Reliability, Construct Validity, and Sensitivity of Goalkeeper-Specific Movement and Reactive Agility Tests in Youth Soccer Goalkeepers.

Abstract

Purpose: To develop and evaluate the reliability, construct validity, and sensitivity of 2 field-based goalkeeper-specific tests designed to quantify movement performance (goalkeeper-specific movement [GSM]) and reactive agility (goalkeeper-specific agility [GSA]), alongside derived reaction-time indices in youth soccer goalkeepers.

Methods: Fifty-eight goalkeepers (12-19 y) completed a standardized test battery (countermovement jump, 5- to 20-m linear sprints, V-cut change-of-direction, GSM, and GSA) on 2 occasions separated by 7 days. Absolute reliability was assessed using the coefficient of variation (CV) and standard error of measurement, relative reliability using intraclass correlation coefficients (ICC; 95% CIs), and sensitivity by comparing standard error of measurement with the smallest worthwhile change. Construct validity was examined through Pearson correlations with established performance tests and known-groups comparisons across age categories.

Results: Linear sprint and V-cut outcomes showed acceptable-to-good reliability (CV ≤10%; ICC: .70-.93). For the goalkeeper-specific tests, total-time outcomes demonstrated acceptable reliability and sensitivity across age categories (GSM: CV 4.2%-7.0%, ICC: .75-.85; GSA: CV 4.2%-6.0%, ICC: .50-.79), whereas side-specific measures and reaction-time indices were less consistent. GSM and GSA total times were strongly interrelated (r = .71-.82) and moderately-to-strongly associated with sprint and change-of-direction performance, and differed by age group, with generally lower times in older goalkeepers.

Conclusion: GSM and GSA total-time outcomes offer goalkeeper coaches and practitioners practical, age-sensitive field-based tools to profile, monitor, and guide goalkeeper-specific movement and reactive agility training. In contrast, reaction-time indices derived from difference scores show limited consistency and are not recommended for monitoring.