TY - JOUR

T1 - Associations among Workload Dimensions, Performance, and Situational Characteristics

T2 - A Meta-Analytic Review of the Task Load Index

AU - Hertzum, Morten

PY - 2022

Y1 - 2022

N2 - Workload is an important explanatory variable in human-computer interaction and commonly measured with the Task Load Index (TLX). Thus, it is important to understand the qualities of TLX and its relations to other variables. By reviewing 384 papers that apply TLX, this study analyzes how differences in TLX and its six subscales are associated with one another and with differences in performance, user experience, and situational characteristics. Six findings stand out. First, the TLX subscales measure associated, but somewhat independent, dimensions of workload. Second, people compensate for demanding conditions by putting in more effort and, as a result, sometimes avoid a drop in performance. Third, differences in workload are associated with differences in error rate, task completion time, and user experience but the strength of association is merely slight to fair. Fourth, differences in opposite directions between workload and either error rate, task completion time, or user experience are few but occur for all TLX subscales. Fifth, differences in workload dimensions are more often associated with differences in tasks and contexts than users and systems. Sixth, the TLX subscales – not just the composite TLX score – are widely used for testing cross-system differences in workload.

AB - Workload is an important explanatory variable in human-computer interaction and commonly measured with the Task Load Index (TLX). Thus, it is important to understand the qualities of TLX and its relations to other variables. By reviewing 384 papers that apply TLX, this study analyzes how differences in TLX and its six subscales are associated with one another and with differences in performance, user experience, and situational characteristics. Six findings stand out. First, the TLX subscales measure associated, but somewhat independent, dimensions of workload. Second, people compensate for demanding conditions by putting in more effort and, as a result, sometimes avoid a drop in performance. Third, differences in workload are associated with differences in error rate, task completion time, and user experience but the strength of association is merely slight to fair. Fourth, differences in opposite directions between workload and either error rate, task completion time, or user experience are few but occur for all TLX subscales. Fifth, differences in workload dimensions are more often associated with differences in tasks and contexts than users and systems. Sixth, the TLX subscales – not just the composite TLX score – are widely used for testing cross-system differences in workload.

KW - Faculty of Humanities

KW - Mental workload

KW - NASA-TLX

KW - Task load index

KW - TLX

KW - Workload

U2 - 10.1080/0144929X.2021.2000642

DO - 10.1080/0144929X.2021.2000642

M3 - Journal article

VL - 41

SP - 3506

EP - 3518

JO - Behaviour and Information Technology

JF - Behaviour and Information Technology

SN - 0144-929X

IS - 16

ER -