Uncovering Asymmetric Temporal Dynamics Using Threshold Dynamics Parameters

Structural Equation Modeling A Multidisciplinary Journal, vol. 32, pp. 949–965

Abstract

Statistical models to analyze longitudinal data often include parameters that capture temporal dependencies. These dynamics parameters are typically thought to operate independently of the time series value. Here, we argue that this leads to overlooking important information on psychological processes. We propose the DYNamics of ASymmetric TIme series (DYNASTI) approach, allowing dynamics parameters to differ above and below the time series mean. Through extensive simulations, we show that DYNASTI implementations of two commonly-used time series models (DSEM and RI-CLPM) adequately recover symmetric and asymmetric temporal dynamics. Importantly, we also show that assuming symmetric dynamics (as in the vast majority of the literature) when processes are in fact asymmetric leads to incorrect conclusions about these dynamics. We further illustrate how DYNASTI implementations can lead to new insights in three empirical examples. We believe asymmetric dynamics are widespread and hope, by providing open and easy-to-apply code, to aid researchers in uncovering them.

Authors 5

1. Jessica Vera Schaaf corresponding

   Radboud University Medical Center

   Affiliation as printed

   Radboud University Medical Center

2. Øystein Sørensen

   University of Oslo

   Affiliation as printed

   University of Oslo

3. Ethan Michael McCormick Aachen

   Leiden University

   Affiliation as printed

   Leiden University

4. Michael Aristodemou

   Radboud University Medical Center

   Affiliation as printed

   Radboud University Medical Center

5. Rogier Kievit

   University of Cambridge · Radboud University Medical Center

   Affiliation as printed

   Radboud University Medical Center

   University of Cambridge