Cognitive, Behavioral, and Social Factors Are Associated with Bias in Dietary Questionnaire Self-Reports by Schoolchildren Aged 9 to 11 Years
Accepted 11 April 2008.
Abstract
Background
Measuring children's dietary behavior is central to evaluating interventions and identifying predictors and outcomes of dietary behaviors. Systematic biases may obscure or inflate associations with self-reported intakes.
Objective
To identify cognitive, behavioral, and social correlates of bias in children's reporting of breakfast items on a self-completion questionnaire.
Design
Cross-sectional survey. Children completed standardized tests of episodic memory, working memory, and attention, and a questionnaire assessing attitudes toward breakfast. Teachers completed a classroom behavior measure. Associations between measures and children's underreporting of breakfast foods (ie, cereals, bread, milk, fruits, sweet items, and potato chips) on a self-completion questionnaire relative to validated 24-hour recall were examined.
Subjects and setting
Subjects were aged 9 to 11 years (n=678). Data were collected from 111 schools throughout Wales in 2005.
Results
A larger percentage of less-healthful breakfast items (ie, sweet snacks and potato chips) than more healthful items (ie, fruit, cereals, bread, and milk) were omitted from questionnaire self-reports. Children from lower socioeconomic status schools omitted more items than those from wealthier schools (Kruskal-Wallis H=12.51, P<0.01), with omissions twice as high for less-healthful items than for more-healthful items within the lowest socioeconomic status schools. Those with positive attitudes (H=23.85, P<0.001), better classroom behavior (H=6.26, P<0.05), and better episodic memory (H=8.42, P<0.05) omitted fewer items than those with negative attitudes, poorer behavior, and poorer episodic memory. Children who ate more items omitted more than those who ate fewer (H=47.65, P<0.001). No differences were observed in terms of attention and working memory.
Conclusions
Episodic memory, classroom behavior, attitudes, socioeconomic status, and total items consumed are associated with bias in questionnaire self reports. Such biases have implications for examination of associations between breakfast eating and cognitive and behavioral factors, examination of effect modification by socioeconomic status in intervention trials, and for the sensitivity of measures to detect intervention effects.
Address correspondence to: Graham F. Moore, MSc, Cardiff Institute of Society, Health and Ethics, Cardiff University, 53 Park Place, Cardiff, Wales, UK CF10 3AT
ABSTRACT