Raw data for A perceptual mechanism of cybersickness in virtual reality systems

<p>The uploaded data contains research data supporting Chapter 3 and 4 of the thesis “A Perceptual Mechanism of Cybersickness in Virtual Reality Systems”.</p> <p> </p> <p>For Chapter 3, the cybersickness severity was evaluated using peak MISC scores. In particular, a one-way repeated-measures ANOVA was conducted on the peak MISC scores to test for significant differences between conditions (visual, oscillation, combined, control). As some MISC baseline scores (MISC-Pre) were not zero, the baseline was subtracted from the MISC scores (MISC -1, 2, 3, 4) to obtain stimulus-related scores. Then, a two-way repeated-measures ANOVA (Condition × Time-point) was conducted on the MISC scores to test for the temporal dimension of differences between conditions.</p> <p> </p> <p>Then, the severity of cybersickness symptoms was evaluated using the SSQ. The total severity score was computed in the recommended manner and analysed using nonparametric methods, as previous studies suggested that SSQ scores are not parametric. A related-samples Friedman’s two-way ANOVA by ranks and Dunn-Bonferroni tests were conducted on the relative SSQ scores to test for significant differences between conditions. As some baseline scores (SSQ-Pre) were not zero, the baseline was subtracted from the SSQ scores (SSQ-Post) to obtain stimulus-related scores. Then, related-samples Friedman’s two-way ANOVA by ranks were conducted on each subscore of the SSQ under the four conditions.</p> <p> </p> <p>The data from the balance board was computed into a time series CoP position data in AP and ML axes. To ensure that participants reached a steady posture, postural data sampled in the first 10 s was removed. The positional variability of the CoP position was computed to evaluate the spatial magnitude of postural activity. In particular, the <i>SD</i> of CoP positions of the two axes was used. First, to ensure that the low-frequency heave oscillation manipulation led to increased postural activity, the positional variability before and after the stimulus exposure were compared using paired-samples <i>t</i>-test for the oscillation and control condition in each body axis (AP, ML). Then, the differences in the positional variability before and after stimulus exposure were compared by performing ANOVA per body axis. In particular, two-way repeated-measures ANOVAs were conducted to test for significant differences between time (before, after stimulus exposure) and conditions.</p> <p> </p> <p>For Chapter 4, the cybersickness severity was evaluated using peak MISC scores. In particular, a one-way repeated-measures ANOVA was conducted on the peak MISC scores to test for significant differences between conditions (fixation, FOV restriction, combined, control). As some baseline scores (MISC-Pre) were not zero, the baseline was subtracted from the MISC scores (MISC-1, 2, 3, 4, Post) to obtain stimulus-related scores. Then, a two-way repeated-measures ANOVA (Condition × Time-point) was conducted on the MISC scores to test for the temporal dimension of differences between conditions.</p> <p> </p> <p>Then, the severity of cybersickness symptoms was evaluated using the SSQ. The total severity score was computed in the recommended manner and analysed using nonparametric methods, as previous studies suggested that SSQ scores are not parametric. A related-samples Friedman’s two-way ANOVA by ranks was conducted on the relative SSQ scores to test for significant differences between conditions. As some baseline scores (SSQ-Pre) were not zero, the baseline was subtracted from the SSQ scores (SSQ-Post) to obtain stimulus-related scores.</p> <p> </p> <p>The data from the balance board was computed into a time series CoP position data in AP and ML axes. To ensure that participants reached a steady posture, postural data sampled in the first 10 s was removed. The positional variability of the CoP position was computed to evaluate the spatial magnitude of postural activity. In particular, the <i>SD</i> of CoP positions of the two axes was used. To understand the changes in postural activity, the <i>SD</i> were sampled in four 60 s time period sequentially, leaving the last 14 s unused. A two-way repeated-measures ANOVA (Condition × Time period) was conducted in the positional variability per body axis to test for the differences between conditions. The data from the balance board were processed in MATLAB using custom code without the use of any additional filter.</p> <p> </p> <p>Raw and pre-processed data are available upon request from adriang@connect.hku.hk.</p>