This difference is likely caused by the sampling methods used; she examined citations in 22 physical chemistry journals, while the present study sampled publications of all faculty in the Chemistry Department. Perhaps if the results of this study had been limited to just the physical chemists, or if she had looked at a broader range of titles, the data would have correlated better.

Table 4. Median ages and percentage of total citations that were 5 years or less in age.

	Departments
	Astro		Chem		Math		Phys		Stat
	2002	2007	2002	2007	2002	2007	2002	2007	2002	2007
Median age	5 yrs	5 yrs	6 yrs	8 yrs	12 yrs	14 yrs	6 yrs	7 yrs	12 yrs	10 yrs
Percent at 5 years	50%	50%	45%	40%	23%	21%	47%	42%	18%	25%

The position taken by the ECOS stated that a five-year collection would be adequate for the needs of the faculty. However, the citation data (Table 4) showed that a five-year collection would include less than 50% of the total number of cited references in every department except Astronomy and Astrophysics, which had a median percentage at the five-year mark. For mathematics faculty a five-year collection would have only included 23% of what they cited in 2002 and 21% of their total citations in 2007. The other departments fell between these extremes. The median ages for Penn State chemistry faculty in 2002 and 2007 are comparable to those obtained by Ortega (2008). Statistics faculty produced the only anomaly--they actually experienced a decline in median age and a higher percentage of references cited that were 5 years or less in age between 2002 and 2007. Ultimately these results helped in justifying the collections space needed to serve the user community, and they provided the core of information used to allocate the available collections space between the disciplines represented by the five departments.

The final space allocations were not a perfect match for the data. Citation data reflects what people choose to cite and not necessarily what they actually look at, so we used internal reshelving and document delivery data to identify other older medium-to-high use physical sciences materials that would otherwise not be retained. Since many of our users were faculty from other departments and colleges or were undergraduate or graduate students and post-docs, this helped counterbalance the bias caused by restricting the study to the faculty in the five departments. In addition, many current print journals were switched to electronic only subscriptions and corresponding print volume were transferred off-site or withdrawn. This allowed some unique print works to be returned to the on-site collection, which particularly benefited the Mathematics and Statistics departments.

Local politics also played a role in the final collections space allocations. The data (Table 3) showed that Astronomy and Astrophysics and Chemistry only needed about 20 years of on-site collections, while Physics and Statistics required about 30 years, and Mathematics nearly 40 years. In the final distribution, Physics was limited to shelf space for a 20-year collection, and Statistics was increased to a 40-year collection. However, the usage data obtained from other sources were used to justify keeping some additional materials on site for physics.

Conclusion

Ideally, data from all user populations should be considered in determining collection use; however, local politics required the inclusion of only faculty from the five ECOS departments. This analysis was relatively quick and easy to do and could be easily implemented at other institutions. The results show a significant difference in the ages at which the citations from each department met the target amounts, with Mathematics faculty being much more likely to cite older material than any of the others. Recognizing that citation data have their limits, other data sources were used to augment citation data, and the data were used to determine collections space allocations in the new library. A follow-up study five years later confirmed the validity of the data and suggested trends for Physics and Mathematics that may need additional follow-ups. The results show that these data can be used for collection management purposes. Four of the five departments showed a fairly constant level of use at the different age ranges, while mathematicians demonstrated a much higher level of use of older material when on-site availability of that part of the collection was increased.

Note

URLs for faculty lists have changed over the years, but the current ones can be located on the departmental web sites: Astronomy and Astrophysics Department http://www.astro.psu.edu/; Chemistry Department http://www.chem.psu.edu/; Mathematics Department http://www.math.psu.edu/; Physics Department http://www.phys.psu.edu/; Statistics Department http://www.stat.psu.edu/

Acknowledgments

The author wishes to thank Sharon Pagano, Douglas McAnallen, and Jon Wallace for their assistance with searching faculty names in Web of Science in the original survey, and Sally Kalin and Linda Musser for their thoughtful comments and suggestions regarding the manuscript.

References

Ackerson, L.G. 2001. Is age an appropriate criterion for moving journals to storage? Collection Management 26(3): 63-76.

Basile, V.A. & Smith, R.W. 1970. Evolving the 90% pharmaceutical library. Special Libraries 61: 81-86 (February).

Bourne, C.P. 1965. Some user requirements stated quantitatively in terms of the 90 percent library. In: Kent, A. and Taulbee, O.E., editors. Electronic Information Handling Washington, DC: Spartan Books. pp. 93-110.

Buchanan, R.A. Accuracy of cited references: the role of citation databases. College & Research Libraries 67(4): 292-303.

Butler, D. 1999. The writing is on the web for science journals in print. Nature 397: 195-200.

Chrzastowski, T.E. & Anthes, M.A. 1995. Seeking the 99% chemistry library: extending the serial collection through the use of decentralized document delivery. Library Acquisitions: Practice & Theory 19(2): 141-152.

Kurtz, M.J., et al. 2006. The future of technical libraries. [Internet]. [Accessed September 28, 2006]. Paper no.: arXiv:astro-ph/060979v1 Available from: http://arXiv.org.

Mullin, R. 2008. Vanishing books. Chemical and Engineering News 86 (17): 30-31.

O'Donnell, J. 2001. Analysis of faculty citation behavior in the electronic age: a study of one institution's recent publications. Proceedings of the Geoscience Information Society 32: 69-74.

Ortega, L. 2008. Age of references in chemistry articles: a study of local authors' publications from selected years, 1975-2005. Science & Technology Libraries 28(3): 209-246.

Smethurst, J.M. 1977. A practical case study. In: Blackwood, J.W., editor. The Future of Library Collections [Loughborough, U.K.]: Library Management Research Unit, University of Technology, p 22-31.

Youngen, G.K. 1999. Using current awareness search results to measure a journal collection's relevancy. Library Collections, Acquisitions, & Technical Services 23(2): 141-148.

Previous

Contents

Next