Bland-Altman plots of OBSERVATION vs. INTERVIEW (n = 26) for energy (panel A), carbohydrate (CHO, panel B), and protein (PRO, panel C) intake. The dashed lines represent the mean bias and 95% limits of agreement of the raw data. Correlation results between absolute values of the means vs. differences (indicating non-randomness of errors) and recalculated mean bias ratios (with 95% limits of agreement, LOA) based on transformed data are also shown for each comparison.
Baker et al. Nutrition Journal 2014 13:41 doi:10.1186/1475-2891-13-41