An increasingly popular form of data collection in health psychology research is Ecological Momentary Assessment (EMA); that is, using diaries or smartphones to collect intensive longitudinal data. This method is increasingly applied to the study of relationships between state-based aspects of individuals’ functioning and health outcomes (e.g., binge eating, alcohol use). Analysis of such data is challenging and regression tree modelling (RTM) may be a useful alternative to multilevel modelling for investigating the association between a set of explanatory variables and a continuous outcome. Furthermore, RTM outputs ‘decision trees’ that could be used by health practitioners to guide assessment and tailor intervention. In contrast to regression, RTM is able to easily accommodate many complex, higher-order interactions between predictor variables (without the need to create explicit interaction terms). These benefits make the technique useful for those interested in monitoring and intervening upon health and psychological outcomes (e.g., mood, eating behaviour, risky alcohol use, and treatment adherence). Using real data, this paper demonstrates both the benefits and limitations of RTM and how to extend these models to accommodate analysis of nested data; that is, data that arise from EMA where repeated observations are nested within individuals.
- decision tree
- ecological momentary assessment
- experience sampling
- Regression tree analysis
- tailored intervention