Alternative Approaches to Analyzing SWAT Data

Abstract

The present study sought to determine the sensitivity of three alternative approaches to deriving a workload composite measure based upon data gathered using the Subjective Workload Assessment Technique (SWAT) and to determine through the use of multivariate statistical procedures (MANOVA) if anything is to be gained by retaining the individual scale information of SWAT. The three rating scale dimensions of SWAT (time load, mental effort load, psychological stress load) were combined into a single workload composite using three techniques: conjoint measurement; a simple sum of the three scales weighted equally; an empirically determined weighted-linear combination of the three scales (from MANOVA). Using data gathered by having subjects perform a continuous memory task under twelve levels of task difficulty, it was found that the three composite measures were equally sensitive and highly correlated (the minimum correlation among the three composites being 0.9913). The results of the MANOVA performed on the same data indicated that the individual scales of SWAT were differentially sensitive to different task demands and that individual scale information should be retained rather than rely on a simple composite.

Estimating a Mental Workload Redline in a Simulated Air-to-Ground Combat Mission

“…However, rate manipulations did not just affect the Time Demand subscale. Other subscales also have been found to be sensitive to rate manipulations just as the Time Demand subscale has been found to be affected by complexity or difficulty manipulations (Biers & Masline, 1987;Potter, 1986;Potter & Acton, 1985). It is not clear whether this lack of scale independence is a consequence of response strategies or whether it occurs because the underlying felt demands on these subscales are not orthogonal.…”

“…It should be noted, however, that the airspeed manipulation may have more general effects than just increasing felt time demand. One of the subscales that contributes to the overall SWAT score is a Time Demand subscale, which has been found to be differentially sensitive to rate manipulations (Biers & Masline, 1987;Potter, 1986;Potter & Acton, 1985). However, rate manipulations did not just affect the Time Demand subscale.…”

SWAT: Do We Need Conjoint Measurement?

“…Biers and his colleagues in a series of laboratory studies (Biers and Masline, 1987;Biers and McInernep 1988;Masline, 1986), investigated the unweighted sum as an alternative to using conjoint measurement in the formation of the workload composite with SWAT. In these studies, subjects engaged in a continuous recognition task in which they were presented twelve different levels of task difficulty formed from the factorial combination of number back (1 vs. 2), number of digits (1,2, or 3), and presentation rate (slow vs. fast).…”

“…Immediately after performing the task under one of the task combinations, subjects rated the workload on the three 3-points SWAT scales. Biers and Masline (1987) applied the unweighted sum metric to a set of previously collected data (Masline, 1986) in which the standard SWAT procedure had been used to form the workload composite. In the Masline (1986) study, subjects first completed the card sorting task and then their rankings were converted to workload values using conjoint measurement (a group solution).…”