Learn how to use RealEye.io - distributed webcam eye-tracking platform.

Beata Lewandowska
Written by Beata Lewandowska

Participant Quality Stats Explained

How do we grade eye-tracking result quality?


For each study, as we evaluate each participant's performance, there are Quality Stats available. You can enter those stats for the whole study and for each participant separately.

Each result has assigned one of the following grades:

  • Perfect
  • Very Good
  • Good
  • Average
  • Low
  • Very Low

Grades are based on the following parameters' values:

Where:

  • sampling rate informs about how many samples per second are collected (30 Hz means that each second on average 30 samples are collected),
  • fixations can be computed for a minimum 20 Hz sampling rate,
  • eye-tracking data length (called also eye-tracking data integrity) informs about how many gaps in the data collected there are,
  • gaze on the screen parameter informs about for how much of the eye-tracking test time participant looked at the screen.

Results marked as Perfect/ Very Good/ Good are great quality results, ready to analyze. Usually, the "Average" results are also usable, just please remember, that below 20Hz it may happen that there are gazes collected, but fixations are not available. The quality grade does not take the "eye-tracking issues" flag (i.e. "Many head movements") into the account, only the numeric data.

"Low" result may be marked as low quality because a lot of the gaze was missing (tester wasn't looking at the screen or his gaze was lost because he/she moved), but also because there was a low sampling rate (i.e. 9 Hz), but if the eye-tracking data is collected for all items/time and it seems that the recording makes sense (if the tester was looking at the products, not in between them or only at one site of the stimuli) then you may consider using such result.

Categories: