E-GAZE · PREVIEW MODE
Notification recall test
This preview skips the camera, participant details, calibration, and shelf tasks.
15 secNotification
→
5 secPause
→
VoiceNumber recall
Allow microphone access when asked. If recognition is unclear, a numeric fallback will appear.
✓
Return to the full E-Gaze study
PREVIEW COMPLETE
Response recorded
- Message version
- —
- Recorded answer
- —
- Response method
- —
VIRTUA-SHOP · E-GAZE MODULE
See what earns attention—and how quickly.
E-gaze measures notification recall and two yogurt shelf-search tasks after one guided calibration.
01
Not checked
Camera, microphone, and positioning
Your preview appears here
Estimated viewing distanceCheck the camera to begin
Sit about an arm’s length from the screen. The app will confirm a comfortable, steady viewing position.
Face centeredScreen at eye levelEven front light
Wearing glasses? Reduce reflections by facing a soft light and tilting the screen slightly. Keep your usual glasses on during calibration and testing.
02
Participant and study
The last name is used in the PDF filename.
Study sequence
- 01Notification recallOne randomized 15-second message
- 02Plain yogurt searchGuided randomized shelf
- 03WIC shelf-talker searchGuided re-randomized shelf
BEFORE CALIBRATION
Follow one circle at a time.
Keep your head comfortable and mostly still while your eyes follow each illuminated point. Missed points return after the grid; Stage 2 adds live estimated-gaze feedback.
1Listen for the location
2Keep your head comfortable
3Hold for five seconds
Stage 1 of 2 · Eye-led learning
CenterFollow the circle and hold
Face the illuminated point and hold for five seconds.
LIVE GAZE VERIFICATION
Measuring accuracy…
Continue looking at each target.
NEXT
Get ready
3Pause for a moment
QUICK RECALL
How many yogurt tubs are remaining?
Please say only the number.
Listening…Voice response was unclear. Select the number you said.
Find all plain yogurtHold each match for 3 seconds · Ends when all are identified
We temporarily lost your gaze. Face the camera and remain still.
VIRTUA-SHOP · E-GAZE
Calibration report
——
Accepted calibration holds · Points 1–4
E-GAZE CALIBRATION
Accepted point evidence
—Points 5–8
E-GAZE CALIBRATION
Final point and verification
——
Final calibration report
- Training points
- 9 / 9 accepted
- Training time
- —
- Mean signal confidence
- —
- Viewing position
- Accepted position held
- Training pattern
- Passed
- Live verification
- 5 / 5 passed
- Mean verified error
- —
- Overall quality
- —
Live verification points
Signal confidence describes feature stability during each training hold. Screen accuracy is measured separately by the five live verification targets.
Webcam accuracy and probable gaze region
At an assumed 60 cm viewing distance, a planning allowance of 2–3° corresponds to approximately 21–31 mm on the screen. The webcam confirms relative viewing position, not physical distance. The rectangle represents the probable region around the estimated gaze point—not certified measurement accuracy. The shelf buffer is calculated after verification.
An opposite-side or opposite-corner estimate is treated as unreliable tracking, not ordinary angular error.SHELF SEARCH
Plain yogurt attention and motion path
SHELF SEARCH
WIC attention and motion path
Research interpretation
Webcam gaze represents probable attention zones. Results use the verified calibration error as an area-of-interest buffer. Coverage and completion time are reported separately, while high calibration uncertainty remains explicitly inconclusive.
DESCRIPTIVE REPORT
Session interpretation
——
How to read these findings
—
Measurement boundaries
Results indicate probable visual attention under this session's calibration. They should not be treated as clinical eye-tracking measurements, proof of comprehension, or proof that an item was consciously evaluated.
Research use
The notification task measures recall of a numeric benefit after brief exposure. The shelf tasks measure search behavior, dwell confirmation, target coverage, revisit behavior, tracking availability, and time required to identify every target. Compare participants or experimental conditions only after applying the same device, distance, lighting, instructions, and target layout rules.
CALIBRATION DATA
Point-by-point analytical table
—Nine training + five verification points
Nine-point training
| Point | Target | Total time | Stable hold | Confidence | Attempts | Direction result |
|---|
Live verification
| Point | Median error | P90 error | Usable coverage | Attempt | Interpretation |
|---|
TEST DATA
Trial-level analytical table
—Notification and shelf outcomes
| Trial | Condition/task | Valid duration | Primary outcome | First target | Tracking | Pauses | Path / dwell |
|---|
Metric definitions
| Metric | Definition | Recommended use |
|---|---|---|
| Coverage | Unique correct targets confirmed by a continuous three-second dwell. | Primary shelf completion measure. |
| Completion time | Valid tracked time from shelf onset until the last unique target is confirmed. | Compare search efficiency when coverage is complete. |
| First-target latency | Valid tracked time until the first unique target confirmation. | Measure early findability and first-glance support. |
| Revisit dwell | A completed dwell on a target already confirmed. | Indicates rechecking or uncertainty. |
| Off-target dwell | A completed three-second dwell that did not overlap a correct target area. | Indicates distraction, misclassification, or mapping error. |
| Tracking availability | Valid tracked time divided by valid time plus paused tracking time. | Quality-control covariate; interpret low values cautiously. |
PLAIN YOGURT TARGETS
Target-level analytical table
——
| Position | Target ID | First attention | Confirmation | Order | Total dwell | Core dwell | Buffer dwell | Visits | Status |
|---|
Interpretation note
Core confirmations fall inside the product area. Probable confirmations fall inside the calibration-derived buffer. A target may receive attention without completing the continuous three-second confirmation dwell.
WIC-APPROVED TARGETS
Target-level analytical table
——
| Position | Target ID | First attention | Confirmation | Order | Total dwell | Core dwell | Buffer dwell | Visits | Status |
|---|
Data files
The PDF provides the visual and human-readable record. The companion JSON contains structured calibration, trial, target, dwell-event, and gaze-sample tables for statistical analysis.
Study limitation
Webcam estimates are sensitive to glasses glare, lighting, face position, device geometry, screen size, and calibration drift. Replicate important findings with dedicated hardware before making high-stakes claims.