Subjective evaluation of perception of accuracy in visualization of data

This paper focuses on the human's perception of information quality and describes the results of a study on how accuracy is estimated for data shown through a visual representation. The subjective assessment of quality appears to be non-linear in relation to the actual degree of errors in the dataset. Users are sometimes unable to distinguish between datasets with different quality, and their ability to estimate is better for certain quality levels than for others. The study also shows that adding complementary information does not always help users to better assess the accuracy of the visualization, and thus of the data. The implication of these results is that, for subjective measures of quality, traditional statistical methods of assessing quality may need to be extended with additional methods to account for the non-linearity and the behavior of data integration.