Skip to main content

SILS Professor David Gotz Receives NSF Grant for Study on Data Visualizations and Causal Inferences

In August, SILS McColl Term Professor David Gotz received a $1.2 million grant from the National Science Foundation (NSF) to conduct research on improving inferences that people make from data visualization.

Professor David GotzResearchers from the Visual Analysis and Communication Laboratory (VACLab), a research lab housed at the UNC School of Information and Library Science (SILS) and led by Gotz, will launch a study to identify the ways in which current data visualization methods can lead people to interpret visualized patterns as indicators of causal relationships between visualized variables. People often make inferences of causality based on visualizations even when such conclusions are not supported by the data. This can lead people to draw the wrong conclusions even when visualizations are used to accurately represent the underlying data.

“People often look at visualizations of data and assume the relationship they’re looking at is somehow causal,” Gotz said. “However, current data visualizations can mislead users into drawing causal inferences because they often fail to communicate many of the interactions that exist between explanatory variables and their effects.”

As part of this research study, the lab will work to develop a new approach to visualization based on the concept of counterfactual reasoning – a central pillar in casual analysis – that is designed to help users draw more robust and generalizable inferences.

“While studies have shown that users often interpret visualizations in a causal way, very few visualizations actually incorporate the underlying statistics that can reliably support causal inference,” Gotz said. “So, our study is focused on trying to understand exactly how people make these causal inferences, and to develop new approaches to visualization that can help make the causal inferences that people make more accurate.”

In addition to helping advance our understanding of causal inference within the context of data visualization, the research also aims to produce open-source software tools that prototype the new approaches that will be developed over the course of the project. The researchers also aim to evaluate their work within the context of large-scale population health projects – offering potential for societal impact through improved health outcomes.

To learn more and keep up with the project and VACLab, please visit