We appreciate Lolli`s1 interest in our work, yet his comments are grounded in a conceptual framework misaligned with the aims and applied context of our work. By convention, inferential thresholds are often applied directly to practice. Nevertheless, inferential statistics quantify long-run probability by determining what would constitute a "real" change across repeated samples from an assumed population,2 not for the athletes being assessed. Because samples can vary substantially, the magnitude of change required to exceed inferential thresholds (eg, ±1.96 × SD) becomes disproportionately large and, in well-trained athletes whose expected performance adaptations are relatively small, may be practically unattainable. Imagine a practitioner in London basing an intervention on findings from a group of athletes in Australia. While inferential statistics might treat these groups as drawn from the same population, in practice, their environmental constraints, training methodology, and athlete characteristics may differ considerably. This illustrates why generalizing inferential thresholds across contexts can be misleading and contributes to overly conservative metrics such as the minimal detectable change (1.96 × v2 × SEM).3 Methods reflecting the variability of one`s own athletes are preferable in practice, which was the aim of our study.4 We used the signal-to-noise ratio, as in recent elite applied studies,5,6 at both group and individual levels. This compares seasonal performance variation with between-day noise (ie, coefficient of variation) and quantifies squad-specific responsiveness, explicitly integrating the between-day reliability that Lolli claims we contradicted1.
© Copyright 2026 International Journal of Sports Physiology and Performance. Alle Rechte vorbehalten.