Challenging Current Practice

Challenging Current Practice

by Dr. Bob Bailey

April, 2002

Is it better to have more items on a page and requiring fewer pages to be accessed (wide breadth), or to have fewer items per page and require more pages to be accessed (more depth)? Based primarily on studies reported three years ago by Larson and Czerwinski (1998) and Zaphiris and Mtei (1998), designers have been encouraged to construct broad, shallow sites.

Is it better to show one page containing frames with simultaneous menus, or to use the more traditional sequential strategies? In the former situation, users stay on the same page and select from different menus that are located in different frames. There is no need to change pages. In the latter case, users click from page-to-page until they reach a target or answer a question. Designers have been encouraged to use simultaneous menus whenever the task requires users to continually use the Back button (Hochheiser and Shneiderman, 2000).

Can users tell when they experience an interface that helps or hinders performance? Many designers still ask users whether or not the interface is adequately designed.

A recent study from Japan (Tsunoda, et.al., 2001) challenges some of our previous conclusions. Based on their review of the literature, they proposed three hypotheses:

H1: No matter what users are doing, fewer layers will result in faster task performance times ("broad" is always better than "deep").
H2: No matter what users are doing, task performance times using simultaneous menus within frames on one page will be faster than using the more traditional method of clicking from page-to-page.
H3: Task performance time and ease of use will be closely associated, i.e., when users do well with a task they will prefer the interface, and when they do poorly they will dislike the interface

Their first experiment had 89 participants search for information using five different formats:

Four-pages showing 3 items on each page for a total of 12 items, and requiring users to make a minimum of 4 clicks,
Three-pages showing 9 items on the first page and 3 on the other two pages for a total of 15 items, and requiring users to make a minimum of 3 clicks,
Two-pages showing 27 items on the first page and 3 on the other page for a total of 30 items, and requiring users to make a minimum of 2 clicks,
One-page showing all 81 items and requiring users to make only 1 click, and
One-page with three frames showing 9 items on the first frame and 2 items in each of the other frames for a total of 15 items, and requiring users to make a minimum of 3 clicks (no new pages are selected)

They found no reliable difference in task performance times across the four "page" conditions, but did report that the frame condition took reliably longer than all of the "page" conditions. The "Four-page" and "Three-page" layouts were preferred.

In their second experiment they kept the breadth and depth the same, but added "price" to the final page or frame. Also, they changed the task to one of finding either the best or worst price. This required users to move back and forth between pages (or frames) to compare the prices of products before making an answer. By changing the task they also changed the minimum number clicks that were required:

The Four-page format required users to make a minimum of 22 clicks (not 4),
The Three-page format required users to make a minimum of 15 clicks (not 3),
The Two-page format required users to make a minimum of 8 clicks (not 2),
The One-page format required the same 1 click as before, and
The One-page with three frames required users to make a minimum of 9 clicks (not 3)

In this experiment, the Four-page and the Three-page layouts produced reliably slower task performance. The "One-page" and "One-page with frames" were reliably preferred over the other three formats.

None of their hypotheses were true. Having fewer layers did not guarantee faster task performance times, and using simultaneous menus within frames did not guarantee better performance times. Finally, users preferred some formats that did not help their performance, and disliked others that helped. This study reminds us (again) to use caution when applying research results in actual situations. Using research-based guidelines can help direct the initial decisions, but then these decisions should be validated with appropriate usability testing.

References

Hochheiser, H. and Shneiderman, B. (2000), Performance benefits of simultaneous over sequential menus as task complexity increases, International Journal of Human-Computer Interaction, 12(2), 173-192.

Larson, K., and Czerwinski, M. (1998), Web page design: Implications of memory, structure and scent for information retrieval, CHI 98 Conference Proceedings, 25-32.

Tsunoda, T., Yamaoka, T., Yamashita, K., Matsunobe, T., Hashiya, Y., Nishiyama, Y. and Takahasi, K. (2001), Measurement of task performance times and ease of use: Comparison of various menu structures and depth on the Web, Proceedings of the Human Factors and Ergonomics Society 45th Annual Meeting, 1225-1229.

Zaphiris, P. and Mtei, L. (1998), Depth versus breadth in the arrangement of Web links

Zaphiris, P. (2000), Depth vs. breadth in the arrangement of web links, Proceedings of the IEA 2000/HFES 2000 Congress, 453-456.