Conclusions

Although the selection list navigation bar is more complex than the standard navigation bar, we feel that the benefits are real and that users will be willing to learn this new style. Properly implemented, selection list navigation bars could make navigation of hierarchical web sites easier and faster in the future.

Selection list navigation bars can provide many benefits. The results of this study indicate that, with practice, users can use them to find information faster and more directly than with standard navigation bars. On a web site, this could make it easier for users to locate information, products, and services, which could in turn make it more likely for them to purchase something from that company and return to that site in the future.

The selection list navigation bars do have several drawbacks. They are more complex and harder to learn than standard navigation bars. Many users will still prefer the standard navigation bars. We suggest that web developers provide users with the option of which type of navigation bar they use. This will be especially important as selection list navigation bars are first introduced. If they are used in the future, they will only gradually fade into mainstream use rather than exploding onto the scene over night. Developers should not expect sudden increases in the number of visitors to their sites or online purchases just because they use a new style of navigation bar.

Implementing the selection list navigation bars can also be more difficult than standard navigation bars. The selection lists do not require much programming, but a little bit for each page can add up quickly on a large site. Minor changes to the structure of a web site, such as just adding or removing a single page, can require updating the links of many navigation bars. We suggest storing the structure of a web site in a database that will make it possible to generate the selection list navigation bars automatically. This should greatly reduce development and update time. The web sites used in this experiment were generated automatically.

Because of the costs associated with selection list navigation bars, they are probably not for everyone. The types of sites that would probably benefit the most from implementing them would be large sites with many pages that are divided into a hierarchy of categories and subcategories. At large web sites, users will gain the most by using selection list navigation bars. At small web sites, understanding the structure and searching for information quickly are not as difficult, so the benefits might not be worth the implementation costs.

As an initial experiment in the use of selection list navigation bars, our conclusions are only preliminary. More research will be needed to verify and support these results. A wider variety of subject ages and educational backgrounds should be tested to determine whether the current findings are generalizable. As part of our justification for why the selection list navigation bar should decrease search time, we theorized that it would provide the user with a more complete understanding of the structure of the web site. With only a single subjective response question on the survey regarding this issue, we could not test this part of our theory. Future research might test subjects on their ability to recall or reproduce the structure of web sites using different types of navigation bars.

The design of this study leaves some room for improvement in future studies. To get relative subjective ratings on the surveys, we had to test each subject on all three treatments. Each treatment had to have a different structure and different questions. It is difficult to determine how much these differences affected the results of the study. In particular, question 9 of the standard treatment asked, "Which company offers sky-diving lessons?" Most of the subjects searched for the answer under the Sports or Recreation categories, but the answer was listed under Extreme Games within the Games category. That one question may have artificially increased the amount of time and number of page loads subjects spent in the standard treatment. In future research, either subjects should be tested on only one type of navigation bar or every combination of navigation bar and question set should be tested to balance out any potential biases.