Change Processes in Ecological Momentary Assessment: Addressing Zero-Inflation in Count Data

Count or zero-inflated count data is often collected in Ecological Momentary Assessment (EMA) studies where understanding the mechanisms of the change process serves as the primary research goal. Traditional methods often fail to capture the complexities inherent in EMA data, leading to discrepancies between empirical findings and theoretical expectations. To address these limitations, this study presents two novel approaches. The first integrates autoregressive effects individually, and the second utilizes a model framework to pinpoint autoregressive effects among groups.
These approaches are extended for both count and zero-inflated count data. Through simulation studies and empirical applications, the proposed models show enhanced accuracy and interpretability at both individual and group levels. Furthermore, models tailored for zero-inflated data, referred to as ZIP-CAR, can distinguish zero patterns at both individual and group levels. The dissertation concludes with discussions on the practical implications, limitations, and future directions of the proposed methods. This work is expected to improve method development for EMA studies, ultimately enhancing the understanding of behavioral change processes in zero-inflated count data.