Integrating Multiple Timescales and Process Models: An Illustrative Application to Affect Dynamics
Open Access
- Author:
- Wood, Julie Katharine
- Graduate Program:
- Human Development and Family Studies
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- February 20, 2018
- Committee Members:
- Nilam Ram, Thesis Advisor/Co-Advisor
Zita Oravecz, Committee Member - Keywords:
- Process modeling
Multiple Timescales
Intensive longitudinal data
Affect Dynamics - Abstract:
- Process models are robust tools for developing, testing, and refining models of psychological processes. However, process models are typically applied only to intensive data collected over relatively short periods of time (seconds, minutes). A key postulate of many developmental theories is that these short-term processes also change over longer time scales (months, years). Measurement burst designs combine collection of intensive longitudinal data that provides for description of short-term process with collection of longitudinal panel data that provide for description of how different aspects of those processes change over time. This paper illustrates an application of a multilevel process model to measurement-burst data, to examine interindividual differences and intraindividual change in intraindividual process dynamics. Ratings of affect arousal were collected from 150 adults age 18 to 90 years during three 21-day measurement bursts of intensive data collection, spaced approximately evenly over the course of a year. A 3-level autoregressive model with heterogeneous variance was used to examine how three key parameters of arousal dynamics (attractor point, inertia, and “reactivity” to biopsychosociocultural (BPSC) inputs) were related to interindividual differences and/or intraindividual changes in depressive symptoms, life events, perceived control, and relative age. Both interindividual differences and intraindividual burst-to-burst changes in depression were associated with lower arousal attractor positions. Older age (relative to others) was related to higher arousal attractor positions but less demonstrated BPSC-related reactivity. Experience of major life events was associated with higher arousal inertia, as well as higher BPSC-related reactivity at both the inter- and intraindividual level. These analysis and results illustrate how a multilevel model, cast in a process model framework, can be utilized by researchers to identify sources of inter- and intraindividual differences on multiple timescales. These kinds of models move developmental research to closer representations of plastic, multideterminant developmental theory.