WebET 3.0 – Validation Study

iMotions released WebET 3.0 in May 2023, which is the best webET algorithm in the development of the product to date. The purpose of the validation study was to run a large scale study to evaluate how the algorithm performs on a truly diverse, global, sample, “in-the-wild”. Data was globally collected from 255 participants over 35 days. Participants conducted a short study comprising of gifs, images, videos, and surveys. Self-reported parameters for ethinicity, eye-color, wearing glasses or not, having facial hair or not, and lighting conditions in the room were evaluated against accuracy. Over 50% of participants had an accuracy of 2 degrees of visual angle (dva) or lower. Over 70% had an accuracy of 3dva or lower and over 90% had an accuracy of 5dva or lower. Of the parameters measured, only the presence of glasses had a significant effect on accuracy. The individual differences for ethinicities, regions, eye-colour and the presence of facial hair did not have a significant impact as people collected data in their natural environments Over time, fixation classification stays stable in the center of the screen but classification may reduce in accuracy towards the bottom corners of the screen. Longer studies and internet problems can cause problems with participant compliance and a suboptimal user experience. Therefore, researchers are advised to keep studies as short as possible in order to ensure high levels of participant compliance and optimal data quality throughout the study.

View on ResearchGate