Proc. of INTERACT 2005, Springer, Berlin (861-872).
Springer Link: http://dx.doi.org/10.1007/11555261_68
HCIL-2005-05, CS-TR-4709, UMIACS-TR-2005-17
Most visualization tools fail to provide support for missing data. We identify sources of missing, and categorize data visualization techniques based on the impact missing data have on the display: region dependent, attribute dependent, and neighbor dependent. We then report on a user study with 30 participants that compared three design variants. A between-subject graph interpretation study provides strong evidence for the need of indicating the presence of missing information, and some direction for addressing the problem.