News about COVID-19 often comes packaged with charts, maps and graphs that help the public understand the state of the pandemic and to justify policies around mask-wearing and social distancing. Such visual representations can succinctly reveal critical information from vast datasets and they are essential for communicating scientific findings to the general public and policymakers.聽
However, the way data is presented can influence its interpretation. 鈥淗ow data is represented can introduce bias, dramatically changing the conclusions drawn and ultimately affecting policy-making and other important decisions,鈥� says Danielle Szafir, an assistant professor of computer science with the ATLAS Institute. As director of the VisuaLab, Szafir researches what scientists, policymakers and the general public take away from such visualizations and whether their conclusions reflect what the data actually means.聽
Recognizing the value of this research, the National Science Foundation recently awarded Szafir a CAREER award for a project titled, 鈥淒eveloping Perceptually-Driven Tools for Estimating Visualization Effectiveness.鈥� An NSF CAREER award is one of the most prestigious given to faculty in the early phases of their careers. For Szafir, the grant provides $550,000 over five years to support research and outreach activities.聽
The award will allow Szafir to fill a gap in data visualization research. Past visualization studies have demonstrated which types of charts perform well for specific tasks, but there isn鈥檛 yet a concrete way to rapidly gauge the efficacy of different types of visualizations. Their goal聽is to offer designers automated solutions for rapidly estimating visualization effectiveness, including revealing what different types of visualizations fail to communicate or communicate incorrectly. By establishing a set of effective visualization design practices that are universally accessible, they will聽help designers make聽better choices that minimize misleading data representations and make data exploration more efficient and easier.聽
Another key goal for the initiative is to develop a curriculum for a Coursera Massive Online Open Course (MOOC) for an online Master聽of Data Science program to help students from a broad set of fields develop essential data visualization skills.聽
Befitting an ATLAS research lab, VisuaLab's work is highly interdisciplinary, touching on cognition, perception and the most advanced applications of visual technologies, including virtual reality (VR) and augmented reality (AR). To help connect researchers specializing in these various fields, Szafir cofounded VisXVision, an interdisciplinary initiative aimed at聽promoting collaboration between data science and cognitive science.聽
鈥淲e have seen an unprecedented increase in public communication using data,鈥� Szafir says. 鈥淏y聽offering designers a means to readily understand what patterns people will see in a visualization, we can rapidly improve the ways we use data.鈥�
Commenting on Szafir's research, ATLAS Director Mark Gross says, "It's hard to overstate the potential of Danielle's work. Inventing tools that make it easier to communicate scientific findings can make us all more informed decision-makers."
Szafir鈥檚 CAREER award (NSF 2046725) is funded by the NSF鈥檚 Directorate for Computer and Information Science and Engineering Faculty Early Career Development Program.聽