Project Overview
Eye-tracking is a technique used to measure where and how long people focus their gaze within a
visual environment. By analyzing eye movements, we can gain insights into cognition, attention,
behavior, and decision-making processes.
In this case study, I used a publicly available dataset from the Nemo Science Museum in Amsterdam.
Participants viewed a feature-rich image for 10 seconds while their gaze behavior was recorded.
The project focuses on fixation durations and how these may differ by age and over time during the experiment.
Goals
The analysis focuses on two primary research questions:
Age and Eye Movements
Do fixation durations decrease with age during childhood and increase again later in life?
Fixation Patterns Over Time
Are there systematic changes in fixation durations throughout the experiment?
Data
The dataset contains recordings from more than 2,600 participants. Each participant viewed an image
for a fixed duration while gaze coordinates, timestamps, pupil information, and fixation-related
features were recorded or derived.Link
Data Structure
-
Pre-processed data: gaze positions, x/y coordinates, pupil size, timestamps,
and experimental event messages.
-
Processed data: detected fixations with average x/y coordinates and fixation duration.
-
Demographics data: participant-level information such as gender and year of birth.
Methodology
The project used statistical analysis methods to investigate the relationship between fixation
behavior, age, and temporal patterns during the viewing task. The analysis ranged from descriptive
exploration to inferential modelling.
Analysis Pipeline
-
Data preprocessing: converted raw or pre-processed gaze data into fixation-level data.
-
Exploratory analysis: inspected distributions, missing values, age groups, and fixation durations.
-
Statistical modelling: investigated age-related and time-dependent patterns in fixation behavior.
-
Robustness checks: evaluated whether results remained stable under alternative processing choices.
Robustness Analysis
As an additional step, the project considered how sensitive the results were to changes in the data
processing pipeline. This is important because eye-tracking results can be affected by preprocessing
choices and fixation detection parameters.
- Modified hyperparameters in the fixation detection algorithm.
- Compared alternative processing pipelines.
- Considered noise or missing-data scenarios to evaluate stability.
Outcome
The project strengthened my ability to work with experimental data, convert raw measurements into
analysis-ready features, and apply statistical reasoning to behavioral data. It also improved my
understanding of how preprocessing choices influence downstream modelling results.
Eye Tracking
Fixation Data
Statistical Modelling
Data Preprocessing
Robustness Analysis
TU Dortmund