Can Referee Psychological Factors Predict Performance? A Machine Learning Approach to Basketball Classification Referees

Referees are important in ensuring justice, continuity, and integrity in sporting events, with their performance shaped by both technical and physical skills as well as psychological characteristics. This research examines how self-efficacy, emotion management, decision-making style, and physical self-esteem impact performance among top basketball referees. Data were gathered from 58 Class-B referees within the Turkish Basketball Federation during the 2024–2025 season via standardized psychometric instruments in conjunction with official end-of-season performance metrics. A systematic machine learning methodology was used, including three experimental stages: feature selection using p-values, correlation analysis, and dimensionality reduction using Principal Component Analysis (PCA). The model, including characteristics selected through correlation analysis, had superior performance, achieving a Mean Squared Error (MSE) of 3.930 and an R² value of 0.879, indicating a robust predictive link between psychological factors and referee performance. These results highlight the predictive significance of psychological preparedness, namely self-efficacy and emotion management, and illustrate the utility of data-driven feature selection in enhancing model accuracy. This study proposes the incorporation of psychological evaluation in referee development initiatives and underscores the promise of machine learning in improving talent discovery and performance assessment in sports officiating.