In a 2001, the California Digital Library Ebook Task Force conducted a survey of fourteen large academic libraries to learn about their e-book experiences.¹⁰ At that time, some libraries believed that “the role of e-books was not to replace print but to serve as a duplicate copy.”¹¹ Studies conducted by Columbia University (in 1997) and the University of Rochester (in 2001) on user perceptions and their use of e-books found that users were receptive to using electronic reference books for quick answers but less interested in reading electronic versions of academic books.¹² An ebrary survey in 2007 also indicated that users found e-books difficult to read, e-book platforms difficult to use, and training in using electronic resources in short supply.¹³ E-book suppliers had yet to develop a satisfactory purchase model that would make bulk e-book purchases easy for libraries.¹⁴ Not until the mid-2000s did e-books start to gain traction with academic library users and become an important academic library service.¹⁵

User perception of e-books also can be seen from usage analyses. Findings of recent usage studies of e-books have varied. Studies conducted at the University of Idaho and Oakland University documented low e-book use.¹⁶ In contrast, a Duke University study found that more e-books than print books were used.¹⁷ In determining what users need, investigating whether, for any given title, the electronic version is used more heavily than the print version is important. A 2010 report of a study at the Texas A&M University Libraries compared use of e-books and their print equivalents in the physical sciences and technology.¹⁸ The study results suggested strong user preference for e-book versions:

For NetLibrary, the top 10 science e-books were used over six times more than the [equivalent] print books, and the top chemistry e-books were used over 3 times more than their print counterparts. For ebrary, the top 17 science e-books were used at least 17 times more than the same print books. In Safari, the top 10 computer science e-books were used 207 times more than their print counterparts.¹⁹

The implementation of patron-initiated purchases by libraries also has changed the landscape of library collection development. The 2008–9 ARL statistics report documented a sharp increase in ILL borrowing during the previous two decades.²⁰ This clearly indicated that “local collections do not meet needs adequately.”²¹ In response to the unmet needs, purchase-on-demand (POD, sometimes called patron-initiated purchase), based on interlibrary loan (ILL) requests, was adopted by several libraries in the early 2000s.²² Some librarians have been skeptical of using patron-initiated ILL purchases as a way to develop library collections. The fear was that such a practice would result in unbalanced and possibly expensive collections of little use.²³ However, research has proven otherwise. Recent studies have presented consistent findings that the high percentage of the titles ordered through ILL POD was appropriate for library collections.²⁴ Further, the circulation rate of the ILL POD titles outperformed that of titles acquired via the conventional system.²⁵ A survey performed at Oregon State University also indicated high patron satisfaction with their Buy Request pilot initiative.²⁶ As a result, the Oregon State University Library has adopted it as a permanent service. Tyler and colleagues incorporated usage into their measurement of the cost of books (i.e., books become more cost-effective when they circulate more times) and found that the cost per title of ILL POD books was lower than books purchased through traditional means because of their higher circulation rates.²⁷ Patron-driven acquisitions programs have proved worthwhile on many levels, such as “cost-effective collection development; user satisfaction; high subsequent circulation; and flexibility in meeting local constraints for price, content, and processing.”²⁸

With e-books gaining popularity, along with the success of the print-based ILL POD model, libraries began to extend the ILL POD service by adopting PDA of e-books. The low circulation rate of print monographs and the rising number of ILL requests indicate an unmet demand among academic users.²⁹ Many libraries have implemented PDA of e-books with the goal of better meeting user needs in a timely manner.³⁰ Price and McDonald found evidence of heavy post–purchase usage of e-book titles acquired via PDA programs in their study—further justification for investing in an e-book collection.³¹ They found that user-selected e-books were used two to five times more often than librarian-selected e-books. Their findings also indicated fewer unused titles in their user-selected e-book collections.

Despite the rapid development of the e-book market, is completely replacing new acquisitions of print materials with e-books wise? Can e-book suppliers fulfill academic libraries’ needs in supporting campus teaching and learning? Price and McDonald conducted an e-book study in which they compared the print book purchases of five academic libraries with the content currently made available by four major e-book aggregators.³² Their study showed that 70 percent of print titles purchased by the participating libraries in 2006 and 2007 were not available from any of the aggregators included in the study. The average matching rates between actual library purchases and the content available from the aggregators was low, ranging from 5 to 25 percent. Inspired by this study, Pomerantz undertook a similar study at Adelphi University with an emphasis on nursing and business print titles acquired by the library.³³ Her study had similar findings: less than one-third of print books had e-book equivalents available from aggregators.

