Supporting Non-Experts' Awareness of Uncertainty: Negative Effects of Simple Visualizations in Multiple Views

Video analysis tools can provide valuable datasets for a wide range of applications, such as monitoring animal populations for ecology research, while reducing human efforts for collecting information. Transferring such technology to novel application domains implies exposing non-expert users to unfamiliar datasets and technical concepts. Existing data analysis practices must adapt to the new data characteristics and technical constraints. With such changes, uncertainty is of major concern as it can yield misinterpretation of data, or distrust and rejection of valid results. We present a study of an interactive visualization of computer vision results and uncertainty. We evaluate the correctness of users' interpretation of data, and their confidence in their interpretation. We compare the impact of either data features (i.e., the true level of uncertainty) or visualization features on user perception of uncertainty. Visualization features had a similar impact on user responses than the data uncertainty itself, thus biasing user awareness of uncertainty. We conclude with the opportunities (intuitive navigation in complex unfamiliar data) and limitations (poor extrapolation and memory loss) of our visualization design which integrates simple graphs in coordinated multiple views. Our design and insights contribute to other cases where non-experts need to familiarize with novel datasets and explore their uncertainty.