Longitudinal and Latent-variable Extensions of the Triadic Actor-Partner Interdependence
Abstract
When modeling interpersonal processes within triadic systems, researchers commonly encounter two types of analytical complexity: observations collected repeatedly over time, and constructs measured via multiple items. The current study introduces longitudinal and latent-variable extensions of the standard triadic Actor-Partner Interdependence Model (tAPIM) to accommodate these data structures. The longitudinal extensions, the concurrent tAPIM and the temporal tAPIM, account for temporal dependence and temporal precedence in repeated observations, while the latent-variable extension models each member’s construct as a latent variable measured from multiple observed items. In addition, we identify measurement invariance across triad members as a challenge specific to triadic data analysis and conduct a simulation study to investigate the consequences of its violation for the estimation of actor and partner effects. The results indicate that: 1) imposing invariance constraints that are not empirically supported may bias the estimation of nonzero actor and partner effects associated with the noninvariant member, while effects of invariant members and zero effects are accurately recovered; 2) relaxing invariance constraints avoids bias in point estimates, but standard errors of effects associated with the noninvariant member were still inflated, reducing power to detect true effects. The current paper offers practical guidance for researchers modeling interdependence in triadic systems using multi-item or longitudinal data.
Related
- Smartphone behavior and loneliness among older couples in China: Actor-partner interdependence modeling
- Beyond Actor-Partner Associations: A Target-Trial Framework for Causal Spillover Effects in Longitudinal Dyadic Data
- September 2026 Working Notes on 3D Navier–Stokes Conditional Regularity Criteria, the Hodge–Navier–Stokes Architecture Referee Audit, and Triadic Multiscale Homogenization
- High-Order Triadic Functional Connectivity in the Brain and Beyond
- Local Convergence of Triadic Closures of Sparse Random Graphs
BibTeX
@misc{wang2026longitudinal,
title = {{Longitudinal and Latent-variable Extensions of the Triadic Actor-Partner Interdependence}},
author = {Shiyao Wang and Jennifer Dang Guay and Karla Van Leeuwen and Evelien Schat and Eva Ceulemans},
howpublished = {PsyArXiv},
year = {2026},
month = sep,
doi = {10.31234/osf.io/cmzha_v3},
url = {https://osf.io/preprints/psyarxiv/cmzha_v3/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Longitudinal and Latent-variable Extensions of the Triadic Actor-Partner Interdependence
AU - Shiyao Wang
AU - Jennifer Dang Guay
AU - Karla Van Leeuwen
AU - Evelien Schat
AU - Eva Ceulemans
PY - 2026
DA - 2026/09/26
PB - PsyArXiv
DO - 10.31234/osf.io/cmzha_v3
UR - https://osf.io/preprints/psyarxiv/cmzha_v3/
AB - When modeling interpersonal processes within triadic systems, researchers commonly encounter two types of analytical complexity: observations collected repeatedly over time, and constructs measured via multiple items. The current study introduces longitudinal and latent-variable extensions of the standard triadic Actor-Partner Interdependence Model (tAPIM) to accommodate these data structures. The longitudinal extensions, the concurrent tAPIM and the temporal tAPIM, account for temporal dependence and temporal precedence in repeated observations, while the latent-variable extension models each member’s construct as a latent variable measured from multiple observed items. In addition, we identify measurement invariance across triad members as a challenge specific to triadic data analysis and conduct a simulation study to investigate the consequences of its violation for the estimation of actor and partner effects. The results indicate that: 1) imposing invariance constraints that are not empirically supported may bias the estimation of nonzero actor and partner effects associated with the noninvariant member, while effects of invariant members and zero effects are accurately recovered; 2) relaxing invariance constraints avoids bias in point estimates, but standard errors of effects associated with the noninvariant member were still inflated, reducing power to detect true effects. The current paper offers practical guidance for researchers modeling interdependence in triadic systems using multi-item or longitudinal data.
ER -