Comparison of cognitive load measures

To test for agreement between the different cognitive load measures, we constructed a separate MLM model for each measure (i. e., NASA-TLX, pupil diameter mean, pupil diameter range, blink rate, total scanning time, gaze shift rate). In each model, the dependent variable was the Paas scale rating and the measure to be tested was modelled as a fixed effect. Observations from the four scenarios were clustered by participant. The NASA-TLX model was based on 116 observations (29 participants × 4 tasks), while the remainder of the models were based on 112 observations (28 participants × 4 tasks).