Subject: UWired Usability Committee - March 12 Meeting Minutes UWired Usability Committee March 12, 1999 Present: Mark Alway, Rick Ells, Scott Macklin, Judy Ramey, Kathryn Sharpe, Stacey Waters Next Meeting - Friday, April 16, 2-3pm, Room 35F Thompson Hall. Enter Thompson by the south entrance, which is on Padelford side of the building. Accessibility - Scott reported that Sheryl Burgstahler is working on accessibility guidelines: * Helen Remick, Assistant Provost for Equal Opportunity, asked Sheryl to work on the project. * Dan Comden is working on gathering current information. * Should UWired develop a method guide on accessibility for Catalyst? * Accessible designs work for everyone. * Accessibility should be a topic for a future UWired Usability Committee meetinng. Catalyst - Usability test has taken place, reports Judy. A report is due by the end of the quarter. Using site statistics in usability studies * Page Visit Statistics - Rick distributed site statistics for the Teachnology site, including example of records from access_log and referer_log files, wwwstat tabulation of the access_log, and a manually created graphic showing visits to each page on the site and jumps from one page to another within the site. The access_log lists pages connected to and the host the connection came from. The referer_log lists connections to a page from a link on another page, either on the site or elsewhere. Creating the graphic required using data from both the access_log and referer_log. University of Twente has a Java applet that constructs graphics depicting patterns of visits. * Mark and Scott demonstrated the use of Webalizer (http://www.mrunix.net/webalizer), which they are using to analyze visits to the UWired site. Webalizer provides pie chart and bar chart depictions of breakdowns of access_log tabulations. * Drawing Rhetorical Inference - Visit patterns can help you discern if you are expressing concepts in the same terms and categories as your audience. Are you (the designer) thinking in the same terms the users are thinking in? * Collecting Search Terms - Collecting the terms users search for, and then correlating them against patterns of use would give you clusters of terms that get to specific pages. Of all the words on a page, which ones are people searching for that successfully get them to the page they want? The value of search terms is that they come directly from the user's brain and thus give a glimpse of the concept set, conceptual mapping, and vocabulary of the audience. * Correlates - Can visit patterns and search terms be used to profile users, which then could be used to help users more successfully use the site? For example, could a person doing a search get a response that includes a "People like you seem to be interested in ..." list. * Living Web Site - Good visit statistics and search term analysis can make the Web site a negotiated space between visitors and site maintainers. Rhetoric of Web Structures * Sites are built on various rhetorical models o Goal/Action Model - Main page lists goals, click on a goal and you see instructions on actions to reach that goal or actual form to perform the desired action. o List of Lists - List of major topics linking to lists of subtopics, etc. o Linear Step-by-Step Sequence - First do step one, then do step two, then... o Decision Tree - If your age is under thirty, choose riskier investment strategies. If betweent thirty and fifty, be more conservative. If over fifty, avoid unnecessary risks. * Sites are designed to with a particular type of rhetoric in mind. Are visitors actually using the site according to the intended rhetorical model? o UWired wants visitors to Catalyst site to go through method guidelines before they access the technology pages. Uses For Visit Statisitcs * Justification for past expenditures and continuation of funding. Is site meeting needs of intended audience? * Aid in search for new funding. Stats can prove a site is useful to other audiences than originally intended. * Picking up on ways your site is being used. Why are visitors hitting some pages heavily and skipping others you thought would be important? * Are your design features (quick guide, most frequently visited pages, alphabetical list, etc.) helpful to visitors? Intel Customer Support Monitors Use Patterns * Changes in use patterns over life of a product or of a site give feedback on how well things are working. * Watch for... o Surge in connections to specific pages or searches on specific topics o Changes in who is going where within the site o Are visitors finding what they want, is what they find useful? * Monitoring can be passive and active o Active monitoring directly queries users with forms, email windows, pop-up windows, etc. o Passive monitoring tries to infer an understanding of the user from patterns of use, as indicated by log files, cookie tracking, etc. * Nurturing Lurker - Behind a healthy site (one that dynamically evolves with the needs of its users) is a monitoring process continually adjusting the site as circumstances evolve. * Intel uses use patterns for internal business monitoring. o Calls per product shipped is used as a metric on how well designed and supported the product is. o Feedback through help calls and Web page visits guides design - product release is successful if it prompts a smaller number of calls and visits than anticipated. Amounts to a radical redesign of relationship with user - feedback directs design process. o Major value is penetrating developer self-deception and hubris. Ways Use Statistics Can Be Deceiving * Browsers or proxy servers can cache pages - since home site is not visited as often, stats will underrepresent actual use of site content. * Difficult to measure time on page. Only have time of connect to specific page. Even you estimate from times given host visits a sequence of pages, longer times do not mean user actually read page content - they might have just minimized the page and gone to lunch. * A log entry recording a visit to a page does not indicate whether page actually met visitor's needs. * Logs must be carefully evaluated. Access_log file records retrievals of page and all files associated with page, such as graphics. You can't just count the number of entries and declare your site has been visited 15,000 times since Monday - multiple entries are made each time each page is accessed. * Use patterns may be strongly influenced by unrecognized factors. For example, choices on a page may be ignored by users simply because the page is long and the choices do not show when page first appears. What factors of layout, page placement, graphical presentation, vocabulary , and character size might influence visitor actions? |- Rick Ells - 543-2875 - rells@cac.washington.edu - Rm 301 4545 Bldg -| |- http://staff.washington.edu/rells/ -|