In order to establish calibration factors of the Adolescent Food Frequency Questionnaire (AFFQ), 74 boys and girls from Piracicaba (SP, Brazil) with ages ranging from 10 to 14 took part in the study. Dietary intake (assessed by the food frequency questionnaire and 24-hour recall) was assessed and adjusted for energy intake. Descriptive statistics, variance analysis using one classification factor, Pearson's correlation coefficients and linear regression were performed. The average of two 24-hour recalls was used as a reference for calibration of data. Calibration coefficients (λ) ranged from -0.07 (iron) to 0.40 (vitamin C) revealing substantial error in the dietary method tested, albeit similar to those observed in the literature. As these coefficients were low, they indicate the need for reformulating the instrument regarding some nutrients, though application was not considered advisable for correcting information on iron and retinol. The methodology used to calibrate dietary data can consider measurement error in the assessment when its assumptions are respected, since violations of these assumptions may lead to other errors that are difficult to predict.
Calibration; Diet; Measurement error; Linear regression
