MSU Libraries have historically allocated a portion of the funding available for monographic purchases to the university’s academic departments. Selection of materials for each department’s funds was made by departmental faculty and the librarian assigned as liaison to the department. Until 1992, allocations were made based on historical spending patterns with occasional adjustments to support new programs and accreditation needs. In 1992, the library faculty, in consultation with the University Library Committee, decided to implement a fund allocation formula using eight variables: undergraduate credit hours, undergraduate majors, graduate credit hours, graduate majors, average cost of book in discipline, publishing output, relative importance of books and serials, and local use. Average book cost, a critical factor in the formula, would be determined by identifying subject areas pertinent to each department and using average prices from “North American Academic Books: Average Prices and Price Indexes” published in Bowker.²

Determining the departments’ average book costs in this manner was not without challenges. One of the most serious problems was that the subject breakdown used in Bowker did not reflect the curricular and research needs of a land grant institution like MSU. For example, the term “Engineering and technology” was listed with no subdivisions to distinguish price differences among MSU’s engineering departments. The problem was exacerbated by an increase in the number of interdisciplinary departments; for example, the Geosciences Department includes programs in geography, geology, meteorology, and geographic information systems.

Another problem was that data in the Bowker table represented hardcover, trade, and paperback books. Belanger noted that the data used in the calculation of the average price of North American academic books is derived from titles included in the approval plans of Blackwell’s Book Services and YBP, as well as books supplied through all order types from Baker and Taylor.³ The increasing number of paperback books included in approval plans and sold through firm orders deflate the average prices and price indices in a manner not indicative of purchasing patterns typical at the MSU Libraries. While a few departments prefer paperback whenever available, most prefer hardcover almost exclusively.

Another challenge was the preference by some departments to collect materials in nonbook formats and in non-English languages. For example, the Music Education Department is concentrating on expanding the libraries’ collection of music recordings on compact discs while the Philosophy and Religion Department requires numerous titles in German. Although Bowker published data on average costs of audiovisual materials, this information was omitted after the 46th (2001) edition.⁴ Foreign language titles represent a very small percentage of the titles used to compute the average prices and price indices in Bowker. Cost coverage reports from the Blackwell and YBP Web sites suggest that about 2 percent of the titles included in approval plans are published in languages other than English. The MSU Libraries had made no attempt to adjust the average prices based on selection of nonbook media or foreign language materials.

The University Library Committee’s periodic reviews of the formula had found no fault with the use of an average price variable, but the Libraries’ collection development unit grew concerned with the appropriateness of determining the average price data for each department from the Bowker information. In 2004, the collection development unit chose to use average prices derived from “Approval Program Coverage and Cost Study” (Blackwell study) compiled by Blackwell’s Book Services.⁵ This study details subject areas and list prices of monographs covered by Blackwell’s New Titles/Approval program for academic library collections. The collection development unit customized price information by matching each academic department’s approval slip profile with the pertinent subject disciplines, a time-consuming procedure. For the 2005/06 academic year, the collection development unit decided to explore options available for determining average prices using the most current data available from several sources.

Allocation formulas and their components have been widely discussed in library literature. Critics have pointed out that allocation formulas can lead to an aggregation of specialized materials and a lack of general interest and interdisciplinary materials; they also are often tied to faculty selection, and minimize the librarian’s role in building a balanced collection.⁶ Proponents say that they can help distribute funding equitably, minimize the tendency for rapid and vocal selectors to acquire a disproportionate share of funding, and cope with the political tensions of departments fighting for a fair share of materials funding.⁷ While acknowledging the limitations of formulas, Lowry observed in 1992 that allocation formulas serve two purposes: they help allocate limited resources equitably and “are extremely useful in solving the dilemmas of unequal resource allocation.”⁸ Lowry proposed a matrix formula including as many as twelve variables chosen based on institutional goals. Many libraries apparently agree with Lowry that formulas are useful since about 40 percent of academic libraries surveyed by Budd and Adams, and Tuten and Jones, have used formulas.⁹

