Elderly Text-Entry Performance on Touchscreens
Hugo Nicolau, Joaquim Jorge · 2012 · Proceedings of the 14th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '12) · doi:10.1145/2384916.2384939
Summary
This paper presents an empirical study of how elderly users perform text-entry tasks on touchscreen devices, examining both mobile phones and tablets. Fifteen participants aged 67 to 89 (mean 79) were asked to copy sentences on an HTC Desire smartphone and an ASUS Transformer tablet using standard QWERTY virtual keyboards with no error correction or word prediction enabled. The researchers measured typing speed, accuracy using Minimum String Distance error rate, and categorized errors into three types: insertions (extra characters from accidental or bouncing touches), substitutions (wrong characters from poor aiming), and omissions (missing characters largely attributed to cognitive errors). Hand tremor was assessed through both clinical Archimedes spiral drawings and accelerometer measurements capturing hand oscillation across multiple axes and frequency ranges. The study was motivated by the observation that while touchscreens are increasingly replacing physical keypads in everyday devices like ATMs, kiosks, and mobile phones, the lack of tactile feedback and physical stability creates particular challenges for elderly users with declining motor abilities. The research aimed to produce empirical data to inform the design of more accessible virtual keyboards for this population.
Key findings
Elderly participants typed at an average of 4.7 WPM on mobile and 5.1 WPM on tablet, with the fastest reaching 11.5 WPM on tablet. Tablets reduced the overall MSD error rate by approximately 9% compared to mobile phones (16.6% vs. 26.0%), a statistically significant improvement attributed to larger key sizes and the device resting on a stable surface. Omissions were the most common error type (9-12.6% of keystrokes) and were consistent across both devices, suggesting cognitive rather than motor causes — participants forgot characters, lost their place in sentences, or struggled with the concept of the space key. Substitutions (3.8-7.8%) were strongly correlated with task-specific tremor measured through Archimedes spiral drawings, while insertion errors were linked to hand oscillation causing bouncing and accidental touches. A right-bottom substitution pattern emerged, likely related to hand dominance. Simple touch models showed promise: an inter-key time threshold removed 30-50% of insertion errors, and a key-centroid reclassification reduced substitution error rates by 9.8-11.5% on mobile. Critically, personalized (user-dependent) models significantly outperformed user-independent ones, indicating that keyboard adaptations should account for individual tremor profiles. 86.7% of participants preferred the tablet device.
Relevance
This study provides foundational empirical data for designing accessible touchscreen text-entry interfaces for older adults — a rapidly growing user population. The finding that different error types correlate with different tremor features has direct design implications: insertion errors can be filtered with timing thresholds, substitutions can be reduced with personalized key-centroid models, but omissions require cognitive support strategies rather than motor compensation. The strong case for personalization over one-size-fits-all approaches is particularly relevant for modern adaptive keyboard design. Practitioners building virtual keyboards for aging populations should consider larger key targets, stable device positioning, individual tremor-profile calibration, and cognitive scaffolding such as progress indicators. The work also highlights that prior QWERTY experience correlates with speed but not accuracy, meaning that the physical challenges of touchscreen interaction cannot be overcome through familiarity alone.
Tags: touchscreen accessibility · text entry · aging · motor accessibility · tremor · mobile accessibility · virtual keyboard · older adults · input methods