Each eye tracking session finishes with an accuracy test.
We ask the viewer to click on 5 dots. While they are clicking, we assume they are looking at the dots. We then estimate their gaze point and measure the distance between the estimate and the dot.
The root-mean-square error over the 5 dots is expressed as a fraction of screen height is then provided as an accuracy measure.
We do not deduct view credits for views with less than 70% accuracy (30% RMS error). However, you can still view and use these results – often the data is accurate but the viewer has looked away from the screen or shifted seating position after the viewing but before the accuracy test.
Both the viewer and the study owner are able to view these accuracy scores. The purpose of showing scores to viewers is to give them feedback on their performance during calibration and viewing.