The average cost of materials per discipline is a common, though not easily established, variable in allocation formulas. As Genaway noted, the collection of an academic library should proportionally reflect all the programs, instruction, and research of the institution.¹⁰ One could infer that subject fields receiving equitable library support would acquire the same proportion of titles published in their field. Since the average cost would significantly affect the library’s ability to acquire titles, a variable to compensate for the broad variation in average prices would help insure proportional subject acquisitions. Biblarz, Bosch, and Sugnet noted that book price disparities among disciplines should be considered as a major piece of data that drives resource development in respective subject areas.¹¹

Librarians have used numerous sources for average price information and noted problems with each. When Randall discussed allocating book funds based on average publishing output and cost of books published in a subject, he relied on A List of Books for College Libraries as his source of information.¹² However, he acknowledged that he omitted two foreign language departments from the formula because his source contained too few priced titles in foreign languages to be considered reliable.

Ellsworth tried using data from Publishers’ Weekly and Bookseller to determine cost and publishing output by fields, but experienced difficulty matching the publications’ general subject categories to academic departments.¹³ He also noted that these sources did not cover foreign books. Sweetman and Wiedemann identified subject categories and restriction to books published in the United Kingdom as problems in using price data published in the Library Association Record.¹⁴ When Werking and Getchell used Choice to estimate publishing output and cost, they identified matching Choice categories with academic departments, lack of vocational literature, and Choice’s focus on undergraduate materials as concerns.¹⁵

The costs compiled from approval plan vendor data and published in Bowker have been used by some libraries to determine average cost. When Axford compared average prices of books purchased through approval plans at three academic libraries to average prices reported in Bowker, he concluded that the differences between the published price index and the locally generated data were too large for Bowker to be used reliably for allocating book budgets among departments.¹⁶ In a letter replying to Axford’s article, Lynden and Birkel pointed out that the published indices were intended to indicate overall price trends and provide data for state and national policy decisions, not to reflect a particular library’s buying patterns.¹⁷ Lynden and Birkel opined that published indices could be useful in preparing budget justifications, but could not substitute for local statistics.

Rein discussed using data from the most recent Blackwell study.¹⁸ He reported the challenge of matching the study’s subject areas to the academic departments of George Mason University. Rein noted that data from two tables had to be combined to cover all the subjects relevant to a department. Rein also suggested that the library should consider foreign publications and nonbook materials in future price studies.

Cubberly also used approval plan data, noting the necessity of sorting the data by Library of Congress (LC) classification before matching it to department interests as identified in the collection development policy.¹⁹ Cubberly mentioned the lack of data on nonbook, foreign language, and retrospective materials as a concern, especially for the University of Southern Mississippi’s large music program. Similarly, Goehner noted a faculty committee’s recommendation that the library and foreign language department at his university work together to identify a book cost for foreign language materials, as data was not available from the library’s approval plan vendor.²⁰

Bourgeois reported that South-west Texas State University uses a faculty-determined allocation formula that includes two price averages for monographs.²¹ One average comes from vendor’s tables and the other from the price of monographs bought by the library in previous years. As O’Connor explained, the University of Technology, Sydney, uses actual prices paid by the library to calculate average cost, because that reflects reality for a library purchasing substantial amounts of overseas publications.²² Evans discussed the decision to change from using average price of previous purchases to price indices published in Bowker when Monash University Library (AUS) adopted a new allocation formula.²³ As Evans explained, the allocation formula committee felt that using average cost based on past purchases served to “enshrine past purchasing practice.”²⁴ The committee decided that adding the number of monographs published by subject and average price from a published index would address that concern.

