Assessing dynamical associations in dyadic interactions across multiple time scales via a Bayesian hierarchical vector autoregressive model

Abstract

Human behaviors are complex and composed of changes on multiple time scales. Recent advances in data collection technology contribute to a fast-growing number of studies with rich and intensive longitudinal data, allowing researchers to examine the underlying change processes on the time scale(s) of their choice. Processes unfolding across different time scales can be interrelated in different ways. One way is through reflecting the same underlying construct. For example, attachment styles in couples may lead to a pattern of dyadic coregulation that is reflected both in their physiological synchrony and in their daily affective coherence. Although previous research has examined romantic relationships both from a physiological and an affective perspective, the association between the two has seldom been formally evaluated. In this article, we describe a hierarchical Bayesian vector autoregressive model that enables researchers to examine whether two processes are associated by having coherent patterns. We demonstrate the specification and implementation of this model using data on two different processes between romantic partners: their second-by-second physiological synchrony and daily affect coregulation.