Test-retest reliability and comparison of children’s reports with parents’ reports of young children’s fruit and vegetable preferences.

Abstract
The aim of this study is to investigate the test-retest reliability of a short computerized assessment of young children’s fruit (F) and vegetable (V) preferences, and to compare children’s responses with their parents’ responses. A paper-and-pencil F and V preference and F and V food frequency questionnaire was completed by 194 parents. Data on 139 preschoolers was available for test and retest of F preferences and data on 135 children for V. F and V preference scales were computed, including the ten most commonly consumed F and the ten most commonly consumed V. Alpha reliabilities were good (F=0.78; V=0.76) and test-retest intraclass correlations were high (ICC for F=0.74; V=0.75). Agreement between parents’ and children’s reports was moderate (F: ICC=0.48, V: ICC=0.41). Children reported more often to have never consumed an item, and less often the midcategory « not yummy, not yucky but ok ». Covariance analysis indicated differences in agreement by school and mothers’ education level with a lower agreement for those of lower social status. A weak, but significant association was found between children’s reported preferences and children’s consumption (F: r=0.19, V: r=0.25). The results are promising, but additional validation is needed in a representative sample and should further explore the sources of disagreements.