Visualizing Uncertainty in Assessment and Institutional Data

Session Abstract: In IR/IE contexts, data visualization is essential for decision support and departmental planning. Visualizations often neglect measures of uncertainty (e.g., confidence intervals); yet in smaller sample contexts, this is a crucial element of decision-making, with implications for action steps and resource allocation. To model ways of visualizing uncertainty, this poster presentation outlines practices based on theories of visual information processing. Specifically, it reviews research on gradient plots, quantile dot plots, and encoding uncertainty using visual features (e.g., blurriness). The presentation concludes with a template for visualizing rubric results developed by UNC Wilmington’s Department of Student Affairs Assessment, Research, and Planning, which uses quantile dot plots to represent 90% confidence intervals for estimated proficiency rates on rubrics––thus allowing Student Affairs practitioners to make better calibrated decisions for improvement based on evidence of student learning.

Keywords: data visualization, institutional effectiveness, assessment, data-informed decision-making