Algorithms for the automated correction of vertical drift in eye tracking data


Behavior Research Methods


January, 2021


Jon W. Carr, Valentina N. Pescuma, Michele Furlan, Maria Ktori, Davide Crepaldi

We evaluate nine algorithms for the correction of “vertical drift” in eye tracking data — the vertical displacement of fixations that results from a gradual loss of eye tracker calibration over time. This correction is particularly important for experiments that involve reading multiline texts because it is critical that fixations on one line of text are not erroneously assigned to an adjacent line. Our results — based on both simulated and natural eye tracking data — show that certain algorithms are better than others on particular kinds of vertical drift and reading behavior. We offer evidence-based advice to researchers on how to choose an appropriate technique and concrete suggestions on how drift correction software can be improved going forward.

Published paper: