Modeling Nonstationary Emotion Dynamics in Dyads using a Time-Varying Vector-Autoregressive Model

Area of Science:

• Psychology
• Computational Social Science
• Statistical Modeling

Background:

• Interpersonal emotion dynamics are complex and often assumed to be static in traditional research.
• Stationarity assumption in standard models limits the accurate capture of evolving emotional relationships.
• Real-world emotional changes can be triggered by internal or external factors, necessitating dynamic analytical approaches.

Purpose of the Study:

• To introduce and validate a semi-parametric time-varying vector-autoregressive (TV-VAR) model for analyzing dynamic interpersonal emotions.
• To demonstrate the superiority of the TV-VAR model over traditional time-invariant models in capturing temporal dependencies.
• To apply the TV-VAR model to real-world couple data to detect changes in emotion dynamics.

Main Methods:

• Development and implementation of a semi-parametric time-varying vector-autoregressive (TV-VAR) model in R.
• Conducting a simulation study to compare TV-VAR with standard time-invariant vector-autoregressive (VAR) models.
• Applying the TV-VAR model to daily positive affect (PA) data from a couple.

Main Results:

• The simulation study confirmed the TV-VAR model's superior performance when emotional dynamics change over time.
• Analysis of couple data revealed significant changes in the male's emotion dynamics.
• No significant changes in the female's emotion dynamics were detected, independent of her or her partner's affect.

Conclusions:

• The TV-VAR model effectively detects and models changes in interpersonal emotion dynamics.
• This approach provides a more realistic framework for studying evolving emotional relationships in couples.
• The findings highlight the utility of dynamic modeling in understanding nuanced interpersonal emotional processes.