Abstract
Human error is pervasive (Kirwan, 1994), and its consequences are often costly (e.g. time, money) or even deadly (e.g. Neumann, 1995). Yet, many types of human fallibility are not anomalous occurrences, but instead flow from psychological processes that normally produce correct behavior (Mach, 1905). It is therefore reasonable to argue that some types of errors may be predictable (so long as situational demands and human capabilities/limitations are properly taken into consideration) and subsequently, amenable to computational modeling techniques. This research seeks to incorporate and implement psychological theory of human error in GOMS models of human performance by first classifying errors under a simplified GEMS schematic (Reason, 1990), then extracting observed error probabilities from simpler mobile phone tasks, and validating model predictions in a more complex, mobile phone task. Results revealed no differences between model predictions and human production of error across all types of error classification, and models accurately predicted error rates for younger and older users based upon previously validated, age-informed processing parameters included in each age-sensitive model.
Get full access to this article
View all access options for this article.