Abstract

Time-varying coefficient modeling (TVCM), which represents regression coefficients as smooth functions of continuous time, provides a flexible framework for uncovering complex patterns of change in levels and associations among variables in intensive longitudinal data. However, conventional TVCM remains limited to investigating directional effects across individuals. By introducing a TVCM formulation of the multivariate normal distribution, the present study extends TVCM to explore temporal dynamics in undirected associations (couplings) and variability, thereby broadening its utility for psychological research. We discuss three implementations: an aggregate-level model and two hierarchical implementations capturing interindividual differences in temporal dynamics, either via person-specific intercepts accounting for differences at the onset of the period of interest or fully person-specific coefficient functions smoothed via partial pooling. To illustrate the model developments, we apply them to six weeks of intensive longitudinal data from 16 anxiety patients undergoing therapy, examining heterogeneity in changes in the level and volatility of negative emotions and metacognitions, as well as in the temporal dynamics of their coupling. We further show how inspecting first-order derivatives of the coefficient functions enables the identification of periods of stability and change in symptom levels, volatility, and couplings. Finally, we discuss extensions incorporating person-level characteristics to explain heterogeneity in temporal dynamics and predict outcomes (e.g., therapeutic success).