摘要：Principal components analysis (PCA) is a widely used techniquein nutritional epidemiology, to extract dietary patterns. To improve theinterpretation of the derived patterns, it has been suggested to rotate theaxes de ned by PCA. This study aimed to evaluate whether rotation inuencesthe repeatability of these patterns. For this reason PCA was appliedin nutrient data of 500 participants (37 15 years, 38% male) who werevoluntarily enrolled in the study and asked to complete a semi-quantitativefood frequency questionnaire (FFQ), twice within 15 days. The varimax andthe quartimax orthogonal rotation methods, as well as the non-orthogonalpromax and the oblimin methods were applied. The degree of agreementbetween the similar extracted patterns by each rotation method was assessedusing the Bland and Altman method and Kendall's tau-b coecient. Goodagreement was observed between the two administrations of the FFQ forthe un-rotated components, while low-to-moderate agreement was observedfor all rotation types (the quartimax and the oblimin method lead to morerepeatable results). To conclude, when rotation is needed to improve foodpatterns' interpretation, the quartimax and the oblimin methods seems toproduce more robust results.
关键词：Multivariate analysis; principal components analysis; repeatability;rotation type.
