The purpose of eye tracking is to measure a person’s point of gaze. However, there is almost always disparity between actual gaze location and the location recorded by the eye tracker. In most eye tracking research, people just take the data from eye trackers and accept the manufacturer’s stated accuracy. We did a pilot study to test the accuracy of a commercial remote eye tracker. In the experiment, we inserted a slide between every two stimuli. The slide had a cross at the center of the screen located on an invisible triggering area of interest (AOI). Observers were asked to look at the cross to enable the next stimulus. There was a built-in calibration and validation at the beginning and after every five stimuli. After the experiment, we examined both the raw gaze data and the fixations exported from analysis software on the test slides. We found that the disparity between the actual target position and the reported gaze point was larger than the reported system calibration error on 40% of the trials. Two eye trackers of the same model were used in the experiment and we found that the two trackers were different in terms of tracking error. The result of the study helped us better determine the actual tracking error, which could help us better visualize and understand peoples’ viewing patterns. The study also indicates that properly operating eye trackers is not as simple as claimed by manufactures. 


D.Wang, A.Haake, S.Simpson,J.Pelz (2013). Head motion compensation for remote eye tracker. In K.Holmqvist, F. Mulvey & R. Johansson (Eds.), Book of Abstracts of the 17th European Conference on Eye Movements, 11-16 August 2013, in Lund, Sweden. Journal of Eye Movement Research, 6(3), pp. 225.

Dong Wang, Preethi Vaidyanathan, Anne Haake, and Jeff Pelz, Are eye trackers always as accurate as we assume?,  Society for Computers in Psychology, Minneapolis, MN, November, (2012)