These studies focused on the availability of e-book equivalents of current print purchases in academic libraries. However, no study thus far has investigated whether libraries can meet current user needs with a greater reliance on e-book acquisitions. Meeting patron needs is more than a matter of determining whether e-book equivalents of the totality of current in-house collections are available. User needs are best reflected in actual materials used. By establishing the levels of e-book equivalency for materials that have been circulated, used in-house, or requested through ILL, academic libraries can ascertain whether the demand for those materials can be met through the acquisition of e-book content.

The authors turned to the local Voyager system to extract the first data source—circulation transaction logs of monographs in the General Collection for the period of July 2008–May 2011. The circulation transactions provided sufficient information about the checkouts of print titles and allowed the authors to examine the extent to which e-book formats could be expected to satisfy current circulation needs. The data for June 2011 were not included because they were not yet available at the time of data collection for the present study. The same limitation applies to the other two data sources discussed below.

To obtain an accurate circulation count for the print collections, the authors first used Microsoft Access queries to clean up the original circulation transaction logs for the study. While the circulation logs listed all individual items loaned to users during the three-year period, the unit of analysis in this study had to be titles rather than items. If users had checked out multivolume titles also available in e-book format, the unadjusted item-level circulation data might well affect e-book availability rates because such titles would be counted multiple times rather than once. For that reason, the authors treated item records for a multivolume title loaned on the same day as a single circulation transaction. Circulation transactions were counted as separate cases, however, when duplicate copies were loaned on the same day or when multivolume titles were checked out more than once in a single day. The authors removed titles lacking ISBNs from the data set because the vendors could not readily supply reports about the electronic counterparts of the print titles without ISBN data.

To obtain ILL book requests, the authors extracted the ILL data for the same three-year period from the OCLC Usage Statistics system (WorldCat Resource Sharing). The authors used Access queries to de-duplicate the ILL data using OCLC control numbers. This produced a final list of unique titles requested via ILL.

Finally, the authors used the “historical browse” data recorded in the local Voyager system to collect records of circulating monographs used in-house (i.e., General Collection titles taken off the shelves and left on tables or in other places). These data were recorded under TCNJ Library’s long-standing policy requiring that barcodes of books used in-house be scanned in Voyager before being reshelved. This scan triggers “historical browse” transactions in the system. The in-house use data collected have some limitations. The Voyager system does not register the “historical browse” transaction date, and pinpointing when particular in-house uses were recorded is not possible, meaning the data include print titles that may have been used before the timeframe of this study. Therefore, for data collection purposes, the authors only used in-house use data on titles that were cataloged during the three-year study period. Thus titles cataloged before the study period but used during that time were not included. Also, because TCNJ Library does not specifically ask users not to reshelve used items, books used in-house and reshelved by users are not tracked. Furthermore, without data for in-house use dates, determining when individual items in a multivolume title might have been used or whether they were used together or separately is impossible. The authors concluded that the number of such cases was small, constituting less than 3 percent of the entire “historical browse” data and not significant enough to skew the data. Thus the historical browse data extracted from the Voyager system do not provide a complete picture of in-house use of library materials. Regardless, because the goal was to determine the e-book availability of heavily used titles, the authors believed that historical browse data would provide helpful usage information.

The authors prepared the three data sets reflecting the different aspects of local print materials use and sent them to two e-book providers: YBP and Coutts. YBP is the main book vendor for TCNJ Library and has partnered with three of the main aggregators of academic e-book titles. Coutts is the academic arm of Ingram Content Group, which owns the MyiLibrary platform. The vendors then ran TCNJ Library’s usage data against their e-book title lists of the four aggregators to locate e-book equivalents of the print titles used by TCNJ users. YBP provided e-book data from three aggregators: ebrary, EBL, and EBSCOhost. Coutts provided MyiLibrary e-book data. Additionally, YBP noted which titles were being offered by their publisher partners outside of the three aggregator platforms and grouped these under the rubric “Publisher Direct,” a term the authors have kept in reporting results. The authors also believe that the four aggregators whose data were used in the present study are the major suppliers of e-books in the academic library market. The sample data studied, although not exhaustive, are considered to be representative of e-book offerings currently available to academic libraries.