Overview of Metalib Usability Testing

Student Types

1. The first or second year Yale student with modest needs for Library resources, and mostly want an easy way to see if the library has a book they want.
2. The more advanced Yale student who is beginning more intensive research, and may be thinking of starting a thesis. This student has some experience with databases beyond Orbis. This student still wants easy searching, but also needs help with identifying appropriate databases. This could be a key group for using Metalib.
3. The graduate student is well acquainted with research, and who has very specialized research needs. This student is probably very familiar with one or more research databases. May need to occasionally reach beyond the databases they are most comfortable with.

There is confusion about which results come from books and which results come from articles. There is sometimes confusion about where to search for these different items. By putting all these things into one database we may add to this confusion. Additionally there is confusion about how to find databases in a subject area. Students do not have a clear idea what is being searched. Asked to find databases about american history, students may go to QuickStart and type in "american history."

Subject searching leads people to trouble. Surprisingly 3 of 6 undergraduates tested in September 2005 went out of their way to select subject search, usually because they wanted to be more focused in their approach. Subject has been removed as a search option.

Almost uniformly (5 of 6) undergraduates were not able to find book holdings in the Metalib full record display, and mistook the LC call number for a Yale call number. To address this, the LC call number was removed from the full display in Metalib. However, the four students tested recently continued to have some trouble locating holdings. Only one student of four went to the simple holdings screen.

Similarly with articles, 3 of 6 undergraduates did not find it obvious how to get full text or how to check for print. The one Metalib feature that helped was the prevalence of full text brought into the Metalib record from some databases. By chance, several undergraduates hit records containing full text in Metalib, and it was apparent to them that they had found full text. Again, in the most recent testing the most successful result to find an article came when the students went to a fulltext resource, Project Muse.

Multidatabase searching is problematic. The most common response from people when asked to find a set of databases in a topic area was to go to MultiDatabase and then search for the category (that is, they entered "history" in the center search box). 3 of 6 undergraduates after a great amount of fumbling are able to correct for this problem, but it is not intuitive. Although the configuration of Multidatabase search bothered people, it was encouraging that half were able to correct their mistakes without being shown what to do. In response the multidatabase page was simplified somewhat to remove a step in choosing Category.

In the more recent November test, no student went directly to multidatabase search to find a set of databases in US history. 3 of 4 tried to enter the term history or american history in a search box. One clicked on H in the Find Database section of Metalib. Once in multidatabase, students seemed able to use the revised menu structure well.

In the September test of undergraduates were often unable to make a connection between the process of clicking on "Add to My Research" and then going to the "My Research" section of Metalib(4 of 6 failed to do this). The Add to My Research and My Research sections of Metalib remain troublesome. It is not clear to students what Add to My Research does. If students try to Add databases to their research before they are logged in, this is particularly misleading (they won't find the databases listed under My Research). The lack of check boxes for selecting citations is a problem. As one student pointed out, check boxes are a standard feature.