2.2.2. Nutrient Pattern Analysis

For the sub-sample (*n* = 971), we explored the data using PCA with orthogonal varimax rotation and Kaiser normalization to enhance interpretability. PCA was applied to the correlation matrix due to the scale differences between nutrients. Components were retained considering the scree plot, eigenvalues greater than 1 (the average of the eigenvalues when using the correlation matrix) and interpretability. High-loading nutrients were defined as having an absolute loading of at least 0.4 and were used to interpret the component. To enhance and support the interpretation, nutrients with absolute loadings between 0.3 and 0.4 were also considered. Food items were allocated to groups corresponding to their highest PC score. The chi-square test was used to test for an association between the FCDB and PCA groupings. The Kruskal–Wallis test was used to test for differences in nutrient values between the PC groupings. The PCs identified by the nutrient analysis were termed 'nutrient patterns (NPs)'.
