User Friendly Web Needs Rigorous Study

By John Makulowich, Senior WriterAdditional systematic study and hypothesis testing are needed for the World Wide Web to become more user friendly, according to experts who attended an influential conference on human factors and the Web."We need to develop better tools for analysis and better Web development tools. And we need to get the message out about the cost benefits of good sites, especially with the rise of e-commerce," said Sharon Laskowski, manager of the Visualization and Virtual Reality Group at the Commerce Department's National Institute of Standards and Technology, which hosted the event at its Gaithersburg, Md., facility."Usability is important. It is more than just making a pretty site," said Laskowski, who chaired the Fifth Conference on Human Factors & the Web, held June 3.But proper analysis of Web usability takes time and funding and only now is the Web reaching a stage of maturity where such experimentation can be pursued adequately."We are now producing solid research that can be transitioned to industry. Part of our role at NIST is to do exactly that," she told Washington Technology in an interview.Attracting the most attendees in its five-year history, the conference served as a forum for information sharing among 325 human factor engineers, designers and developers from industry, academia and government agencies.The one-day event was sponsored by AT&T Labs, Bell Atlantic Corp., the National Association of Securities Dealers Inc. and Oracle Corp. The sixth conference is set for Austin, Texas. Among the sponsors are IBM Corp., Armonk. N.Y., and SBC Communications Inc., San Antonio.There are two kinds of research on Web usability, according to Laskowski.One type looks at a Web site and asks is it usable. Here, the research, generally a smaller study that does not demand statistical significance, only requires a few people to uncover what she calls "major show stoppers," or elements on the site that are counterproductive.The other kind of research looks carefully at how users interact and navigate through any given Web site. Here, the researcher needs enough users and good experimental methods to reduce the number of confounding variables to get at reproducible findings.Among the more interesting of the 11 peer-reviewed papers presented at the conference were one on suggested uses for the back key inside Web browsers and another on a proposed Web event-logging tool.In the paper, "Getting Back to Back: Alternate Behaviors for a Web Browser's Back Button," Saul Greenberg, researcher in the Department of Computer Science at the University of Calgary in Alberta, Canada, highlighted the role of the back key and proposed a new approach based on what he termed a "recency model."Greenberg's research showed that people revisit pages with surprising frequency, with about six of 10 sites visited being seen before.More to the point and supporting his argument for paying attention to the back button, he noted that pressing back accounts for 30 percent to 37 percent of all navigational acts. Furthermore, 43 percent of the 60 percent of pages previously visited are returned to within the last 10 pages viewed."Clearly, revisitation is important for browser design and how you structure Web pages," said Greenberg.In an informal survey of the fairly sophisticated browser-using attendees, Greenberg revealed through a show of hands that very few realize that the back key creates problems in reaching some previously seen pages.As Greenberg said during his presentation: "The typical stack-based behavior underlying Back is problematic because some previously seen pages are not reachable through it. The advantage of recency [model] is that all previously seen pages are now available via Back."The other paper, presented by Michael Etgen of AT&T Labs, Middletown, N.J., was titled, "What Does Getting WET (Web Event-logging Tool) Mean for Web Usability."In proposing the new WET tool and reviewing different approaches to usability data collection on the Web, Etgen pointed out that it has proven to be a challenging task, even as many usability engineers crave a device that makes the process easy, automated and useful.Among the current approaches are hand-coding interactions in person or from videotape, examining the logs saved on servers and running software on the client, or modifying the browser software to allow some automated data collection. A new tool from NIST called WebVIP takes a different approach by copying an entire Web site and adding identifying and event handling code to the HTML links on the site.Etgen claimed that his proposed WET tool overcomes many of the important limitations of the other usability data collection methods by taking advantage of the global event handling capabilities built into browsers produced by Netscape (now part of America Online) and Microsoft Corp.Specifically, WET works on the client side, does not need to be installed on each of the client machines of the usability participants, uses a technique built into the browser so it is not proprietary, works across platforms, and is relatively easy to configure for a site of any size."WET can be placed on a production site. It collects data on a full range of client-side interactions and requires the insertion of a small modular block of code which does not interfere with the site's existing event handling," said Etgen.He quickly added that although WET shows promise, it is still in the early stages of development and requires further testing. He stressed that WET simply collects usage data and does not capture the important aspects of user interaction, such as motivation, frustration and satisfaction, all of which can never be logged by the computer.He also cautioned usability engineers to be practical in their approach to data collection, as tools like WET could gather much more data than necessary or useful."The ideal usage model for WET would be as a method for relieving the usability engineer of the tedious interaction logging during or after usability testing in a lab or on the road," he stated in his co-authored paper.Apropos the nature of the event, proceedings were distributed on CD-ROM in lieu of paper.Also, both presentations and a list of attendees with their organizations and e-mail addresses were made available on the conference Web site (www.nist.gov/hfweb).