Measurement and scaling of workload in complex performance

Two groups of paid male volunteers (Groups I, N = 51, tested with identical schedules on two successive days; Group II, N = 43, tested on one day only) performed over nine intervals on various combinations of the six tasks of the CAMI Multiple Task Performance Battery. Two of the tasks involved monitoring static (lights) and dynamic (meters) processes, and four more-active tasks involved mental arithmetic, elementary problem solving, pattern identification, two-dimensional compensatory tracking. Five of the nine intervals provided different complex tasks consisting of concurrent monitoring tasks and two of the active tasks. Other trials provided data on the single active tasks as well as the combined monitoring tasks. The results indicated that all performance measures--a total of 12 for the six tasks--varied significantly as a function of the different task-combination conditions. A standard psychological scaling technique (Thurstone Case V) was applied to the monitoring data (response times for green and red lights, and for meter monitoring) to develop an index of workload for the five complex task combinations. Since better performance was presumed to indicate a lower workload, workload was inferred to increase as performance declined across conditions. The best performances, which were assigned scale values of 0, were found to be associated with single-task performances, as expected. Scale values for the complex task-combination conditions were found to be consistent between groups and between the two days of testing of Group I (r's of 0.947 to 0.993). Although the scale values are specific to the task and task-combination conditions employed, the scaling-procedure application shows promise for cases in which quantitative measures of performance can be acquired with moderately large samples of subjects (N's greater than 50).