Title: Multilevel models for longitudinal dyadic family data (in-person)
Authors: Fiona Steele - London School of Economics (United Kingdom) [presenting]
Abstract: The family is one of the most important examples of a social network. There is substantial interest in studying family interactions to understand child development and, in later life, exchanges of intergenerational support between adult children and their parents. In their simplest form, these data have a dyadic structure with a bivariate response describing bidirectional interactions between the members of each dyad. However, dyadic family data often have a more complex structure: data are increasingly longitudinal, dyads may be nested within families, and a family member may belong to multiple family dyads leading to a cross-classified structure. Multilevel models offer a flexible way of analysing complex longitudinal and multivariate data from dyadic designs. We consider random-effects models for longitudinal dyadic data collected under two different designs: a round robin study of the behaviour observed between each pair of family members over the course of a task, and household panel data on the support that mothers and fathers provide to and receive from their adult children.