Differences in Average Power Output Values from Computational Models of Repeated Vertical Jump Tests: A Single-Group...

Background: Repeated vertical jump tests are widely used to assess neuromuscular function and lower limb performance. However, inconsistent formulas for average power output produce large discrepancies, limiting comparability across studies and limiting practical applications. This study aimed to compare three different models for the calculation of average power output, Bosco, Miron Georgescu (MG), and Modified Miron Georgescu-15s (MGM-15), applied to identical jump test data, in order to evaluate their computational behavior and practical relevance in athlete performance profiling. Methods: A single-group quasi-experimental study was conducted with 25 physically active male university students (mean age: 21.4 ± 2.7 years), who performed a 15 s repeated vertical jump test on the OptoJump Next system. Raw parameters including flight time, contact time, and jump height were recorded and exported. Average power output (W/kg) was subsequently calculated using three distinct computational models, each applied to the same dataset of flight and contact times. A repeated-measures ANOVA was used to compare outputs across models, with Bonferroni-adjusted pairwise comparisons for post hoc analysis (α = 0.05). Results: Significant differences were observed (p < 0.001). The Bosco model produced the highest values of average power (40.13 ± 8.56 W/kg), followed by MG (21.07 ± 5.92 W/kg), while MGM-15 yielded the lowest and most consistent outputs (4.08 ± 0.61 W/kg). Effect sizes were very large (η²_p = 0.952), confirming that calculation models strongly influenced the outcomes. Conclusions: The findings demonstrate that average power output differed markedly across formulas, despite identical performance data. Bosco and MG models tended to overestimate values due to simplified assumptions, whereas the MGM-15 method produced lower and more consistent outputs that may better capture repeated jump demands. The standardization of computational models is fundamental to ensure comparability and to improve athlete performance profiling in research and practice.