Abstract
In visual querying, users analyze data for their decisions and problems by interacting with graphics that are dynamic and linked. This querying paradigm is new, a dramatic break from the more familiar retrieving of data via search statements and displaying of it in static charts and graphs. For this new visual querying paradigm, analysts conceptually and operationally need to master new approaches. To discover salient relationships, they need to manipulate displays. To drill down for detail or causes, they have to select data of interest directly from a graph. And to draw inferences, they have to consider meanings across several dynamically linked graphics. With the aim of studying users' success in these new approaches, particularly focusing on the approach of directly selecting data from graphs, I conducted a scenario-based usability test with 10 data analysts. They interacted with visualizations to complete a realistic complex analysis—evaluating employee performance. Test findings reveal a range of difficulties in visual selection that, at times, gave rise to inaccurate selections, invalid conclusions, and misguided decisions. To overcome these difficulties, support for visual selection needs to be built into interfaces and help. Results and recommended improvements are presented.
Get full access to this article
View all access options for this article.