Ellen Hamaker (Utrecht University)
Dynamic Multilevel Modeling for Dyadic Data
Intensive longitudinal data such as daily diaries or observational measurements from dyads allow us to investigate the reciprocal dynamics that underlie dyadic fluctuations over time. A promising approach to such data is based on defining a dynamic model at the within-dyad level, and allowing for differences between dyads at the between level. Such multilevel models have been referred to as dynamic multilevel modeling.
In this talk I introduce this modeling framework and discuss two applications to dyadic data. The first application is based on daily diary data from couples in which they evaluated their general affect, but also their relationship specific affect. We investigate whether yesterday’s scores are predictive of today’s scores on these variables and whether there are indications that spouses affect each other’s affect. The second application is based on Gottman’s Equations of Marriage. In this approach the effect of one spouse on the other depends on the experience of the affected person. We use a threshold autoregressive model for this and extend it into a dynamic multilevel model. These examples illustrate the potential of dynamic multilevel modeling for studying dyadic interactions.