QUICK AND DIRTY GOMS - A CASE-STUDY OF COMPUTED-TOMOGRAPHY INTERPRETATION

GOMS (goals, operators, methods, and selection rules) has the potentia l for a high degree of theoretical accuracy. However, this accuracy is time-consuming to realize under field conditions, and, consequently, GOMS generally is not used by software engineers. Nevertheless, in our experience, GOMS models can be practical if the effort required to pr oduce and use the model is commensurate with its limited practical acc uracy. To speed and simplify the development of GOMS models, we develo ped ''quick and dirty'' GOMS-or QGOMS. In this work, we detail an 8-ye ar case study in which QGOMS models were used by a software engineerin g team to develop and evolve an interface for the electronic display o f computed tomography (CT) medical images. Experimentally generated ti ming results were compared to model estimates to provide an indication of expected accuracy for QGOMS models under field conditions. We have also implemented a direct-manipulation graphical tree editor that all ows rapid development and analysis of QGOMS models. In addition to rap id model construction, QGOMS provides ''probability selection rules'' allowing a more refined GOMS analysis.