CHS:EAGER:Aiding Reasoning about Correlation and Causation

4/1/19-3/31/21

PI: Francesco Cafaro
Clark University
Award Details

People are increasingly exposed to data and datasets in everyday life, in domains from health and science to news and policy. This raises important questions about how to help non-specialists make sense of those data, in particular, around understanding how to think about correlation and causation. These concepts can be slippery; confounding correlation with causation may lead people to assume causality when there is none, but correlations do often provide precious hints to causation. This project will investigate how theories of cognition that emphasize the relationship between thinking and physical action can be used to design full-body and tangible ways to interact with data-based museum installations that prime people to think in ways that improve their understanding of causation and correlation. The goal is to develop bridges between theories of embodied cognition, the design of data visualizations, and long-term learning effects about science, technology, engineering, and mathematics concepts discussed in the installations. The project will be deployed in real contexts, having direct potential impacts on visitors’ understanding, and will be used to inform educational curricula in ubiquitous computing and design for informal learning.

The project is organized in four phases that will be conducted at Discovery Place, a science museum in Charlotte, NC. In part one, the team will design visualizations of geo-referenced datasets on a wall-size projected screen. Using a semi-experimental design, groups of visitors will interact with one of several variations of the installation, including full-body and tangible interaction styles based on different physical metaphors for correlation as well as a tablet-based control condition. In part two, the team will experiment with different styles of data visualization (e.g., line charts or heat maps), and in part three visitors will be asked personalize the dataset on display; these extensions are necessary to assess the generalizability of the results from phase one to different data presentations and domains. Part four addresses transferability of learning across time and to other contexts, following up with museum visitors weeks or months after their visit and asking them to evaluate the likely correctness of data-based claims about correlation and causation in science articles in domains such as health remedies.

This award reflects NSF’s statutory mission and has been deemed worthy of support through evaluation using the Foundation’s intellectual merit and broader impacts review criteria.

Tags: ,