Abstract
This study examined the effectiveness of different models in predicting performance in retail point-of-sale terminal operation. Newly trained participants performed nine different transactions on a retail point-of-sale terminal. We used a similarity rating task to measure participants' conceptual representation of the domain. Predicted performance times based on physical actions were developed from a Keystroke-Level Model (KLM) analysis, while predicted times based on number of cognitive steps were developed from GOMS models of the transactions (Card, Moran, & Newell, 1983). A measure reflecting the similarity of participants' conceptual representation of the task to that of experts did not predict performance. The KLM-based predictions did significantly predict performance times for both groups, though they somewhat underpredicted them. The GOMS-based predictions of cognitive steps did not substantially add to the predictions based on the KLM analysis.
Get full access to this article
View all access options for this article.
