Related Experiment Video
Updated: Sep 9, 2026
An Application for Pairing with Wearable Devices to Monitor Personal Health Status
Published on: February 3, 2022
Within-person reciprocal associations between problematic smartphone use and exercise adherence among university
Yihe Gai1, Xia Liu2, Shuai Sun1
1Department of Physical Education, Hongyi College, Chengdu Technological University, Chengdu, China.
Background:
Problematic smartphone use (PSU) and difficulties maintaining exercise-related routines are common among university students, yet the prospective within-person relationship between PSU and self-reported exercise-adherence tendency (EA) remains unclear. Previous cross-sectional and traditional cross-lagged studies cannot cleanly separate stable between-person differences from within-person fluctuations.
Objective:
This four-wave longitudinal study examined reciprocal prospective associations between PSU and self-reported EA tendency using a random-intercept cross-lagged panel model (RI-CLPM).
Methods:
A total of 1,241 adult university students from six universities in Chengdu, China, provided valid baseline data in a four-wave offline survey conducted from March to December 2025. Longitudinal measurement invariance was evaluated with three subscale-mean indicators per construct. The primary observed-score RI-CLPM used robust maximum likelihood and full-information maximum likelihood. Directional differences were tested using time-specific within-person standardized coefficients and robust Wald tests. Sensitivity analyses included complete-case, partial-equality, multiple-indicator latent, covariate-adjusted, exerciser-only, auxiliary-variable FIML, and delta-adjusted missing-data models.
Results:
Scalar invariance was supported at the subscale-indicator level. EA-to-PSU associations were β = -0.293, 95% CI [-0.373, -0.213], β = -0.185, 95% CI [-0.271, -0.099], and β = -0.136, 95% CI [-0.208, -0.064] (all p < 0.001). The PSU-to-EA paths were β = 0.017, 95% CI [-0.067, 0.102], β = 0.027, 95% CI [-0.059, 0.112], and β = -0.064, 95% CI [-0.142, 0.014] (all p > 0.05). The joint directional test was significant, Wald χ2(3) = 59.159, p < 0.001; interval-specific differences were significant at T1-T2 and T2-T3 but not T3-T4. The within-person pattern was reproduced across alternative standardizations and sensitivity analyses. The observed-score random-intercept association was not reproduced in the multiple-indicator latent model.
Conclusion:
Within-person elevations in self-reported exercise-adherence tendency were prospectively associated with lower subsequent PSU, whereas the reverse paths were not statistically significant. The directional difference was not supported in the final interval, and the observational design does not establish a causal or preventive effect.
