Measuring usability: Categorically modeling successful Websites using established metrics

Although website usability continues to be a hot research topic and sets of usability guidelines abound, many websites are still plagued with severe usability problems. Many owners/designers cannot afford the expense and effort of a thorough professional website usability evaluation. It is difficult for a novice developer to digest the glut of available guidelines and guideline sets in order to determine which particular guidelines to include when designing a usable website in a specific category. Furthermore, guidelines are often too vague to be implemented properly. The time from inception to completion of a website is relatively short (as compared to a software package) and as most websites are designed by amateurs, an economical and rapid usability assessment of their websites is necessary. Many of the currently available automated website rating systems are based on criteria that is often subjective and not based on “usability” per se. Using a traditional website categorization that includes e-commerce, informational, interactive and portal, this dissertation employs datamining algorithms to develop predictive, categorized models of an ideal website, based on an established set of usability metrics in order to enable the novice—the individual or small business owner—to design better websites. It establishes an order of relevance for a quagmire of usability metrics and their associated guidelines. It also serves to validate or refute previous research. It introduces a framework illustrating the relative importance of guideline metrics gathered from a selection of automated web usability tools. Finally, a more user-centered categorization is proposed which may serve as an impetus for future research. While acknowledging the importance of user interface quality in HCI, web user interfaces are differentiated from desktop GUIs. The focus of this dissertation will be strictly usability, using guidelines established specifically for websites, as opposed to graphical user interface guidelines that may or may not apply to websites