A Study of Input Device Manipulation Difficulties
Shari Trewin · 1996 · Proceedings of the Second Annual ACM Conference on Assistive Technologies (Assets '96) · doi:10.1145/228347.228351
Summary
This paper from the University of Edinburgh describes a pilot study designed to gather detailed, quantified data about physical input errors made by people with motor disabilities when using standard keyboards and mice. The author identifies a significant gap: despite widespread recognition that people with motor disabilities make unintentional input errors, virtually no quantified data existed on the nature, frequency, and patterns of these errors. Such data is essential for developing automatic recognition of specific difficulties and dynamically configuring systems to accommodate individual users. The study used an InputLogger program that recorded detailed timestamped logs of all keyboard and mouse events, combined with video recording and observer notes. Four volunteer adults with motor disabilities participated, completing typing tasks (word processing passages) and mouse tasks (pointing, clicking, double-clicking, dragging). The subjects had diverse disabilities: two had motor difficulties caused by muscle wastage and spasticity, one had neurological damage, and one had impaired dexterity due to incomplete left hand function. The study design deliberately captured normal performance patterns for each user first, so that errors could be identified by comparing actual input against expected input for the task.
Key findings
The study identified six types of keyboard performance errors: additional key errors (pressing an adjacent key alongside the intended one), doubling errors (holding a key too long, triggering key repeat), missing key errors (failing to press a key at all), dropping errors (releasing a modifier key like Shift before pressing the modified key), remote key errors (pressing a completely wrong key), and transposition errors (swapping two keys). Each subject had a very different error profile — individuals tended to be prone to one or two specific error types rather than making all types equally. The most common error was additional key presses (type "ad"), where timing analysis revealed that accidental presses overlapped with the intended key in 83% of cases, with the intended key being the last raised. This distinctive timing pattern makes automatic detection feasible. Doubling errors were easily detectable from key-down/key-up timing and could be eliminated by extending the key repeat delay. Dropping errors showed a pattern where the Shift key was released no more than 2 ticks before the modified key — recognisable from log data. For mouse usage, six error types were identified: click length errors, timing between clicks errors, click movement (moving while clicking), positioning errors, dragging difficulties, and dropping the mouse button during drags. Dragging and hierarchical menu selection were the most difficult mouse tasks. One subject showed a consistent diagonal movement pattern while clicking, suggesting that click acceptance zones could be customized per user.
Relevance
This paper laid groundwork for what became a significant research area: automatic adaptation of input device settings based on observed user behavior. The taxonomy of keyboard and mouse performance errors developed here has been widely influential and directly informed the design of accessibility features now standard in operating systems — StickyKeys (addressing dropping errors), FilterKeys/SlowKeys (addressing additional key and doubling errors), and enhanced pointer precision settings (addressing click movement errors). Shari Trewin went on to become a leading researcher at IBM in this area, extending this work into automatic keyboard and mouse configuration tools. For practitioners, the paper provides a systematic framework for understanding why a user with motor disabilities struggles with input and what specific configuration changes would help. The finding that each user has a distinct error profile underscores that accessibility settings must be individualized rather than one-size-fits-all. The methodology of detailed input logging combined with video analysis remains the gold standard for understanding input device difficulties.
Tags: motor disability · keyboard accessibility · mouse accessibility · input errors · performance errors · input logging · assistive technology · user research · accessibility testing · personalization