Visual search behavior and decision-making in expert and less-experienced football referees across different foul situations: an eye-tracking study.

INTRODUCTION

Modern football imposes considerable demands on referees' rapid foul judgements, yet the dissociation between accuracy and efficiency of officiating expertise across different foul situations remains poorly characterized.

METHODS

The present study compared 15 national-level and 15 Class-3 male football referees in three typical situations: ball-contesting, tactical foul and handball. Participants viewed 30 real-match video clips and made one technical and one disciplinary decision per clip while an EyeLink 1000 Plus eye tracker recorded global eye-movement indices and area-of-interest (AOI) measures. Data were analyzed with 2 (group) × 3 (situation) mixed-design ANOVAs and independent-samples t-tests.

RESULTS

Decision accuracy did not differ between groups for either technical or disciplinary judgements, although a robust situation main effect was observed [technical accuracy F (2,56) = 116.45, p < 0.001, η2 p = 0.81; disciplinary accuracy F (2,56) = 13.73, p < 0.001, η2 p = 0.33]. For decision efficiency, disciplinary reaction time showed a Group × Situation interaction [F (2,56) = 4.16, p = 0.021, η2 p = 0.13]; simple-effect analysis indicated that national-level referees responded faster than Class-3 referees in ball-contesting situations [F (1,28) = 6.07, p = 0.020, η2 p = 0.18]. Pupil diameter showed an analogous interaction (p = 0.026), although within-situation simple effects did not reach significance. AOI analyses revealed that national-level referees fixated the upper body of the fouling player earlier in ball-contesting situations [t(26.02) = -2.38, p = 0.024, d = 0.87] and revisited the attacker's lower body and the defender's upper body less often in handball situations (both p < 0.05, d ≥ 0.80).

DISCUSSION: Within this video-based task, expertise-related differences emerged primarily in efficiency-related indices rather than overall accuracy; sample size, the simulated nature of the stimuli and ceiling-or-near-chance effects in some conditions limit generalization, and the findings warrant validation in real-match settings.