PubMed1997
OBJECTIVE: To validate fourth-graders' self-reports of school lunch by comparing their reports to lunch observations, and to determine the impact on accuracy of gender, ethnicity, meal component, and time interval between eating and reporting. DESIGN: Students were randomly selected, observed eating lunch, and interviewed the same day, next day, or Monday regarding Friday's lunch. Accuracy of reporting items was determined by tallying matched foods (items reported and observed), phantom foods (items reported but not observed), and omitted foods (items not reported but observed). Accuracy of reporting amount eaten was determined by calculating absolute and arithmetic differences. SUBJECTS: Subjects were 260 students: 89 same-day, 148 next-day, and 23 Monday recalls; 59 whites (30 boys) and 201 blacks (103 boys) from four schools. STATISTICAL ANALYSES: Rates for matched, phantom, and omitted foods; analysis of variance; Friedman's nonparametric analog of analysis of variance; Student-Newman-Keuls' post hoc comparisons. RESULTS: In regard to reporting items, the respective rates for matched, phantom, and omitted foods were 84%, 5%, and 16% for same-day recalls; 68%, 13%, and 32% for next-day recalls; and 38%, 48%, and 62% for Monday recalls. Rates for omitted and phantom foods were higher for Monday recalls than for next-day recalls, which were higher than for same-day recalls (P < .05 for all). In regard to reporting amounts, analysis by gender, ethnicity, and time interval failed to yield significant main or interaction effects. When children correctly reported items eaten, they were quite accurate in reporting amounts eaten. Omitted food rates were lowest for beverage, followed by entree, and highest for miscellaneous and condiment. APPLICATIONS: Even under the best conditions (ie, reporting within 90 minutes after eating school lunch), children have difficulty accurately reporting what they have eaten. As the time interval between eating and reporting increases, accuracy decreases markedly. Techniques that improve reporting of items eaten should result in improved accuracy of reporting amounts eaten.