When is enough enough? Empirical guidelines to determine participant sample size for scene viewing studies

Area of Science:

• Cognitive Psychology
• Neuroscience
• Human-Computer Interaction

Background:

• Eye tracking is vital for understanding spatial attention allocation.
• Establishing participant sample size guidelines for gaze analysis is challenging.
• Existing guidelines for classical statistical tests do not directly apply to gaze distribution or area-of-interest analyses.

Purpose of the Study:

• To provide evidence-based guidelines for determining participant sample sizes in eye tracking studies.
• To address the lack of clear recommendations for gaze distribution map and area-of-interest analyses.
• To enable researchers to achieve reliable and reproducible gaze estimations.

Main Methods:

• Analysis of two large-scale eye tracking datasets: dataset 1 (1248 participants, single image) and dataset 2 (84 participants, 200+ images).
• Evaluation of sample size effects on gaze distribution map similarity using Normalized Saliency Score.
• Assessment of sample size effects on area-of-interest analysis outcome variance.

Main Results:

• Diminishing returns observed for increasing participant sample sizes in both dataset types and analysis methods.
• Specific examples provided: a 5% increase in distribution map similarity requires sample sizes of 13-34 or 10-32 participants.
• A 25% decrease in outcome variance for area-of-interest analysis requires sample sizes of 13-44 participants.

Conclusions:

• No single sample size fits all eye tracking studies; recommendations depend on dataset and analysis type.
• The study provides practical guidelines and reference tables for determining optimal participant sample sizes.
• Findings benefit researchers in academia and industry for efficient and reliable eye tracking study design.