From web designer to researcher: Using experiment patterns to research web design decisions

Web designers are increasingly expected to justify design with empirical evidence. While laboratory-based usability testing helps identify some problems, it is less successful in predicting how sites are used in the wild. Laboratory tests on informational websites can miss subtle effects of writing style, navigation, and graphic design on users' perceptions and performance. During recent years, tools that support automated, remote data collection of web users' behavior on live websites have come to market. These tools typically measure click-paths that lead users to specific landing pages, and do not apply to many scenarios that involve the use of informational websites. We are developing a software toolkit, WebLabUX, that helps web designers perform online experiments and measure the effects of web design variables on users' behavior, site perceptions, and comprehension. One significant challenge we have faced is the creation of a user interface that embodies best practices in online experiment design such that people without formal training in experiment design can successfully use the toolkit. This paper discusses a set of study patterns derived from Redish and Mirel's taxonomy of actions of users when working with information, and we discuss how these patterns can be incorporated into the user interface of WebLabUX.