Using structural equation modeling to study traits and states in intensive longitudinal data.

Abstract

We wish to thank prof. dr. Peter de Jonge and the team of HoeGekisNL, as well as the participants, for allowing us to use their data in this study. We would also like to thank the Center for Information Technology of the University of Groningen for their support and for providing access to the Peregrine high performance computing cluster.Partial results of this study were presented at the measurement error in longitudinal data workshop held in the university of Manchester in June 2019 and at the IMPS 2019 meeting in Santiago de Chile. Additionally, all the code developed during this research is available in the following GitHub repository: https://github.com/secastroal/LST_Analyses. We are also very grateful for the helpful comments provided by four anonymous reviewers. Finally, we are especially thankful to dr. Markus I. Eronen and María Angélica Acevedo-Mesa, who help us to proofread the revised manuscript.

Mixed-Effects Trait-State-Occasion Model: Studying the Psychometric Properties and the Person–Situation Interactions of Psychological Dynamics

“…The ME-TSO model presented in this study aims to be an additional tool to model psychological dynamics. The model integrates the multilevel TSO (Castro-Alvarez et al, 2021), the LST-RF approach (Geiser et al, 2015), and the DSEM framework (Asparouhov et al, 2018). In general, the ME-TSO model allows studying the carry-over effects of psychological constructs, the person-situation interaction effects, and the psychometric properties of the items per individual across fixed situations.…”

“…This means that it has been presented as a single-level model, which requires the data to be in wide format (i.e., one row per subject and one column for each repeated measure). However, we have previously presented a multilevel version of the TSO model (Castro-Alvarez et al, 2021), the path diagram of which is shown in Figure 1. In contrast to the single-level TSO model, the multilevel TSO model requires the data in long format and is therefore more suitable to analyze intensive longitudinal data.…”

“…In other words, the multilevel TSO model assumes that the measurement of the construct of interest (e.g., positive affect) over time does not change. Lastly, the multilevel TSO model is easier to apply and estimate than the single-level TSO model because it has a smaller number of free parameters and it allows to easily handle many measurement occasions and missing data (Castro-Alvarez et al, 2021;Geiser, 2020;Geiser & Lockhart, 2012).…”

“…Hence, optimism was added to further explain the person-situation interactions. Note that we have previously analyzed these data with the TSO model, without considering random autoregressive effects or person-situation interaction effects (Castro-Alvarez et al, 2021).…”

“…As mentioned before, the ME-TSO model requires that multiple items measure the same construct. This is the case in the HowNutsAreTheDutch data with the items of positive affect activation (Castro-Alvarez et al, 2021). Before fitting the models, we described and visualized the raw data.…”