Kate Fuller, Jill Livingston, Stephanie Willen Brown, Susanna Cowan, Thomas Wood, and Leslie Porter
Print version (Adobe Reader required)
During fiscal year 2006, the University of Connecticut Libraries spent almost two-thirds of its collection budget on electronic resources, making it essential that students, faculty, and staff can find and access these resources without assistance from librarians. To address ease-of-use issues, a cross-functional task team spent a year assessing the libraries’ database locator and worked to create a more functional system. This iterative process of usability testing and design included three sets of usability tests, several design sessions, and revision of database descriptions. The new design now enables users to successfully and quickly find databases without mediation.
Libraries spend an increasing amount of their collections budgets on online resources: In fiscal year 2005, libraries in the Association of Research Libraries spent more than 35 percent of their collection budget on electronic resources, while public libraries allocated 8.75 percent of their collection budget to electronic materials in fiscal year 2004.1 The University of Connecticut Libraries (UConn Libraries) spent approximately 62 percent of its collection budget on electronic resources in fiscal year 2006. Libraries now subscribe to dozens, if not hundreds, of research databases, yet patrons usually stick with the research resources they know: For undergraduates, these are Google and Wikipedia; for faculty, these are the few databases they learned while in graduate school. To support their substantial investment in electronic resources, it is imperative libraries make them easily available to patrons. Unfortunately, there is often a disconnect between the existence of libraries’ electronic resources and patrons’ knowledge of them. How can libraries best showcase the electronic databases available to patrons? A task team at the UConn Libraries recently spent a year working to resolve this problem; the result is a usable, functional solution that improves the resource discovery process for librarians and patrons alike.
The Database Locator: Opportunity for Improvement
As at most academic libraries, the UConn Libraries invest heavily in electronic resources in both dollars and staff time. For this reason, it has become more central to the UConn Libraries’ mission to increase awareness and usability of electronic products. In 2000, the UConn Libraries began delivering electronic resources to end users through a database that became known as the Research Database Locator (RDL). Newly acquired databases were added to the resource with elements such as the database name and access URL, a one to two paragraph description of each resource, licensing and access information, and a list of subjects to which the database was relevant. Patrons and librarians both used the RDL to find databases relevant to their research.
However, the database locator did not include a component for managing the administrative aspects of these resources. In early 2005, the UConn Libraries began to develop a more formal electronic resource management (ERM) system based on the Digital Library Federation (DLF) Electronic Resources Management Initiative (ERMI). This new system combined the public resource discovery tool with modules for managing electronic resource elements. This ERM tracks a variety of information, including all of the UConn Libraries’ licensed database resources, most of the journal packages, individual journals when license information is required, and a handful of freely available Web resources such as AGRICOLA and MedlinePlus. Altogether, there are currently more than 450 resources in the ERM.2
As work was completed on the back end of the ERM in 2006, an ERM public interface team (PERM), assembled to discuss the design of the system for end users. The team wanted to look at the ERM from the users’ perspective with the goal of redesigning the Web interface to enable students, faculty, and staff to use the system without assistance from a librarian. The team spent approximately one year on this project and approached the problem from several different angles. Early work began with a literature review of website usability testing in libraries and continued by scouring the Web for examples of library database locators that met usability heuristics and used simple, clear language to describe databases. The team conducted a preliminary evaluation of the ERM from the patron’s perspective, assessing the site on the basis of common Web design and usability principles and evaluating query logs and usage data for the ERM.
Finally, the group began a several-month iterative process of usability testing and design, beginning with a test of the ERM public interface and continuing with an initial redesign. Database descriptions were rewritten to provide key information in a succinct format. The new design was tested with faculty and students, redesigned accordingly, and retested until the team was satisfied that the new system enabled users to successfully and quickly find useful databases without mediation.
