A Multilevel Logistic Regression Analysis on the Likelihood of Overeating and Unplanned Eating

Abstract

This paper analyzed three models of emotional eating: the restraint disinhibition model, the affect-regulation model, and the externality model (Herman and Polivy, 1975; Haedt-Matt & Keel, 2011; Schachter, 1968). Emotional eating is a risk factor that contributes to the probability that a person will overeat or have an unplanned meal, which can contribute to weight gain and obesity (Goldbacher et al., 2012). Using multilevel logistic regression in Hierarchical Linear and Nonlinear Modeling (HLM) software, two-level models were created with overeating and unplanned eating as the dependent variables (Version 7; Raudenbush, Bryk, & Congdon, 2010). All three models provided valid frameworks for understanding the risk factors associated with overeating and unplanned eating. Level 1 of the models included situational factors (e.g. negative mood, place, people, and type of meal), which were analyzed to understand how context affected the probability of overeating and unplanned eating. Level 2 of the models included within-person characteristics (e.g. BMI, age, race), which were analyzed to understand how individual demographics affected the probability of overeating and unplanned eating.