An Evaluation of Movement Data Analysis Techniques for Virtual Reality
Abstract
Researchers frequently collect position and orientation data from tracking hardware during virtual reality research studies. These data capture important information about participants' movements during experiments, but often result in large and complex datasets that can be challenging to analyze and interpret. We explore the potential of standard statistical measures that could be used to interpret position and orientation tracking data, and discuss advantages and uses of each measure. We further present three new measurement techniques -- 95% range, time spent at mode, and coefficient of determination. We then evaluate the effectiveness of each statistical measure to detect differences in position and orientation on two existing data sets. We find that all of the tested measures are effective in discerning known main effects in position and orientation with varying degrees of sensitivity. This work is intended to guide researchers on future analysis and interpretation of motion tracking data sets.