Fig 1.
Schematic overview of the empirical study.
Table 1.
Description of objective performance classes.
Fig 2.
Partial dependencies of objective performance classes.
Table 2.
Description of research hypotheses.
Fig 3.
Illustration of research hypotheses.
Table 3.
Formal description of research hypotheses (corr a correlation coefficient and acc the accuracy of a prediction).
Table 4.
Mental workload & usability—Groups A, B (G.A/G.B).
Fig 4.
Summary statistics by task.
Fig 5.
Scatterplots of NASATLX, WP vs SUS.
Table 5.
Pearson and Spearman correlation coefficients of the usability and the mental workload scores.
Fig 6.
Density plots of the correlations by task—Group A, B.
Table 6.
Correlations MWL vs usability. Groups A and B.
Fig 7.
Details of tasks with moderate/high correlation.
Table 7.
Correlation MWL-usability by user.
Fig 8.
Density plots of the correlations by user.
Fig 9.
Density plots of the correlations of selected users by task (top, groups A, B) and by users (bottom).
Fig 10.
Relationship between usability and mental workload scores for participants with moderate or high Pearson and Spearman correlation coefficients.
Table 8.
Correlations of the mental workload scores with the usability scores by performance class.
Fig 11.
Correlations of the mental workload scores and the usability scores by performance class.
Fig 12.
Original distribution of the objective performance classes.
Table 9.
Frequencies of classes.
Fig 13.
Independent features and classification techniques.
Fig 14.
Distribution of the accuracies of individual and combined induced models ordered by mean.
Table 10.
Ordered distributions of accuracies of trained models grouped by learning technique (combined highlighted).
Table 11.
Wilcoxon test of distributions of accuracies ordered by independent features with 95% confidence intervals (statistically significant different models highlighted).
Fig 15.
Ordered distributions of accuracies of trained models by mean using full feature sets of original mental workload and usability assessment instruments.
Table 12.
Ordered distributions of accuracies of trained models using full feature sets of original mental workload and usability instruments (combined models highlighted).
Table 13.
Wilcoxon test of distributions of accuracies ordered by independent features with 95% confidence intervals using mental workload and usability attributes (statistically significant different models highlighted).
Fig 16.
Intra-correlations of mental workload and usability questionnaire items.
Table 14.
Cronbach’s Alpha of the mental workload and usability questionnaire items.
Table 15.
Formal description of research hypotheses and their acceptance status (corr a correlation coefficient and acc the accuracy of the model’s prediction).