Introduction

gazerr is a Python package which provide a command line utility to estimate the error distribution of gaze duration measurements. It requires a calibration file from an eye tracking application that indicates the underlying error in the fixation model, then uses that to generate the expected distribiution of true duration for any measurement.

Motivation

The use of eye fixation models to estimate gaze duration (or dwell time) in Area of Interest studies (AOI) was discussed by Holmqvist et al (2012). The authors simulated the impact of gaze fixation error by adding noise according to manufacturer specifications to data sets of low margin areas of interest. We extend this idea to calculate probabilistic bounds on the error in gaze duration using calibration data as a source of noise distribution that is specific to both the study participants and the device/environment of the study. The result is an open source application that may be used by a wide variety of reserachers to provide error bounds on any gaze duration measurements.