Using a novel form of meta-analysis, this research estimates information processing parameter estimates for older adults using the Card, Moran, & Newell (1983) Model Human Processor model and applies these perceptual, motor, and cognitive parameters to mobile phone usability studies across tasks of increasing complexity. Older adult models predicted older adult human performance data extremely well (R = 0.99), and older adult models produced equivalent goodness-of-fits to previously validated younger adult models for task completion time, suggesting that older adult parameters are valid for modeling purposes. Critical path analyses for mobile phone tasks supported and substantiated human factors findings and were also decomposed to highlight places where errors would likely occur as a function of cognitive workload, hardware or software design (including menu depth/breadth, button size, environmental supports), and user characteristics. Errors were then classified in ways amenable to modeling and error probabilities were extracted from known human performance. These were implemented in a novel GOMS model for error prediction in a more complex task, and analyses revealed no differences between model predictions and human production of error across all types of classified errors and across young and old age. As such, this research validates older adult parameters so that capabilities and limitations may be better understood with regard to existing designs, and so future technologies may be better designed around the needs of older adults. Further, this research decomposes errors into classifications amenable to human performance modeling, and extends a modeling technique to account for error prediction. |