Screen or No Screen? Lessons Learnt from a Real-World Deployment Study of Using Voice Assistants With and Without Touchscreen for Older Adults
Zhenyu Chen, Yijun Lin, Jiaxin Li, Qiuchen Qian, Mingzhe Li, Jikun Peng, Yuhang Zhao · 2023 · ASSETS 2023: The 25th International ACM SIGACCESS Conference on Computers and Accessibility · doi:10.1145/3597638.3608378
Summary
This paper presents a 40-day within-subjects deployment study examining how older adults interact with voice assistants (VAs) with and without touchscreen capability. Sixteen participants (mean age 82.5 years), all living independently, used an Amazon Echo Dot (voice-only) and Echo Show (voice plus touchscreen) in their homes for 20 days each. The research addresses a gap in understanding how touchscreen augmentation affects real-world VA usage among older adults, who represent a growing demographic that could benefit significantly from voice-based technology for independent living support. The study employed a mixed-methods approach combining device usage logs, daily diary surveys, pre- and post-study questionnaires (including SUS and NASA-TLX), and semi-structured interviews. Participants completed structured daily tasks including weather checks, timer setting, and entertainment requests, alongside unstructured free exploration. The researchers tracked setup time, response latency, interaction patterns, and subjective experiences across both device conditions. The study is notable for its ecological validity, deploying devices in participants' actual homes rather than controlled lab settings, and for its longitudinal design that allowed observation of learning curves and adoption patterns over time. The research team provided initial setup support but deliberately minimized ongoing assistance to capture authentic usage behaviors.
Key findings
The touchscreen-equipped Echo Show reduced initial setup time by approximately 50 percent compared to the voice-only Echo Dot, as visual feedback helped older adults confirm their commands were understood. Response latency was roughly 20 percent lower with the Echo Show, though participants used touch input only about 27 percent of the time, strongly preferring voice as their primary input modality while appreciating visual output as confirmation. The Echo Show achieved higher System Usability Scale scores (averaging 75 percent) and lower NASA Task Load Index scores than the Echo Dot, indicating better perceived usability and lower cognitive demand. Participants reported that the screen was most valuable for weather information, video content, and as a photo display, but several noted it could be distracting. Error recovery was notably easier with the touchscreen device, as participants could see and correct misrecognized commands. The study identified a learning curve of approximately 7-10 days before participants felt comfortable with either device, with the touchscreen device showing a slightly faster adaptation period.
Relevance
This research offers practical guidance for designing voice-based assistive technology for aging populations. The finding that older adults prefer voice input but value visual output suggests a multimodal design approach is optimal, rather than voice-only or screen-dependent interfaces. The six design implications identified, including simplified setup procedures, consistent wake-word responsiveness, and age-appropriate default volume levels, provide actionable guidelines for developers creating senior-friendly voice interfaces. For accessibility practitioners, the study highlights that voice assistants can serve as effective assistive technology for older adults with motor, visual, or cognitive limitations, but that thoughtful design is needed to address challenges like speech recognition accuracy with aged voices and intuitive error recovery. The relatively small sample size and focus on a specific demographic context should be considered when generalizing these findings.
Tags: voice assistants · older adults · smart speakers · smart displays · touchscreen · aging in place · deployment study · multimodal interaction · usability
Standards referenced: SUS · NASA-TLX