Longitudinal associations between parent–child attachment and prosocial behaviour: the roles of empathy and gratitude
- 1School of Education and Psychology, Shaoxing University , ,
- 2Shaoxing Changcheng Middle School , ,
Abstract
Background: Parent–child attachment is assumed to support adolescent prosocial behaviour through empathy and gratitude, but most evidence is cross-sectional and rarely separates stable between-person differences from within-person change. This study examined whether longitudinal associations among these constructs match the proposed sequential account and whether they hold within persons.
Methods: Chinese junior secondary students (N = 557 linked across three waves) completed an identical self-report battery on all four constructs. Longitudinal measurement invariance was tested for each construct. A random-intercept cross-lagged panel model (RI-CLPM) separated between-person from within-person variance, and a latent sequential indirect-effect model with autoregressive controls was estimated by full-information maximum likelihood with 5,000 bias-corrected bootstrap resamples. Two prespecified robustness analyses varied the placement of gratitude and excluded outliers.
Results: Metric invariance held for all constructs; full scalar invariance failed for prosocial behaviour (ΔRMSEA = 0.021). Random intercepts held 64.5 to 70.9% of total variance, and no hypothesised within-person cross-lagged path was significant (attachment to empathy β = 0.076, p = 0.561; empathy to gratitude β = 0.140, p = 0.160; gratitude to prosocial behaviour β = −0.043, p = 0.694). In the latent model (CFI = 0.974, RMSEA = 0.046), attachment predicted empathy (B = 0.100, 95% CI [0.032, 0.163]), empathy predicted gratitude (B = 0.171 [0.015, 0.329]) and gratitude predicted prosocial behaviour (B = 0.376 [0.241, 0.545]). The sequential indirect association was small (B = 0.0064 [0.0010, 0.0202]; 5.8% of the total indirect association), was not reproduced when gratitude was placed at Wave 2, and its empathy-to-gratitude component lost significance after outliers were excluded. The two prospective two-link within-person indirect products identifiable in the RI-CLPM were also not distinguishable from zero.
Conclusion: The data support stable longitudinal associations, particularly those involving gratitude, but only tentative evidence for a sequential pathway and none for a within-person developmental chain. A four-wave design placing each construct on a separate occasion is needed to test the mechanism.
Related
- Associations of individual exposure to particulate matter with next-day physical activity and sedentary behaviour in older adults: a micro-longitudinal study
- Longitudinal effects of compassion training on emotion regulation and prosocial moral development among adolescents
- Latent Profiles of Primary Emotional Systems and Their Associations with PTSD, CPTSD, Attachment, and Childhood Adversity.
- Associations of pet ownership and pet attachment with a metabolite-based psychosocial distress score and circulating metabolomic profiles
- Gender differences in problematic pornography use: associations with personality traits and attachment styles.
BibTeX
@article{chen2026longitudinal,
title = {{Longitudinal associations between parent–child attachment and prosocial behaviour: the roles of empathy and gratitude}},
author = {Xiaoxue Chen and Jia Xu and Anni Chen and Leigang Zhang},
journal = {Frontiers in psychology},
year = {2026},
month = oct,
doi = {10.3389/fpsyg.2026.1970907},
url = {https://europepmc.org/article/PMC/PMC13730822},
note = {CC BY},
}
RIS
TY - JOUR
TI - Longitudinal associations between parent–child attachment and prosocial behaviour: the roles of empathy and gratitude
AU - Xiaoxue Chen
AU - Jia Xu
AU - Anni Chen
AU - Leigang Zhang
PY - 2026
DA - 2026/10/21
JO - Frontiers in psychology
DO - 10.3389/fpsyg.2026.1970907
UR - https://europepmc.org/article/PMC/PMC13730822
AB - Background: Parent–child attachment is assumed to support adolescent prosocial behaviour through empathy and gratitude, but most evidence is cross-sectional and rarely separates stable between-person differences from within-person change. This study examined whether longitudinal associations among these constructs match the proposed sequential account and whether they hold within persons. Methods: Chinese junior secondary students (N = 557 linked across three waves) completed an identical self-report battery on all four constructs. Longitudinal measurement invariance was tested for each construct. A random-intercept cross-lagged panel model (RI-CLPM) separated between-person from within-person variance, and a latent sequential indirect-effect model with autoregressive controls was estimated by full-information maximum likelihood with 5,000 bias-corrected bootstrap resamples. Two prespecified robustness analyses varied the placement of gratitude and excluded outliers. Results: Metric invariance held for all constructs; full scalar invariance failed for prosocial behaviour (ΔRMSEA = 0.021). Random intercepts held 64.5 to 70.9% of total variance, and no hypothesised within-person cross-lagged path was significant (attachment to empathy β = 0.076, p = 0.561; empathy to gratitude β = 0.140, p = 0.160; gratitude to prosocial behaviour β = −0.043, p = 0.694). In the latent model (CFI = 0.974, RMSEA = 0.046), attachment predicted empathy (B = 0.100, 95% CI [0.032, 0.163]), empathy predicted gratitude (B = 0.171 [0.015, 0.329]) and gratitude predicted prosocial behaviour (B = 0.376 [0.241, 0.545]). The sequential indirect association was small (B = 0.0064 [0.0010, 0.0202]; 5.8% of the total indirect association), was not reproduced when gratitude was placed at Wave 2, and its empathy-to-gratitude component lost significance after outliers were excluded. The two prospective two-link within-person indirect products identifiable in the RI-CLPM were also not distinguishable from zero. Conclusion: The data support stable longitudinal associations, particularly those involving gratitude, but only tentative evidence for a sequential pathway and none for a within-person developmental chain. A four-wave design placing each construct on a separate occasion is needed to test the mechanism.
SN - 1664-1078
ER -