Two New Journal Articles and More from VACLab Researchers at IEEE VIS 2024

The 2024 IEEE VIS Conference will be held in Florida this October. As the world’s premier international scientific conference focused on visualization and visual analytics, the IEEE VIS gathers researchers from around the world and showcases a variety of peer-reviewed publications, presentations, and other events to designed share the communitiy’s latest research results.

VACLab researchers will be active participants in this year’s confernece with two new journal articles, a pair of posters, and an oral presentation of our recently published IEEE CG&A article.

Our first new paper, “Causal Priors and Their Influence on Judgements of Causality in Visualized Data”), examines how people draw causal inferences from traditional visualizations (e.g., scatterplots, bar charts, and line charts) and how those inferences are influenced by names of the variables displayed in the chart. More specifically, we find that the causal association people assume a priori based only on the words used to label variables (which we term a causal prior) has a signficant effect on how people draw inferences from the underlying visualization. This paper will be presented during IEEE VIS 2024 and will appear in a 2025 issue of IEEE TVCG.

Our second new paper, “Beyond Correlation: Incorporating Counterfactual Guidance to Better Support Exploratory Visual Analysis”), proposes a new way to guide exploratory analysis based on counterfactual statistics. Our evaluation of this new approach shows that counterfactual-based guidance, in comparison to traditional correlation-based approaches, can improve how users perform visual causal inference and can shift users’ patterns of exploration. This paper will be presented during IEEE VIS 2024 and will appear in a 2025 issue of IEEE TVCG.

In addition to the two papers mentioned above, VACLab researchers will give an oral presentation to describe our previously published IEEE CG&A article titled “Using Counterfactuals to Improve Causal Inferences from Visualizations." This presentation will share our vision for how counterfactual concepts developed by the causal inference community can be leveraged to enable the development of more effective visualization technologies.

Finally, VACLab-ers will also share two research posters describing two other research contributions. First, we report a short study that examined the similarity and differences between LLM and human judgements of causal relationships from visualizated data. Second, we describe our recently released software tool, CausalSynth, which allows users to generate and visually inspect synthetic data with known causal relationships. Our lab uses this open-source tool to create data with known causal ground truths for use in our own research and prototype development.

All of the work to be presented at IEEE VIS 2024 has been made possible in part by the National Science Foundation under Award #2211845.