Good and Bad Eaters: A Latent Profile Analysis of Eating Behavior in Adolescents

TITLE: Good and bad eaters: A latent profile analysis of eating behavior in adolescents

ABSTRACT BODY:

Introduction: The data for this study are taken from the ORBIT (Habitual and Neurocognitive Processes in Adolescent Obesity Prevention) grant which aims to develop an intervention to improve adolescents’ eating behavior and reduce their risk for obesity. Part of this process includes researching which constructs are related to and can predict poor nutrition behavior and obesity. Analyses with this data have suggested that among adolescents there may exist different latent (unobserved) classes of eating behaviors. Latent profile analysis (LPA) can be used to uncover these latent classes. In particular, the Youth/Adolescent Questionnaire (YAQ) (a measure of food frequency) appears to be a good candidate to provide indicators for these latent classes.

Methods: Several LPA models will be run in Mplus 6. Individual items, as well as subscales from the YAQ, and possibly other neurocognitive or demographic measures from the data set may be included as indicators. Some of these items may also be investigated as covariates that determine conditional class membership. Number of classes for the final models will be determined by comparing the Pearson chi-square statistic, as well as average class assignment probabilities among models.

Results: Preliminary results have suggested the presence of at least two classes of adolescents: those who consume large amounts of unhealthy snacks and beverages, and those who do not. The class profiles of the final model will be presented, along with explanations of the model parameters and measures of data fit.

Conclusions: These results may help obesity researchers plan interventions to improve adolescents’ food intake. Evidence of qualitatively different classes of eating behaviors would help researchers be able to plan interventions specifically tailored to individuals based on their class membership. Also, indicators that show high class discrimination may be used as a screening tool to identify those adolescents at risk for membership in a class that consumes many unhealthy snacks.