The author conducted the analysis using hierarchical linear regression and not not hierarchical linear modeling. The former assumes independence of at the measurement level, but the latter explicitly models non-independence. In this case measures (repeated games on the same players) are not-independent, thus hierarchical modeling should have been used to analyse the dataset. In addition whilst position and match location can be viewed as time-independent predictors (i.e. level 2 predictors), vVO2max can be considered as a time-varying predictor , that is it changes throughout the measurement period. To what extend do preseason differences in vVO2max “explain” differences during late in-season games? vVO2max could have been measured at repeated time-points (i.e preseason, post preseason, midseason, late inseason) within the season and enter the model as a time-varying predictor. Alternatively due to the maximal effrot required to obtain reliable vVO2max values, sub-maximal fitness indices could have been used.
[…] Contextual factors and aerobic fitness influence match running performance in elite soccer Via Sports Performance & Science Reports […]
[…] Contextual factors and aerobic fitness influence match running performance in elite soccer […]
The author conducted the analysis using hierarchical linear regression and not not hierarchical linear modeling. The former assumes independence of at the measurement level, but the latter explicitly models non-independence. In this case measures (repeated games on the same players) are not-independent, thus hierarchical modeling should have been used to analyse the dataset. In addition whilst position and match location can be viewed as time-independent predictors (i.e. level 2 predictors), vVO2max can be considered as a time-varying predictor , that is it changes throughout the measurement period. To what extend do preseason differences in vVO2max “explain” differences during late in-season games? vVO2max could have been measured at repeated time-points (i.e preseason, post preseason, midseason, late inseason) within the season and enter the model as a time-varying predictor. Alternatively due to the maximal effrot required to obtain reliable vVO2max values, sub-maximal fitness indices could have been used.