We examined pair-wise Spearman correlation tests between all variables to check for multicollinearity; if variables were correlated at rs≥0.70, only the variables with the lower Akaike’s Information Criterion value were included in further analyses to reduce redundancy [32–34].