Application of EMA Data to an Obesity Study

The Habitual & Neurocognitive Processes in Adolescent Obesity Prevention is a project consisting of a series of five studies that will culminate in the creation of a novel intervention strategy to improve nutrition behavior and reduce risk for obesity among adolescents.  The first of these five studies seeks to identify cues to habitual dietary behavior among adolescents through the use of Ecological Momentary Assessment (EMA) data.   One of the major goals of the study is to determine the cues that precede and predict different dietary choices in adolescents with the ultimate goal of developing interventions for poor nutritional choices based on these cues.   Participants used PDA-like devices to track their food consumption and other contextual variables over a 7 day period.  In addition to general demographics, traits and daily responses, we collected two types of EMA Data: User-Prompted Responses and Randomly-Prompted Responses.  This paper will provide an overview of the study design and methodology.  The final data consist of a total of 5013 EMA observations from a total of 158 subjects over 7 days.  The data were collected in real time (EMA data) and there are at least three levels of data:  momentary observations nested in days nested in participants.  The nature of the data introduces unique methodological obstacles. I will address some of the methodological issues that arise from EMA data both specifically from this study and EMA in general. Initial findings in our data suggest that having unhealthy snacks or sweet drinks among adolescents were associated with being at school, being with friends, feeling lonely or bored, craving a drink or snack and being exposed to food cues.  These findings identify important environmental cues to poor snacking choices. Particular issues for discussion include combining data from different data types, dealing with the hierarchal nature of the data, understanding the relationship of time, non-compliance and missing data.