Parallel processing modeling in longitudinal designs: An example predicting trajectories of distress and life satisfaction.

Abstract

Parallel process modeling (PPM) can be used to analyze co-occurring relationships between health and psychological variables over time. A demonstration is provided using data obtained from the British Household Panel Survey (years 2005, 2006, 2007, and 2008), examining predictors of ongoing changes in their distress and life satisfaction of a subsample from the survey. In the 2005 survey, data were available from 7,970 participants based on the following demographic variables: gender, age, ever registered as disabled, and ever experienced any strokes (before or at 2005). Time-varying variables included distress and life satisfaction collected yearly from 2005 to 2008. Time-invariant variables included age (65 or older), gender, disability condition, and stroke survivor status. Steps of fitting the PPM are presented. Four distinct distress trajectory groups-chronic, recovery, delayed, and resilient-were identified from the PPM estimates. Resilient and recovery groups showed a positive trend in life satisfaction. The delayed distress and chronic groups had a slight decrease in satisfaction. The time-invariant covariates only significantly predicted baseline levels of distress and satisfaction (i.e., their intercepts). PPM is a relatively simple and powerful tool for simultaneously studying relations between multiple processes. A step-by-step approach on decomposing the significant predictive relation from the change of distress to the change of satisfaction is presented. Properly decomposing any significant growth factor regressed on another growth factor is necessary to fully comprehend the intricate relationships within the results. Practical implications and additional methodological information about fitting PPM are discussed. (PsycInfo Database Record (c) 2024 APA, all rights reserved).