ABSTRACT

Interactive graphs, charts, and other visual representations of data are increasingly common in public life. People bring their own expectations and assumptions about the data and situations these visualizations represent, though most visualizations do not take these expectations into account. Comparing these expectations to actual data is a powerful tool for checking those assumptions, developing better understanding of situations, and making better decisions. To support such expectation visualizations, the project team will use a combination of experiments, software development, and design activities to develop toolkits and best practices for developing visualizations that allow viewers to represent, interact with, and see feedback on their own predictions about the data. The work will focus on helping people better understand scientific research and expert analysis around topics such as health decisions that might impact their own lives. The work will also support a broader educational goal of data literacy education, through course modules that can be inserted into introductory informatics and data science courses and a new course on thinking with data, and outreach goals through developing a research and development platform where designers, researchers, and developers can work together to improve expectation visualization techniques.

To do this, the project has three main research goals. The first is to develop a suite of empirical findings on the effects of expectation visualization, through a series of experiments on how predicting data, receiving personalized feedback on those predictions, and reflecting on gaps between predictions and data affect people's later memory of the data and future expectations. The second thrust builds on the first, using these empirical results along with design studies and comprehensive reviews of existing tools and literature to build a design space with software examples characterizing key decisions in designing expectation visualizations. These decisions will include a range of techniques for graphically eliciting people's expectations, contextualization techniques that help people learn to use those techniques and constrain their choices appropriately, and feedback or reflection techniques that help call attention to places where expectations did and did not match the underlying data. The third thrust is to put these principles into practice by developing applications to support the communication of uncertainty in experimental results, the reduction of spurious pattern discoveries in data analysis, and the integration of problem context and expert analysis with the visualization itself.

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.

Please report errors in award information by writing to: awardsearch@nsf.gov.