Examination of Database Access Tools
There is an abundance of books and articles that describe the value and methods of usability testing. Of note is a guide by Jeffrey Rubin, which discusses the methods and materials needed for testing.3 This practical, start-to-finish guide to testing documents (as well as a primer on the setting, materials, and methods for conducting and analyzing tests) is well known and well referenced. Also well recognized within the field are the strategies of usability expert Jakob Nielsen. In columns on his Alertbox website, Nielsen recommends conducting an initial heuristic evaluation of a website followed by iterative usability testing.4 His studies show that testing with five users will reveal 85 percent of usability problems, and he advocates conducting three rounds of tests with five users each so that a site can be continually tested and improved. Darlene Fichter, a librarian whose interests include human–computer interaction, reinforces the value of rapid iterative testing. She recommends using a casual, try-and-see approach as opposed to the often slow-moving “next major release” approach. According to Fichter, the former “drives creativity and new ideas” and gives designers a chance to try changing different elements on the basis of user testing.5
A number of libraries have successfully conducted usability testing and redesign of library websites and have documented their endeavors in case studies. Cobus, Dent, and Ondrusek provide a detailed, step-by-step discussion of how usability testing was conducted on the Hunter College Library website and how testing evolved as librarians determined problems with their own testing strategies.6 Battleson, Booth, and Weintrop, from the University at Buffalo Library, provide a quality introduction to usability engineering and human– computer interaction in their case study report.7 Additionally, Manzari and Trinidad-Christensen write about their noniterative usability study of the Library and Information Science Library website at Long Island University.8 Other case studies are published that provide additional frameworks for testing in libraries.9
We conducted a wide-ranging examination of other libraries’ database access pages. During our scan, we took note of interface features we found to be effective; however, we were particularly interested in identifying database locator tools that exemplified use of best practices. In defining these best practices, we used as our guide the principles that Jakob Nielsen has defined in his usability research.10 In addition, when we were examining specifically the use of particular terminology, we were guided by John Kupersmith’s seminal article, “Terms that Library Users Understand.”11 We view the usability research of Nielsen and Kupersmith as comprehensive and complementary; therefore, when we examined other libraries’ tools, we evaluated the search functionality, navigation, and icon design in terms of their learnability, efficiency, memorability, errors, and satisfaction. We evaluated the database locator tools’ terminology with the following practices in mind: avoiding words that users often misunderstand, using natural language equivalents, enhancing or explaining confusing terms, providing intermediate pages, providing alternate paths, and consistency. The following are websites we found worthy of note (many of these websites have been redesigned since our initial examination):
- North Carolina State University’s Databases page. What we liked: The streamlined approach. NC State’s “Find Articles” page does not have a keyword search option, but rather offers a subject drop-down menu and a “databases by title” option. The site provides help at the point of need; for example, a “Tips” box is displayed prominently in the upper-right corner of the search screen that says “not sure where to start?” directing users to general databases. This is a good use of providing alternate paths. On the results screen, there are usually only three to five databases displayed; as a result, these results appear “above the fold” so users don’t have to scroll. On these pages, there is another “Tips” box, which contains helpful links on topics, such as how to find newspaper articles, something that often stymies users. All of these features fall in line with Nielsen’s principles.
- University at Buffalo Libraries Resources by Subject page. What we liked: The unique ways to access databases. These include a “Best Basic Resources” page that suggests databases on the basis of eight typical student search needs (e.g., finding statistics). On the databases results screen, the “top resources” on the subject are listed first, using very clear icons that show whether a database is restricted to University at Buffalo or is accessible to the public. On the right side of the screen, the librarian responsible for the subject area is listed prominently, with an e-mail link.
- University of Toronto Libraries homepage. What we liked: The user-friendly language. The “best research resources” pages address a common problem many library users have—making the leap from a specific research topic to a broad subject—by giving a tip beneath the subject drop-down box (“For example, are you writing a paper on ‘panic attacks’? Choose psychology.” This is a good example of using natural language to describe how to search.) The results page employs the wording “best article databases” and includes only three to five results so the user is not overwhelmed and scrolling is kept to a minimum.
After examining other library’s research database locator tools, we selected features that reflected best practices, as we had defined them, for user search interface design. It was clear that avoiding jargon, providing hints, limiting the number of results, and predicting common user problems—such as searching for topical rather than broad subject keywords—were important factors in the design of these exceptional existing database locator tools. We would later take these design elements into consideration when improving the design of our own database locator. But before we could begin to redesign the site, we needed to test the existing interface to determine where users encountered the most difficulty.