Now Shipping! – The EyeLink 3; combined head and eye tracking at up to 1000 Hz.

FAQ: How can I correct drift / calibration issues in my gaze data?
#1
In general it is far better to avoid this kind of situation by ensuring optimal participant set up, a good symmetrical pattern of +s on the Host PC display during calibration, and of course by monitoring the online gaze cursor on the Host PC screen during recording (where any misalignment between gaze and text would be apparent). Please see our Hardware Setup and Usage Video Tutorials (linked  below) for best practice advice for the physical setup of specific systems and participant setup prior to data collection. Occasionally, however, even the most careful experimenter can collect data where there is a clear mismatch between the fixation locations as recorded by the eye tracker, and where the participant was actually looking. If this has occurred it is possible to manually adjust the location of fixations in Data Viewer so that the intended analysis can still be performed. In reading studies for example, there will be Interest Areas around each word/phrase in the text, so the goal of the manual adjustments would be to make sure the fixations are aligned within the appropriate Interest Areas.

There are two approaches: 1) manually adjusting a single fixation or a group of fixations, and 2) performing an automatic drift correction that averages the Y locations of a group of fixations. This latter technique is particularly useful for reading studies. It is best practices to save a copy of the viewing session prior to making any fixation changes as this will allow for a quick restore of the data for any reason.
  1. Manual Adjustments
    Fixations can be selected either in the main Spatial Overlay window, or from the list of fixations in the middle panel of the Data tab in Inspector window. To select a single fixation in the Spatial Overlay window, simply click on it (Temporarily toggling off the Interest Area / Saccade visibility on the main Data Viewer toolbar may help as it will declutter the Trial View window and allow for easier access to select the desired fixations). To select a group of fixations, create a rectangular selection box around the desire fixations by clicking and dragging the cursor in the Spatial Overlay window. To select fixations from the Data tab in Inspector window, simply click on the desired fixations while holding the CRTL (Windows) or Option (Macs) key, or click on the first fixation and then hold down the SHIFT key and click on the last fixation to select all fixation in-between. This technique is useful for ensuring that you are selecting a group of consecutive fixations. Another useful shortcut for some scenarios is pressing CTRL+A (Windows), or Command+A (Mac), which will automatically select all the fixations. If fixations were selected from the Inspector window, click on the top bar of the Spatial Overlay window in order to perform the next step.

    After selecting the fixation / group of fixations, hold down the ALT (Window) or Option (Mac) key and use the cursor arrows to manually move their X and Y positions. For single fixations, the X, Y positions can be directly edit (Ave. X Position, Ave. Y Position) in the lower panel of the  Data tab in the Inspector window.

  2. Automatic Drift Correction
    To automatically drift correct a group of fixations to their average Y value, right clicked the selected group of fixations and select the "Drift Correct" option. If this option is grayed-out it is likely because non-fixation events such as a message or an Interest Area was also selected, or a sequence of non-consecutive fixations has been selected. A useful shortcut for this automatic drift correction is to press CTRL+D (Windows), or Command+D (Mac). Automatic Drift Correction will fail if one or more fixations deviates more than 30 pixels in the Y-dimension from the mean Y position of the selected fixations. This threshold can be modified in the preference settings (Preference tab in the Inspector Window --> Data Views --> Batch Drift Correction Error Threshold). 

       
Please note that adjusting fixation positions by any of the methods described above will affect event data, including fixation location and associated saccade beginning/end points, but the sampled gaze data itself will not be changed.