LAUSR

OBJECTIVE To assess whether feeding questionnaire responses reflect observed mealtime behavior. DESIGN Cross-sectional associations between self-reported and observed behaviors. SETTING Participants' homes. PARTICIPANTS Parents (n = 75) of toddlers (mean age = 24.7 months) in the US. MAIN OUTCOME MEASURES Feeding behavior questionnaires and coded videos of children's dinner meals. ANALYSIS Parents' questionnaire responses of "never" (or "rarely") considered consistent with video observation if behavior was not observed; responses of "always" (or "most of the time") if behavior observed at least once. Proportion (%) of participants observed performing each behavior was calculated for the groups of parents reporting that they "never," "sometimes," or "always" used that feeding practice. These were compared across the 3 response groups. RESULTS Parents reported 6 behaviors consistently (≥70% agreement): allowing child to eat as much as wanted, helping child eat, prompting child to eat, television/screens on during meal, nonfood rewards, and hurrying child. The remaining 8 behaviors fell below the threshold. For many behaviors, all response groups (never, sometimes, always) had similar rates of participants demonstrating the behavior. Only 5 behaviors had observed rates falling in the expected direction (frequency of always > sometimes > never). For some behaviors, the "sometimes" group had a higher (eg, clean plate) or lower (praise) frequency than the other 2 groups. CONCLUSIONS AND IMPLICATIONS Self-reported questionnaire responses predicted whether some, but not all, behaviors were observed. Parents' use of "sometimes" remains difficult to interpret as parents may use "sometimes" inconsistently across behaviors and perhaps to mitigate socially undesirable responses. Self-reports of "sometimes" performing a behavior may have limited utility for prediction of behavior and likely requires additional exploration with the respondent.