Vice Provost for University Libraries
University of Connecticut
Assistant Dean, Graduate School of Library and Information Science
Web-based surveys are often overlooked as a tool for measuring the usage of networked electronic resources. The two major difficulties are generating a truly random sample of users/uses and minimizing the effects of non-respondents, thus assuring the representativeness of the sample. MINES for Libraries™, a web-based survey methodology and one of ARL’s New Measures Initiatives, is proving to be a valid and reliable method for web-based surveying of networked electronic resource usage. This presentation will discuss the following questions:
• In a web-based usage survey, what is the appropriate unit of analysis: the user, each instance of usage, a count of what resource was used, or some other measure?
• Can the web-based survey be made replicable and reliable?
• How serious is the problem of non-respondents in the web environment?
• What is a valid random sampling scheme for web surveys that will permit the use of inferential statistical tests on the results, and not just descriptive measures?
• What happens to the results if the survey is made mandatory or optional?
• Can a web survey method collect valid, reliable and commensurable assessment data for all of the library’s electronic resources, including: electronic journals, electronic books, databases, online catalogs, digital collections, and other electronic services?
• What should the network topology look like to facilitate web-based survey assessment — that is, is there a network infrastructure of assessment?
Data taken from 50,000 surveys in more than thirty North American libraries during the last two years will illustrate the discussion, and provide examples of web-based survey results.