Family- and classroom-level effects on the accuracy of self-reported weight among school-attending children

Objectives: The comparison of direct and self-reported measures of weight has been examined in several studies, but few have probed such issues in underage population. In this study, we extend this focus to children and seek to evaluate possible contextual influences in under-report and non-response of bodyweight.

Methods: The sample was composed of approximately 2000 4th and 6th grade pupils recruited from 28 public elementary schools in the northern Taiwan in 2006-2007. Information on individual and classroom/teacher characteristics was collected via paper-and-pencil self-administered questionnaires; anthropometrics on weight and height were retrieved from health records. Gender-specific multilevel analyses were performed to assess individual-, family-, and classroom-level effects on self-report accuracy.

Results: Roughly 1 in 5 children had underreported bodyweights (ie,³a1 kg difference) and the estimate for non-response was 31% in both genders. For girls, older ages, the experience of menarche, having taller height-rank, and heavier weight-rank were associated with bodyweight underreport; a reduced underreport was found in female pupils having at least one same-sex siblings in family (OR=0.44~0.65). Similar association patterns appear in boys while sibling effects seem less significant. Moderate intra-class correlations surface in class-level (ICC=0.2 for boys, 0.23 for girls); children from classes mentored by male, younger-aged teachers or with higher class-average weights tend to underreport.

Conclusion: These findings suggest the accuracy of children's self-reported bodyweight might differ not by only individual characteristics but also by the living contexts. Further research on possible mechanisms is needed to improve the validity of self-reported information.