| Title |
Magnitude and nature of variability in eye-tracking data / |
| Authors |
Holmqvist, Kenneth ; Zemblys, Raimondas ; Beelders, Tanya |
| DOI |
10.16910/jemr.10.6 |
| Full Text |
|
| Is Part of |
Journal of eye movement research: vol.10, iss. 6: Abstracts of the 19th European Conference on Eye Movements, August 20-24, 2017, Wuppertal, Germany.. Bern : International Group for Eye Movement Research. 2017, vol. 10, iss. 6, p. 179.. ISSN 1995-8692 |
| Keywords [eng] |
precision measures ; noise ; data quality ; raw data signal |
| Abstract [eng] |
Existing precision measures do not adequately describe the magnitude of variability (e.g. noise) in eye- movement data, because they are also affected by the nature of the variability in the signal. For instance, sample-to-sample RMS of an “ant trailing” signal would be low, indicating good precision, while the signal could spread over a large spatial extent. Conversely, for sawtooth signal artifacts that are for instance produced by a spurious corneal reflection, the spatial standard deviation (STD) can be small while RMS can be very large. We have developed two new complementary measures of variability that unambiguously indicate the magnitude of variability independent of its nature, and orthogonally the nature of the variability independent of its magnitude. We hypothesize that the nature of the variability is a constant property of an eye-tracker, while the magnitude varies with many factors specific to the situation when data are recorded (such as gaze position on the screen and pupil size). Data quality studies benefit from such a distinction. Our measure could further be employed to test how robust event detection algorithms are to increases in the magnitude of variability in the eye-movement signal, and which algorithms are most suitable for which type of signal variability. |
| Published |
Bern : International Group for Eye Movement Research |
| Type |
Conference paper |
| Language |
English |
| Publication date |
2017 |
| CC license |
|
