Abstract
This study investigated the impacts of three sorting techniques on various cognitive tasks performed on a tabular representation. The tasks under study were a multiattribute object selection task and selected low-level analytic tasks. Three sorting techniques, including sorting by a column (Typical Sort), sorting by all columns simultaneously (SimulSort), and sorting by all columns with faithful vertical locations (ParallelTable), were compared with a static table without the sorting feature (Baseline). An incentivized controlled laboratory study with 80 participants and a preliminary eye-tracker study were conducted to better understand the strengths and weaknesses of the four different approaches. SimulSort and ParallelTable were found to significantly improve the performance of multiattribute object selection. ParallelTable, however, suffers from an occlusion problem, so it is not an appropriate support for some low-level analytic tasks. The findings were used to propose appropriate sorting techniques for specific tasks performed on a table.
Acknowledgments
This work has been partially funded by the U.S. National Science Foundation (Grant No. 09064963), and we also thank the students who participated in this study.
Notes
1We asked the rank of a certain value in a column instead of asking a shape of distribution because asking a shape of distribution became an open-ended question, which is difficult to be quantitatively evaluated.
2We also employed various transformation techniques (logarithmic, square-root, and box-cox transformations) to alleviate the skewness of the data, but these transformations did not reveal any new findings.
3We asked, “How confident are you that you made the best choices in this round?” Participants answered using a 7-point Likert scale from very confident to not at all confident.