Evaluation of parental reports as a proxy for their children's food preferences

As childhood obesity rates rise, research on children's food preferences gains importance. However, studies on these preferences have predominantly relied on parental reports. Research assumes that parents are accurate informants, given their role as food providers. However, excluding children introduces limitations, as parents may misinterpret or overestimate their children's food preferences. Conversely, children's self-reports are hampered by cognitive and language challenges. Leveraging an online survey, this study examines whether children's reports of their food preferences align with those reported by their parents. A diverse range of statistical methods was implemented to examine agreement in 510 parent-child dyads. These included Spearman's rho, intra-class correlation coefficients, Multitrait-Multimethod analysis complemented with a confirmatory factor analysis (MTMM-CFA), and linear regressions. There were strong correlations across food-related constructs, indicating robust consensus (r ≥ 0.60). MTMM-CFA model comparison analyses also demonstrated strong convergent (r ≥ 0.61) and discriminant validity (r ≤ 0.56), with both challenger models underperforming compared to the focus model (CFI ≤0.48). Moreover, the results replicated previous findings from the literature and found participants' reports to be predictive of downstream food consumption. This provided evidence of good and comparable predictive validity of children's eating habits for both parents and children. The findings reported in this paper suggest that children's responses align well with parental assessments. In consequence, whether to prioritise children's or parents' reports may depend on the research aim and logistical or ethical considerations, seemingly without compromising data quality.