Effects of socio-structural variables in the theory of planned behavior: a mediation model in multiple samples and behaviors

Abstract

Objective: Observed variation in health behavior may be attributable to socio-structural variables that represent inequality. We tested the hypothesis that variability related to socio-structural variables may be linked to variation in social cognition determinants of health behavior. A proposed model in which effects of socio-structural variables (age, education level, gender, income) on health behavior participation was mediated by social cognition constructs was tested. Design: Model effects were tested in correlational datasets (k = 13) in different health behaviors, populations, and contexts. Samples included self-report measures of age, highest attained education level, gender, and net household income, and constructs from the theory of planned behavior (attitude, subjective norms, perceived behavioral control, intention). Ten samples provided follow-up self-reports of health behavior. Results: Path analyses supported sample-specific indirect effects of gender and age on health behavior with comparatively few income and education level effects. Meta-analytic structural equation modeling indicated consistent indirect and total effects of gender on intentions and health behavior through social cognition constructs, and a total effect of education level on behavior. Conclusion: Results provide support for the proposed mechanism by which socio-structural variables relate to health behavior. Replication in larger samples and meta-analytic synthesis across multiple health behavior studies is warranted.

Keywords:

• Health behavior determinants
• social cognition theory
• health-related behavior
• intention
• attitude

Disclosure statement

The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Study materials, data files, data analysis scripts, and analysis output presented in this manuscript can be found at https://osf.io/zhbq6.

Notes

1 Data files, analysis scripts, and data analysis output files can be accessed online: https://osf.io/zhbq6

2 The models estimated in datasets 3 (FH patients physical activity sample), 8 (parents swimming pool safety behaviors – national sample), and 9 (parents swimming pool safety behaviors – regional sample) did not include a behavior measure and were therefore fully saturated, so no fit statistics could be computed.

3 Full parameter estimates including p-values are presented in Appendix K (supplemental materials).

Additional information

Funding

Preparation of this manuscript was supported by a Finnish Distinguished Professor (FiDiPro) grant awarded to Martin S. Hagger from Business Finland (Grant #1801/31/2015).