2. Experiment

2.1 Introduction and Hypothesis

As information becomes more available on the Internet and as the size of web sites continue to grow, the time it takes to navigate through a site to find specific information becomes an important factor. Web pages are usually grouped together under a specific category or sub-category. Most, if not all web sites have some sort of navigation menu that links to a sub-category of links that lead to deeper levels of the site. The location of the navigation menu could play an important role in search time. Many sites employ frames to differentiate the menu of links from the actual pages. Positioning of the menu of links can vary from the top to the bottom of the window, from the left side to the right side of the window. Though many sites seem to use a menu frame on the left-hand side, it may not be the best placement. Our experiment will attempt to prove whether a measurable difference can be discerned from changing the location of a menu in a web site.

It is our hypothesis that the test with the interval links placed in the frame along the top will result in the best navigation performance. Several factors lead us to decide that the positioning of the menu frame at the top of the browsing window is optimal. First, as in many software applications in the Windows environment, menus are usually found at the top and an experienced computer user may naturally gravitate toward the top of the screen to look for the menu. Also, the average distance from the each link in the menu frame to our starting point, the center of the larger frame, is shorter for the links positioned along the top and bottom than along the left and right sides. These circumstances convince us that having a menu for a web site located at the top of the browsing window would be most beneficial.

Our pilot study revealed some improvements to the initial experiment that we developed.

Initially, the first random number showed up in the main frame upon entering the first test. The clicking of an interval in the smaller frame produced the next random number in the larger frame. We observed that during some experiments the random numbers produced were within or close to the same intervals where at other times the random numbers produced intervals spaced farther apart. We found that having the subject return to the mouse pointer to a starting position within the larger frame helped to increase the reliability of the test .

A few other details were ironed out in the pilot tests. The overall look of the experiment was finalized by defining the color scheme and refining the font sizes. A few errors in the timing mechanism were also fixed. A zero was added to the number one in the first interval to provide a consistent clickable area.

2.3 Subjects

Twenty-four students from the University of Maryland with some level of web browsing and computer experience took part in our experiment. Each participant was well versed in navigating web sites and was randomly given a different test order.

The experiment was carried out on the Gateway 2000 computers in the Van Munching computer lab. The computers have 17" monitors running at 800x600 resolution. Microsoft Windows 95 was the operating system and Netscape 4.0 was the web browser used. The application's viewing window was maximized.

We conducted our experiment by the using following procedures:

• Show the subject the experiment introduction.
• Ask the subject if there are any questions/concerns and make sure they know the proper course of action for the experiment.
• Present the subject with the one of the testing orders.
• Allow the subject to complete the tasks.
• Record the results as the subject completes a survey.

Our experiment consists of four tests. Each test contains two frames. The smaller frame has 10 links and these links are labeled with an interval of ten numbers sequentially between 1 and 100. The larger frame has a link that produces a random number. The difference between the four tests is in the placement of the smaller frame, either to the left, right, above, or below the other frame.

By clicking the link in the larger frame a number will appear in the frame. The objective is to click on the interval link in the smaller frame that corresponds to that number. Once the interval is clicked, the time is recorded and a link to get the next random number again appears in the larger frame. This event occurs five times for each of the four treatments. The time it takes for the subject to recognize the number and click the correct interval is timed internally. Once the subject finishes the four tests, the results of the experiment are retrieved by the one member of our group and a survey that gauges the participant's preferences as well as their background in computers is filled out.

Continue
Return to Menu Frame Placement for Optimal Web Navigation