Because the literature review described situations in various types of academic libraries, the authors selected forty-five Carnegie Doctoral/Research Extensive land grant colleges to survey about allocation formula use in institutions similar to MSU. The survey, which appears as an appendix, was developed to determine:

• if the libraries use a formula to allocate all or a portion of their materials budget to academic departments, and if so, what percentage is allocated to departments;
• what formats are included in the formula;
• if the average cost of books per discipline is used as a variable; and
• how that average cost is determined.

The survey also included questions regarding the amount spent annually on monographs and the portion of the monographic budget devoted to approval plans, if any. Twenty-six (58 percent) of the forty-five librarians responded.

As table 1 shows, only eight (31 percent) of survey respondents reported using a formula to allocate all or a portion of the library’s materials budget. A library’s use of a formula did not appear to be related to the amount of money spent annually on monographs. The library with the largest monographic budget ($5,000,000) and the one with the smallest ($241,500) both use allocation formulas.

Half of the libraries using an allocation formula included price as one of the formula variables and half did not. One respondent commented that the difficulty of matching the library’s fund and curricular structure with the book disciplines used in price indices led to a decision to omit price as a variable. Respondents mentioned using data from book vendors, Bowker, and local expenditure information from their integrated library systems to determine average price.

Since neither the literature review nor the survey of institutions similar to MSU pointed to a single best source for determining average book prices, the collection development unit chose to compare price information from four data sources: (1) approval plan profiles for each MSU department, (2) the Bowker study, (3) the Blackwell study, and (4) three-year average of local expenditures.

The authors chose to use the most current data available from each source in June 2006 to calculate average prices. The years for each data source varied, with the Bowker prices from 2004, Blackwell study from 2004–2005, approval plan from 2005, and three-year average expenditure from 2003–2005. The difference in time period covered by each source makes a direct comparison of the average prices inappropriate, but allocation formulas focus on the relationships between prices rather than the actual prices. The authors calculated the average price of books for each department from the most current available data for each source, as using the most current data is normal practice.

After the average price of books for each department was calculated from the four data sources, eleven departments were chosen for further study. The departments represent five of MSU’s seven colleges, with the College of Veterinary Medicine and the College of Architecture, Art, and Design omitted because of their highly specialized content needs. Other criteria determining selection included:

• representation of several disciplines within one department resulting from the merger of two or more departments (counseling, educational psychology, and special education; geosciences; human sciences);
• a single department’s reliance on resources from multiple disciplines (agricultural information science, biochemistry and molecular biology, entomology and plant pathology, industrial engineering);
• a significant percentage of nonbook or non-English language purchases (music education, philosophy and religion); and
• interesting discrepancies among the average prices from the various data sources (English, marketing).

The approval plan and Blackwell study methods tended to produce higher average prices, while the Bowker and expenditure methods tended to produce lower average prices (see table 4). Lower prices were expected from Bowker because the data reflected two-year old prices. The expenditure method produced lower prices because it reflected discounts received by the library, while other methods reflected list prices. Allocation formulas rely on the proportional differences in average book price among departments, not the absolute price. While the various methods produced differing average prices, the authors wanted to determine if all methods produced similar proportional differences among the departments.

The authors compared the four pricing methods using Pearson correlations (r) (see table 5). All four methods are significantly related to each other at the p<0.05 level (n=11, two-tailed). The Blackwell study and approval plan prices were the most closely related; the correlation between them was significant at the p<0.01 level (n=11, two-tailed). Significant correlations between the data were expected, as all four methods are attempts to measure the same variable, the average price of monographic materials in a specific academic field.

The authors computed the coefficient of determination (r[2]) to determine the shared variability among the methods. If the four methods are producing very similar proportional differences in average prices, one would expect the shared variability to be very close to 1.00. As shown in table 6, the shared variability ranges from 0.46 between the approval plan and expenditure methods to 0.83 between the Blackwell study and approval plan methods. Although the average prices calculated by all four methods are significantly correlated, the coefficient of determination shows that the source of the price data does make a difference in the proportional difference among the average prices, and in the department’s allocations.

