The purpose of the present study is to analyze the agreement between different sampling frequencies (SF) to quantify the accelerometer-load in soccer. Eight under-16 male soccer players were registered during an in-season training session. Each player wore four inertial measurement units that registered the accelerometer workload index PlayerLoadRT at different sampling frequencies: (a) 10 Hz, (b) 100 Hz, (c) 500 Hz, and (d) 1000 Hz. Additionally, a down-sampling method was performed: (e) 1000 to 100 Hz and (f) 500 to 100 Hz for comparison purposes. Agreement and correlation analysis were determined using the Pearson correlation coefficient, intraclass correlation, Bland-Altman bias, and t-student of independent samples with Cohen`s d effect size. Very large to nearly perfect correlations were found between all SF (r > 0.704). An almost perfect agreement was found between all SF (ICC>0.864), except regular to substantial agreement between 10 Hz and the rest of the sampling frequencies (ICC = 0.357-0.658). Comparison analysis showed statistical differences between all sampling frequencies (p < 0.01) with the highest differences between 10 Hz and all other sampling frequencies. If trying to compare data collected at different frequencies, researchers should explain their rationale for the chosen sampling frequency to provide greater context for the reader. Accelerometers with greater than 100 Hz frequency should be used to provide more robust data regarding the dynamics in soccer. Based on the goals of the research, a sampling frequency can be selected to register suitable accelerometry-based data.
© Copyright 2021 Proceedings of the Institution of Mechanical Engineers, Part P: Journal of Sports Engineering and Technology. SAGE Publications. All rights reserved.