Unraveling behavioral sequences in family interactions: A dynamic network approach to binary time series

Micro-level family interaction processes are considered to be the engine of child development and (re-)adaptation. To study the dynamics of these interaction patterns in moment-to-moment exchanges between parents and their children, we developed JacNet. This data-analysis approach quantifies the relative frequencies and the contingency of different behaviors over time and depicts them in an easy-to-grasp network picture. In this proposal we want to extend the approach towards multi-step processes and develop a test to directly compare the strength of the links between different behaviors in the network. Further, we aim to address high-dimensionality and non-stationarity. To make the approach more widely available, we will extend the method to event-based coded data and develop easy-to-handle software.

Keywords:dynamic network, family interactions, binary time series, micro-coded data

Disciplines:Statistics and data analysis