Modeling Psychological Processes Over Time: Leveraging Theoretical Knowledge and Data Complexities

Abstract

Psychological processes are characterised by variation over time. In order to accurately study these processes, it is necessary to measure them in an appropriate way—repeatedly over time—for example, using daily diaries or momentary assessments. Researchers who collect such data are often interested in the dynamic relationships between variables. For example, how a person's current mood relates to their own or their partner's mood a moment earlier. However, the complexity of psychological processes and data means that most dynamic models are not suitable for accurately capturing the behavior under study. In her dissertation, Sophie Berkhout shows how theoretical knowledge and complexities in data can indicate why a model is or is not suitable for studying psychological processes. She further argues that these can serve as a guide for the development of more suitable models. She introduces a Shiny app that shows the model-implied behavior of different types of dynamic models using simulated data and visualizations, to help researchers understand what kind of behavior these models can capture. Furthermore, she developed new dynamic models that take into account night gaps in data collection when participants are sleeping and variables measured at different frequencies. Finally, she presents multilevel extensions to study both within-person and between-person differences in psychological processes.