Objective: Guided by theory, this study explored cross-sectional differences in factors influencing adolescent eating behaviour including gender and socio-economic status (SES), and subsequently tested the longitudinal predictive power of the models.
Design/setting/subjects: Data were collected by questionnaires in a longitudinal study of adolescents (age 13 years at baseline) and their parents from Hordaland County, Norway. Association of personal and environmental variables (family, friends, school/society) with the consumption of fruit and vegetables (FV) and selected sources of fat and of sugar were assessed at age 15 The final cross-sectional models were subsequently employed in groups stratified by gender/SES and to predict consumption at age 21
Results: The model explained more of the variation in the sugar score (21%) and the FV score (13.5%) than in the fat score (5%). SES was associated with both the sugar and FV scores. The strongest associations with the sugar score and FV were for antisocial behaviour and evaluation of own diet, respectively. The former association was significant in all gender/SES groups, whereas the latter association was only significant in the low SES groups. For all three types of food, the strongest significant predictors in the longitudinal models were frequency of consumption at age 15.
Conclusion: The model's ability to explain variation in eating behaviours differed by food type, and possibly by gender/SES, but previous eating behaviour was an important predictor for all three foods. Prospective studies should carefully operationalize theoretical constructs when further investigating the influences of and interrelationships between these factors and gender/SES on the development of eating behaviours.