| dc.description.abstract | Multiple-view visualization is a type of data visualization that includes multiple charts (views) in one screen. One of the key goals of multiple-view visualization is to support the investigation of a topic from different perspectives. In a reasoning task, the design of the multiple views should support viewers in linking information from more than one view. My dissertation investigated how goal-directed strategies such as spatial alignment affect people’s reasoning with visualizations. One of the experiments prompts viewers to think about two scatterplots about cars: one view depicts the origin of the cars, and the other view depicts how many cylinders the cars have. The two views depict the same attributes of cars on the axes: the y-axes depict the miles per gallon of the cars, and the x-axes depict the horsepower of the cars. The goal was to ask viewers to make inferences about cars between 10-30 miles per gallon. The two views were aligned if they were horizontally arranged: The ticks in the axes depicting the goal-relevant attribute (miles per gallon) were directly aligned. When the two views were vertically arranged, the axes depicting the goal-relevant attribute (miles per gallon) were impeded. I hypothesized that viewers who saw the two views aligned made inferences more easily based on information from both views in a reasoning task than viewers who saw the impeded version. Five experiments demonstrate that direct alignment consistently increases viewers’ confidence about their inference ratings across different types of visualizations. Results also suggest that direct alignment encourages viewers to consider goal-relevant views in a multiple-view visualization. This project is the first attempt at studying design strategies of multiple-view visualizations in psychology research. In applied settings, this research provides insights into how arrangements of views can affect viewers’ inferences and their experience of making inferences based on multiple-view visualizations. | |
