The Webtango Method

Abstract

Chapter 4 described briefly the WebTango automated evaluation approach [Ivory, 2001; Ivory et al., 2000; Ivory et al., 2001; Ivory and Hearst, 2002a], which we elaborate on in this chapter. We designed WebTango to leverage the benefits of both performance evaluation [Jain, 1991; Sauer and Chandy, 1981] and usability evaluation to evaluate the quality of sites designed primarily to disseminate information (i.e., informational sites). Performance evaluation is used predominately in the assessment of computer systems, and usability evaluation is used for assessing interfaces. In the spirit of measurement techniques in the performance evaluation domain, benchmarking in particular, the methodology entails computing an extensive set of quantitative page-level and site-level measures. These measures can be viewed as a benchmark because they enable objective comparison of web interfaces. Specifically, we use these measures to derive statistical models of highly rated interfaces. As is done with guideline review methods in the usability evaluation domain (see Chapter 4), the models are then used in the automated analysis of web pages and sites. Unlike other guideline review methods, the guidelines are in essence derived from empirical data. Hence, it is also possible to use the models to validate or invalidate many guidelines proposed in the web design literature.