Corrigendum to: “FreeSurfer subcortical normative data” [Data in Brief 9 (2016) 732–736]

摘要

We have recently uncovered a flaw in our statistical method that impacts the semi-partial R? results presented in Figure 1. These values were computed using the option Effect size within the GLM procedure of the SAS statistical software and are labeled by SAS as semi-partial eta squares. These values are, in fact, a partitioned R? according to a given order of predictors’ entry. The main impact of this procedure is that the order of the predictors influences the semi-partial R?. While it does not impact the regression models, for those readers that are drawing conclusions based on the relative importance of the predictors, we felt compelled to provide more accurate and robust results.