Academic journal article Information Technology and Libraries

Classification of Library Resources by Subject on the Library Website: Is There an Optimal Number of Subject Labels?

Academic journal article Information Technology and Libraries

Classification of Library Resources by Subject on the Library Website: Is There an Optimal Number of Subject Labels?

Article excerpt

The number of labels used to organize resources by subject varies greatly among library websites. Some librarians choose very short lists of labels while others choose much longer lists. We conducted a study with 120 students and staff to try to answer the following question: What is the effect of the number of labels in a list on response time to research questions? What we found is that response time increases gradually as the number of the items in the list grow until the list size reaches approximately fifty items. At that point, response time increases significantly. No association between response time and relevance was found.

**********

It is clear that academic librarians face a daunting task drawing users to their library's Web presence. "Nearly three-quarters (73%) of college students say they use the Internet more than the library, while only 9% said they use the library more than the Internet for information searching." (1) Improving the usability of the library websites therefore should be a primary concern for librarians. One feature common to most library websites is a list of resources organized by subject. Libraries seem to use similar subject labels in their categorization of resources. However, the number of subject labels varies greatly. Some use as few as five subject labels while others use more than one hundred. In this study we address the following question: What is the effect of the number of subject labels in a list on response times to research questions?

Literature review

McGillis and Toms conducted a performance test in which users were asked to find a database by navigating through a library website. They found that participants "had difficulties in choosing from the categories on the home page and, subsequently, in figuring out which database to select." (2)

A review of relevant research literature yielded a number of theses and dissertations in which the authors compared the usability of different library websites. Jeng in particular analyzed a great deal of the usability testing published concerning the digital library. The following are some of the points she summarized that were highly relevant to our study:

* User "lostness": Users did not understand the structure of the digital library.

* Ambiguity of terminology: Problems with wording accounted for 36 percent of usability problems.

* Finding periodical articles and subject-specific databases was a challenge for users. (3)

A significant body of research not specific to libraries provides a useful context for the present research. Miller's landmark study regarding the capacity of human short-term memory showed as a rule that the span of immediate memory is about 7 [+ or -] 2 items. (4) Sometimes this finding is misapplied to suggest that menus with more than nine subject labels should never be used on a webpage. Subsequent research has shown that "chunking," which is the process of organizing items into "a collection of elements having strong associations with one another, but weak associations with elements within other chunks," (5) allows human short-term memory to handle a far larger set of items at a time.

Larson and Czerwinski provide important insights into menuing structures. For example, increasing the depth (the number of levels) of a menu harms search performance on the Web. They also state that "as you increase breadth and/or depth, reaction time, error rates, and perceived complexity will all increase." (6) However, they concluded that a "medium condition of breadth and depth outperformed the broadest, shallow web structure overall." (7) This finding is somewhat contrary to a previous study by Snowberry, Parkinson, and Sisson, who found that when testing structures of [2.sup.6], [4.sup.3], [8.sup.2], [64.sup.1] ([2.sup.6] means two menu items per level, six levels deep), the [64.sup.1] structure grouped into categories proved to be advantageous in both speed and accuracy. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.