260 likes | 363 Views
Dynamic Double-Click Adjustment with the ACCESS Framework. Michael Heron. Introduction. Modern computer systems come packed with accessibility options. And they are usually pretty good. Unfortunately: Older people often don’t know what can be changed
E N D
Dynamic Double-Click Adjustment with the ACCESS Framework Michael Heron
Introduction • Modern computer systems come packed with accessibility options. • And they are usually pretty good. • Unfortunately: • Older people often don’t know what can be changed • Older people often don’t know how to change things • Older people often lack the confidence to make changes. • In some cases, people may be physically unable to make the needed changes.
Accessibility and the Older Adult • The issue is further complicated by the nature of the aging process. • Physical and cognitive faculties change at different rates for different people • Accessibility issues may stem from the interrelationship of subtle ailments • Individuals may not consider these ailments in themselves to be sufficient to report as a disability • The ‘dynamic diversity’ of adults makes it difficult to prescribe accessibility.
Accessibility and the Older Adult • There are then two categories of problems that must be addressed. • The difficulty of making changes • Identifying the need for accessibility • In this talk, I will describe the ACCESS framework. • This is an open source accessibility framework designed to resolve these issues.
Developing Accessibility Software • In addition to the end-user issues associated with accessibility, there are issues for research developers. • Accessibility software is usually difficult to write. • It has to link in to very low level O/S processes • Research software is most often goal-oriented • It’s to gather experimental evidence • Cross-platform support is usually neglected. • Focusing on where the problems are. • Tools are usually ‘develop then forget’ • And often unavailable because of IP considerations.
The ACCESS Framework • The ACCESS framework has been written to address these problems for users and developers. • The framework provides low level interaction with the operating system. • And has support for cross-platform deployment. • It also provides a plug-in engine. • And it is the plug-ins that handle actual accessibility support.
The ACCESS Framework • The brief of the grant that funded this research was to support older users in the workplace. • As such, the research has focused primarily on this age group. • It resolves the issues of knowledge and confidence by inverting the locus of control. • Plug-ins algorithmically detect the need for changes • Plug-ins then make those changes • The system is designed to run invisibly in the background until changes are made.
The ACCESS Framework • The framework parcels out I/O information onto the plug-ins. • The plug-ins analyse this data • Every so often, the framework ticks and polls each of the plug-ins. • Plug-ins that believe they can make corrections are placed into a weighted roulette wheel. • This wheel is spun, and the plug-in it selects is permitted to make a change to the system.
The ACCESS Framework • Users can express their consent or dissent with regards to a correction. • When a change is made, a dialog box is presented outlining the change. • The user is asked to ‘like’ or ‘dislike’ the change. • Dislike means: • The change is undone • The weighting of the plug-in in future roulette wheels is lowered • Like means • The change is committed to the system • The weighting of the plug-in is increased.
Correction • Plug-ins themselves get the direction of correction passed on. • So they can adjust their own internal algorithms. • The result is (hopefully) a balance between correction needs and user preferences. • In short, it caters for dynamic diversity. • Changes made by plug-ins are very small. • The ‘sweet spot’ is hit by accumulated adaptation.
Study Design • To see how people would respond to the tool, several studies were run. • In this presentation I will focus on the first. • The study was an hour long, and consisted of 38 participants, each of age 65+ • In the first study, the framework was populated with a single plug-in. • Adjustment of the double-click threshold. • In the second study, the framework was populated with five plug-ins.
Study Design • Before the participants were exposed to the tool, they were asked to fill out a questionnaire. • This was to determine their fit with the problems outlined in an earlier slide. • On the whole, participants were as expected. • Unaware of options in the system • Unaware how to make changes • Lacking confidence in making changes they knew how to make.
Study Design • After each of the tasks, the participants were asked to fill out a short questionnaire. • How easy did you find it? • How responsive was the mouse? • Would you be able to make it easier in future? • During the task, the system kept track of objective measures of performance. • Number of successful double-clicks • Time between double-clicks
The Study • The study involved performing two tasks, with and without the framework. • Task one was double-clicking a static target • Task two was double-clicking a moving target. • Both tasks lasted for five minutes. • The testing system was configured to the extremes of what was legally permitted by the operating system. • So as to ensure there was a need for correction.
Task One– Objective Measures Results are significant at P < 0.04 or better
Task Two – Objective Measures Results are significant at P < 0.02 or better
Summary of Objective Results • Performance with the framework active was significantly better than when it was not. • The actual increase in performance is not due to the plug-in, but instead due to the system’s internal thresholds being changed. • These changes would not have occurred though without the framework. • The use of the framework made an objective difference for participants.
Subjective Results I feel that the task was easy to perform Results for task one are significant at P < 0.001, and for task two at P < 0.05
Subjective Results I feel that the mouse was suitably responsive for the task Results for task one are significant at P < 0.01, and not significant (P < 0.1) for task two.
Subjective Results I know how to change the computer settings so that the mouse is more suited to perform this task No statistically significant difference
Summary of Subjective Results • The framework is perceived to work. • There is a statistically significant improvement in attitudes when the framework is active. • Participants generally felt that they would be unable to make the changes needed to simplify the tasks. • Across all tasks and conditions. • In some cases, the change in attitudes is considerable.
Study Follow-up • At the end of the study, participants were asked to complete a final questionnaire on their perceptions of the tool. • Strong agreement was registered for: • Tractability of the metaphor • Beneficial impact of the tool • Willingness to use the tool on their own computer • Participants also felt that the tool was not intrusive. • Despite the artificially high rate of corrections in the study.
Conclusion • The tool shows considerable promise as a mechanism for addressing end-user concerns. • It both works (improves performance) and is perceived to work. • Both of these are important in potentially interesting users in a real world version of the tool. • A further study strengthened these experimental findings with a suite of five plug-in. • The tool is due to be released as open source. • Interested parties should feel free to get in touch.