Predictive Workload Models and Multiple-Task Performance
DOI: 10.1177/154193129203600105
archive: indexed pipeline: cataloged
Abstract
The goal of the present study was to evaluate three workload models, VACP (Aldrich, Szabo, and Bierbaum, 1988), TLAP (Parks and Boucek, 1988), and WINDEX (North & Riley, 1988), in terms of how well they account for task performance results gathered from a laboratory experiment. The models are discussed in terms of their treatment of five timesharing issues: 1) the nature of workload components (mutually exclusive or partially overlapping); 2) the utility of distinguishing between cognitive processing codes; 3) classifying voice response (psychomotor function or separate component; 4) qualitative vs. quantitative coding of task demands; and 5) the utility of an overload red-line. A correlational analysis was performed on model predictions versus actual performance for 16 different task loading conditions, to evaluate each of the three workload models. All three models did a good job of predicting performance differences across conditions, accounting for between 61% and 77% of the variance, with the TLAP model providing the best prediction. In addition, a hybrid model was developed (using the “optimal” assumptions concerning the five timesharing issues) which accounted for 85% of the variance. The results are discussed in terms of the viability of the assumptions made by each of the models with respect to the five timesharing issues.
Access
Route: Publisher paywall (check institutional access or ResearcherGate)