We designed and evaluated an agent that recommends a pointing device gain for a given user, with mixed success. 12 participants with physical impairments used the Input Device Agent (IDA), to determine a recommended gain based on their performance over a series of target acquisition trials. IDA recommended a gain other than the Windows default for 9 of 12 subjects. Subsequent performance using the IDA gain showed no meaningful differences as compared to the default setting or users’ pre-study settings. Across all gains used by these subjects, however, gain did have a significant effect on throughput, percent of error-free trials, cursor entries, and overshoot. Linear models of gain’s effect on performance showed that its effect on throughput is relatively small, with only a 13% difference from highest throughput (at gain = 10) to lowest throughput (at gain = 6). Cursor entries were more strongly affected, showing a steady increase with increasing gain. Categories and Subject Descr...
Heidi Horstmann Koester, Edmund F. LoPresti, Richa