PhD Thesis Defense: Zack While, Broadening Data Visualization Design to People in Late Adulthood
Content
Speaker
Abstract
Data visualizations have experienced increased ubiquity in daily life, visually communicating data related to various domains such as weather, health, and exercise. However, the vast majority of existing empirical visualization design knowledge has been built using participant pools that skew to a younger demographic (e.g., ages 18-30). This leads to a knowledge gap regarding how visualization design may differ for people in the late adulthood stage of development (PLA), often described as beginning around age 60 or 65. Furthermore, demographic trends are shifting toward an aging global population, with PLA (ages 65 or older) expected to outnumber children (ages 0-14) starting sometime between 2050 and 2075. Thus, it is of great importance and relevance to the visualization community to understand how various changes associated with aging (e.g., changes in perception) can impact design considerations.
First, I help establish a solid foundation for this new research subfield, called GerontoVis. This includes arguing for the importance of expanding visualization research to better-include PLA, discussing how various physiological changes due to aging may affect visual analysis performance, and how adjacent bodies of work in HCI are, on their own, insufficient for supporting PLA in a visualization context. I also discuss challenges and opportunities for work in GerontoVis that can expand on and broaden this growing area of research.
Next, I contribute empirical baselines for understanding the impact of aging on visual analysis performance. This includes evaluating the impact of contrast polarity (i.e., using dark mode or light mode) on the visualization performance of both younger adults (YA) as well as PLA (age 60+), which found similar impacts across both age groups and a mismatch between participants' preferred polarity and their best-performing one. Additionally, I investigated how aging interacts with the choice of low-level visualization task (e.g., finding extreme values) and visualization type (e.g., bar charts). I observed that PLA showed greater heterogeneity in task performance as well as differences from YA in best-performing visualization choices for some tasks.
Lastly, I explore design for PLA in a more application-focused setting. In one study, I analyzed the perceptual speed of PLA using glanceable visualizations on smartwatches. I observed that differences in speed grew with increasing age, with PLA ages 75+ requiring more time to accurately make data comparisons than those ages 65-74. In another study conducted via in-person, semi-structured interviews, I examined the experiences of PLA using embedded interactive information displays (EIIDs) that communicate information relevant to home devices (e.g., microwaves, stovetops, and dishwashers) using a variety of data representations such as lights and LED screens with numbers and icons. Observations of design choices that bottleneck and facilitate PLA's ability to execute information-guided device operations, as well as the identification of habits and compensatory strategies used by PLA, led to design recommendations for both EIIDs and interactive visualizations. used by PLA, led to design recommendations for both EIIDs and interactive visualizations.