Abstract
Management of a broad range of statistical measurements was the focus of Nancy Beals' presentation on electronic resource statistics. Wayne State University, with a student population of more than thirty-one thousand students, has one hundred thirty thousand electronic holdings spread throughout three hundred databases. Data for electronic holdings usage are captured from multiple sources and can be incorporated into easily accessible spreadsheets. The data can then be used to inform collection development decisions.
INTRODUCTION
As Electronic Resources Librarian for Wayne State University, presenter Nancy Beals was confronted with the very large task of managing the statistics that one hundred thirty thousand plus electronic holdings can potentially produce. She spoke to the audience about the analysis and process that goes into planning a centralized information source created from a multitude of decentralized statistical measures of electronic resource usage. Beals noted that measurement tools are now available that venture far beyond impact factors and citation analysis common to print titles to help a library determine how its resources are being used. These tools present both opportunities and challenges by showing how, according to Beals, “statistics can play a more progressively important role in determining the return on investment with library dollars.”
STANDARDS AND DATA COLLECTION ISSUES
Libraries can now capture large amounts of raw data from many different publishers. Use of the data must be considered alongside some of the problematic issues that come with data collection. Some of these data are captured based on industry-developed standards. Although the publishing industry has begun to recognize the need for standards to create meaningful comparisons, standards such as COUNTER (Counting Online Usage of NeTworked Electronic Resources) are relatively new and are not used industry-wide. Data can also be inconsistent or incomplete when systems break down. Data may not be instantly retrievable and can be skewed if library personnel do a significant amount of work with the records.
OPPORTUNITIES FOR USING DATA
The opportunities for using the data are numerous and can help a library develop the criteria necessary in making collection development decisions. Data show what is being used, as well as how it is being accessed. Even taking into account the possible flaws, the data provide numbers which can be used to measure such variables as average cost-per-title, usage of titles within packages, and usage in different subject areas. These numbers can suggest areas where a library may need to promote a resource or to justify spending and/or cancellations. The numbers can also inform a library's decision on space and storage issues. Trends can be analyzed by looking at factors such as full-text downloads, cost-per-use, and use by subject.
REALITIES OF DATA COLLECTION
Collecting usage data has become a necessity rather than an option. Insatiable demands for electronic resources and budgets that cannot handle those demands require quantifiable measures to defend purchases and provide equity in spending across subject areas. Beals has developed a “Stats Master” spreadsheet that describes available measurement tools. Selectors at Wayne State are then brought together for training sessions where data are explained and interpreted.
The Stats Master includes information about the provider, the platform, and COUNTER status, as well as other metrics available from commercial providers. Data for the library are harvested from Scholarly Stats, Serials Solutions, Innovative Interfaces Inc. (III) reports, and the III electronic resources management (ERM) module with the Standardized Usage Statistics Harvesting Initiative (SUSHI) protocol. For example, reports from III can compare statistics between different resource providers. Scholarly Stats can show the top ten journals with the largest number of full-text requests by platform and other variations based on high usage or low usage. Serials Solutions can show click-through statistics by providers, journal titles, and databases.
STATISTICS INTEGRATION
Data from such metrics provide a quantifiable means for Wayne State to cope with decreases in funding by using data-driven decision making. Selectors are trained in how to use the statistics and how to integrate statistical resources into selection criteria. Beals is able to provide data in comprehensive, easy-to-read quarterly reports that provide the selectors with the tools to do their own analysis. Several reports including cost-per-use (CPU), provider details, and duplicate subscriptions were shown as examples during the presentation ().
TABLE 1 Cost-Per-Use Example
Training sessions are informal and begin with a general approach to interpreting statistics, such as searches, sessions, and cost-per-search. Sessions go on to include examples from specific providers such as Serials Solutions and Scholarly Stats. Definitions of National Information Standards Organization (NISO), COUNTER, and SUSHI are provided. Selectors are given a schedule of when reports are available and cautioned regarding some of the issues, such as lag time from the publisher, that may compromise the data.
FINDINGS AND CONCLUSIONS
Through this first detailed analysis, library staff members learned that there was significant overlap, that publisher platforms are not necessarily used more than aggregators, and that cancelling packages could be looked at as an option for savings. Future variables may also become more meaningful as statistical measurements become even more granular. Data analysis in the future may incorporate article, chapter, or specific page downloads. Use of links to graphics, back buttons, and links out to Web information may also be counted.
DISCUSSION
Discussion questions were of both a general and a technical nature. In response to one question regarding the complexity of the project, Beals mentioned that she started with the smaller packages. Faculty input has been limited because this extensive project is fairly new. An audience member noted that in general, faculty members prefer summary data to large amounts of raw data. Selectors, as noted during the presentation, are provided with reports and do not have to compile large amounts of data themselves.
A question about the use of the SUSHI protocol highlighted the need for cost data to be in order records so that Scholarly Stats could produce cost-per-use data. Beals also emphasized that criteria such as session numbers versus searcher numbers, as well as the significance of other variables, must be tailored to the needs of individual institutions.
Anomalies such as platform or publisher problems should be taken into account when making decisions. One audience member suggested that anomalies can be counted against the total number of hits. Another audience member suggested that low numbers for specific database use may indicate that the library needs to better educate users on how to use a particular database.
One very interesting question centered on how to measure the use of social software as a possible access point to a particular source. Are users linking into the library resources through Google or PubMed, or even from a Facebook page? One of the publishers in the audience said that they were indeed trying to monitor such gateways. Another technical issue mentioned was the difference between information databases and e-journals themselves. While e-journals can provide COUNTER-compliant statistics, not all databases are created equal. There are databases, of course, that have full-text articles in them that do provide COUNTER statistics. There are also COUNTER Database Report 1 (DB1) and Database Report 2 (DB2) that provide usage data on searches, sessions, and turnaways for the whole database independent of source material within the database. Differences between databases need to be kept in mind as standards are developed through library and vendor/publisher interactions.