Glanceable Data Visualizations for Older Adults: Establishing Thresholds and Examining Disparities Between Age Groups

TL;DR

Strong evidence of differences for participants aged 75 and older is observed, sparking interesting questions regarding the study of visualization and older adults, and first steps toward better understanding and supporting an older population of smartwatch wearers are discussed.

Abstract

We present results of a replication study on smartwatch visualizations with adults aged 65 and older. The older adult population is rising globally, coinciding with their increasing interest in using small wearable devices, such as smartwatches, to track and view data. Smartwatches, however, pose challenges to this population: fonts and visualizations are often small and meant to be seen at a glance. How concise design on smartwatches interacts with aging-related changes in perception and cognition, however, is not well understood. We replicate a study that investigated how visualization type and number of data points affect glanceable perception. We observe strong evidence of differences for participants aged 75 and older, sparking interesting questions regarding the study of visualization and older adults. We discuss first steps toward better understanding and supporting an older population of smartwatch wearers and reflect on our experiences working with this population. Supplementary materials are available at https://osf.io/7x4hq/.

Figures (10)

• Figure 1: Study setup and apparatus: (a) An example participant during the study. (b) A participant's point of view. (c) The keyboard used at the start of the study, arrow keys emphasized with a red outline (top) and the two separate buttons (bottom) added after P6.
• Figure 2: An example of the staircase method, showing P24 performance in the Donut12 condition. Each circle or triangle glyph is a trial the participant performed. The color orange-red marks errors while triangle shapes mark reversals, i. e., trials when a participant switched from up (most recently increased exposure time) on the staircase to down (decreasing exposure time), or vice-versa. Reversal points R1-R15 are labeled. To compute the time threshold, the time-per-stimulus for the reversal points is averaged, starting with the third reversal (R3). The computed threshold in this example is 238 milliseconds. The minimum exposure time was 100 milliseconds.
• Figure 3: Overall structure of the study. Each participant started and ended their study with a questionnaire. In this example, each chart type presents 7 data items for a staircase, then 12, and then 24; for each participant, this ordering was based on the Latin square design. Each staircase comprises 10 practice trials, and then the staircase, with the number of trials in the staircase depending on performance. After each chart type, we asked participants to describe strategies they used for that set.
• Figure 4: Steps involved in an individual trial. A participant is first shown the stimulus (a chart) for an amount of time, which the current staircase determined. Then, they are shown four intervening images for 20 ms each. Next, the screen prompts the participant to either press the left or right arrow key, after which the participant is told whether their input was correct or not.
• Figure 5: High-level depiction of the study process and progression. We started with four pilot studies and proceeded to the main study with 24 participants. Each box in the graph represents a participant and their age. Colors indicate if that participant was in the young-old (age 65-74) or old-old (age $\geq 75$) group. After the second study (P2), participants were explicitly told at the start that they could skip the Radial24 condition if they were experiencing excessive fatigue, discomfort, or difficulty. Seven of the 22 remaining participants (marked with a ) chose to skip that condition. After the sixth study (P6), we swapped the keyboard with two separate physical keys for greater ergonomic comfort during the study (shown in \ref{['fig:user-periph']}).