Because none of the coefficients of determination are very close to 1.00, the methods of determining average price are not producing nearly interchangeable proportional differences. The MSU Libraries will need to consider all the pros and cons of each method to select the most appropriate method for local use.

Some factors considered by MSU that may be pertinent to other libraries considering sources for price data included difficulty of matching source subjects to departments, inclusion of non-English and nonbook materials or of paper bindings, currency of price data, and staff time needed to arrive at price averages. Table 7 summarizes factors that might be taken into account by MSU and other libraries with similar programs.

Two of the methods, expenditure and approval, do not require matching the source data’s subject divisions to departments. The expenditure method uses the prices of titles specifically selected and purchased to support the departments’ programs. However, the library will need an alternative way to determine average price when new programs are established or departmental programs change dramatically. The approval method uses existing profiles associated with the library’s approval or new book notification plans that already match subjects to departments. Libraries should be aware of how non-subject limits on the approval profile may affect the average price. For example, approval plans can be limited to university press titles or to books without media, which may produce average prices that differ from the average prices of books from all publishers or in all formats.

The Bowker and Blackwell study methods both require matching the source data’s subject divisions to departments. Bowker data are divided into broad subjects, so the library may need to use the same price data for several departments. Interdisciplinary subjects may need to be matched to several Bowker subjects. The Blackwell study data is divided into more than one thousand subject divisions based on LC classifications, allowing close matches between the departments’ interests and the subject divisions.

The library may consider the inclusion of price data on non-English, nonbook, or paper bindings important. A very small proportion of titles in the Blackwell study and Bowker data are non-English, while approval plans may exclude non-English materials completely. The Bowker and Blackwell study data do not include nonbook materials, and approval plans may exclude nonbook materials as well. The Bowker data include paper bindings, while the Blackwell study presents data on cloth and paper bindings separately, and libraries can chose whether to include paper bindings in approval profiles. The expenditure method has the advantage of including price data in the same proportion as non-English materials, nonbook media, and paper bindings are purchased. When non-English materials or nonbook media are a significant portion of the library’s purchases, the library should seriously consider using average expenditure in the allocation formula. Libraries also should consider whether the data source they use for determining average price matches the library’s preference for acquiring paper or cloth bindings.

Currency of data also may be a consideration for libraries. The MSU comparison used the most recent data available for each source at the time the study was conducted. The Bowker data was for books published in 2004, the Blackwell study for books published between July 2004 and June 2005, the approval plan for books published in 2005, and the expenditure method for books purchased in 2003–2005. Allocation formulas rely on the proportional differences in average book price among departments as opposed to the absolute price, so the currency of the data may not be important to the library. When the library uses the same average price data for other purposes, such as budget requests, currency may be a factor.

Finally, the staff time required to compile the data and compute average price should be considered. The expenditure and Bowker methods required minimal staff time. The Blackwell study method was more time-consuming, but had the advantage of allowing the library to match subjects to departments more closely than Bowker allows. The Blackwell study also reflects the English-language academic book market, while expenditures reflect only a small portion of the titles available for purchase. The approval plan method was very time-consuming, a major disadvantage for most libraries.

Libraries use price data in many ways, such as allocating funding among departments or broad subjects, preparing budgetary projections, and assessing charges for lost materials. The Library Materials Price Index Group (LMPI) has noted that the Bowker indices can serve as useful benchmarks against which local costs can be compared, but they cannot substitute for cost data that reflect collecting patterns of individual libraries.²⁶ LMPI has indicated an interest in pursuing studies that correlate individual libraries’ costs with the national prices. This study found a significant correlation between MSU expenditures supporting specific departments and the Bowker price index. Expanding the study to include all MSU departments would be necessary to validate this finding, and could serve as a case study for other libraries considering the most appropriate method for determining average price data for use in allocation formulas.