Effects of Link Arrangements on Search Efficiency
2.1 Introduction and Hypotheses
Internet Web page designs vary drastically from one site to the next. Many books are available on 'how to' design Web pages, but none of them discussed link arrangements in detail. Specifically, they did not discuss link arrangements using column format s or multiple pages requiring scrolling. What is the best way to design the page when there are hundreds of links? While interning at the Naval Academy, the second author noticed that the search links were subdivided into categories. Beneath each category were subcategory links listed individually in a paragraph format and separated by commas. The links were hard to read and distinguish from those above, below or to either side. Thus, we decided to design an experiment to test for a more efficient an d faster Web design for link arrangements.
Our hypothesis: a three-column link arrangement would have faster search times than the one- or two-column link arrangements. The search time would decrease as the number of columns increased. The independent variables were a 30- and a 155-link Web page. There were three treatments: one-, two-, and three-column link arrangements. The dependent variable was the search time (in seconds) to find a specified link.
2.2 Pilot Study Results
Pilot study results showed that the three-column link arrangements had the fastest search times, corresponding to the hypotheses. The one- and three-column link arrangements search times were very close for both the 30- and 155-link groups. The data collected reflects that the two-column link-arrangement is not as efficient for Web link arrangements.
2.3 Subjects
Forty-two midshipmen, staff and faculty members from the United States Naval Academy with varying levels of computer experience, participated in the experiment. Group assignments were made by switching between the 30- and 155-link Web page designs and alternating between the treatments -- one-, two-, and three-column treatments. The satisfaction questionnaire provided the following background information regarding Internet use: responses varied from zero to 60 times for the Naval Academy Web page, while the World Wide Web use varied from zero to +200 times in the past 30 days.
2.4 Materials
Hardware: The experiment took place in a computer-lab classroom at the Naval Academy. Each participant used a 486 DX system, standard VGA CTX1451 monitor, standard 101/102 keyboard and serial mouse. Participants used the mouse and/or keyboard to navigate the Web pages. The monitors were set at an 800 x 600 resolution. The mouse position on the desk was determined by the participant.
Materials: A multi-purpose form was used -- side one had the required informed consent and side two indicated which Web page the participant tested. It also provided spaces for participants to enter start/stop search times. It asked about the past 30 days internet use for the Naval Academy Web pages and the World Wide Web. There were several questions pertaining to subjective satisfaction that subjects answered after completing the study. Participation was voluntary.
Design: The Web pages were designed using links from the U.S. Naval Academy Web pages. These links were selected because participants were familiar with the military terminology used. To ensure participants were not scanning documents prior to the start of each task, there was an internal hyperlink 'START' at the top of each Web page followed by a digital JAVA clock and then blank space. The task search links were then located below the initial screen. The digital JAVA clock showed hours, minutes, and seconds to facilitate participants logging start and stop times for each task. Each of the task links (3 per page) were internally linked to return the participant to the top of the page to the 'START' and JAVA clock.
2.5 Procedures and Problems, Administration, Grading
Procedures
The experiment was completed over a two-day period at the U.S. Naval Academy. Participants came to the computer lab at a time convenient to their schedules. Upon arriving, they were given oral instructions on signing the informed consent form, marking the start and stop times, and completing the subjective satisfaction questionnaire. They were then instructed on how to start the test and select (click) the task search links.
Participants were assigned a Web page (USNA1, USNA2, USNA3, USNA4, USNA5 or USNA6) when arriving. Setup of each Web page was completed prior to the participant sitting down at the computer. No training was necessary. After being given verbal instructions and completing the informed consent form, the subjects selected a task list from the proxy. Six task l lists for independent variable were available to ensure random search orders.
The participants marked the start time using the JAVA clock (12:10:30) and clicked START using a mouse. The page then internally linked to the lower area where the links were located. The participant would search for the first task link and click on it when found. The link would return the participant to the top of the page with the start screen where he would mark the stop time using the JAVA clock. The participant would repeat this two more times. Upon completion each participant completed a subjective satisfaction questionnaire and was thanked for his or her time. The experiments ran smoothly with no problems.
The administration of the experiment required designing six Web pages, three 30-link and three 155-link pages. The questionnaires were designed to gather information regarding the space between the rows and columns of links, search ease and link arrangement satisfaction -- the ratings were from one (hard to follow) to nine (easy to follow). Information was also requested concerning the participants past 30 days use of the Naval Academy Web pages and World Wide Web.
Continue 3.
Results
Title Page for
Effects of Link Arrangement on Search Efficiency