The study, conducted by Zhijia Yang of Henan University, used partial least squares structural equation modeling to test whether work-life balance functions as a shared indirect pathway linking these heterogeneous resources with job satisfaction. The theoretical framing draws on the Job Demands-Resources model, which holds that job demands consume psychological and physical resources while job and personal resources support coping, motivation, goal attainment, and wellbeing. What makes this study distinctive is its attempt to extend that resource logic beyond conventional workplace supports, treating structured organizational training and flexible online learning as professional-development resources, healthy eating and physical activity as lifestyle-related personal resources, and social media engagement as a behavioral-social resource embedded in digital work life.

The survey sample consisted of full-time employees with at least one year of work experience, drawn from information technology, education, healthcare, finance, manufacturing, and service industries. Of 520 questionnaires distributed, 462 responses were received, and 437 valid responses remained after cleaning, an effective valid response rate of 84.0 percent. Respondents were 48 percent female and 52 percent male, with 46 percent aged 21 to 30, 34 percent aged 31 to 40, and 20 percent aged 41 or older. Work experience was distributed across 42 percent with one to five years, 31 percent with six to ten years, and 27 percent with more than ten years. The questionnaire measured seven constructs, each with three indicators on a five-point Likert scale, adapted from prior organizational behavior and wellbeing research and refined through expert assessment and pre-testing.

The measurement model demonstrated strong psychometric properties. Standardized indicator loadings ranged from 0.890 to 0.950, all above the conventional 0.70 threshold. Average variance extracted ranged from 0.807 to 0.890, composite reliability from 0.926 to 0.961, and Cronbach’s alpha from 0.881 to 0.939, indicating strong internal consistency and convergent validity. Discriminant validity was assessed using the heterotrait-monotrait ratio alongside the Fornell-Larcker criterion and cross-loading inspection, and all heterotrait-monotrait values fell below the conservative 0.85 cutoff, with the highest value of 0.706 observed between work-life balance and job satisfaction. Indicator-level variance inflation factors ranged from 2.311 to 4.625, indicating that multicollinearity was not present under the commonly used threshold of 5.

The structural results showed that all five hypothesized paths from the resource variables to work-life balance were statistically significant at the 0.001 level. Online learning participation led with a coefficient of 0.379, followed by social media engagement at 0.332, employee training at 0.312, physical activity at 0.294, and healthy dietary habits at 0.272. Effect sizes clarified the relative contributions: online learning showed the largest antecedent effect size at 0.351, followed by social media engagement at 0.270, training at 0.238, physical activity at 0.211, and diet at 0.181. The path from work-life balance to job satisfaction showed a large effect size of 0.637. Together, the five resources explained 59.1 percent of the variance in work-life balance, while work-life balance explained 38.9 percent of the variance in job satisfaction, and blindfolding-based Q-squared values of 0.374 and 0.257 indicated strong and medium predictive relevance respectively.

Bootstrapped indirect-effect analysis supported work-life balance as a statistically significant indirect pathway for all five resources. The indirect effect through work-life balance was 0.347 for social media engagement, 0.304 for physical activity, 0.266 for employee training, 0.259 for online learning, and 0.257 for healthy dietary habits, with all 95 percent bootstrapped confidence intervals excluding zero. The author interprets these as model-based indirect associations rather than evidence of causal mediation, because the cross-sectional design cannot establish temporal ordering. Because the structural model did not simultaneously estimate direct paths from the resources to job satisfaction, the analysis also cannot distinguish full mediation from partial mediation, and the results are described as supporting an indirect-only theoretical specification.

The strongest single finding concerns online learning participation, which the author attributes to its dual character as both a competence-building and an autonomy-supporting resource. Unlike organization-scheduled training, online learning is typically self-directed, asynchronous, and adjustable to employees’ available time, potentially giving workers greater discretion over when and how they develop skills. This flexibility may reduce time-boundary pressure and support self-regulation, allowing professional development to align with personal responsibilities. Employee training, by contrast, is generally initiated and scheduled by the organization, so its benefit for work-life balance may depend on whether training is integrated into workload planning rather than added as an extra demand on top of existing tasks. The author also cautions that the relatively high full collinearity values for the two professional-development constructs mean their distinct contributions should be interpreted carefully.

The positive association between social media engagement and work-life balance is notable given prior research cited in the study linking greater digital-technology use to work-life conflict. The author emphasizes that the construct was operationalized around professional connection—staying in touch with networks, exchanging information, and receiving support—rather than raw usage intensity. Boundary theory and recovery theory both suggest the relationship may be conditional: social media can weaken work-non-work boundaries when it creates constant availability expectations and after-hours interruptions, yet it may support balance when it facilitates networking, reduces uncertainty, and helps employees resolve work-related concerns more efficiently. The study did not test nonlinear effects, and the author flags this as a priority for future research, along with moderators such as after-hours availability expectations, digital overload, segmentation preference, and perceived boundary control.

Model fit was adequate but not unqualified. The standardized root mean square residual was 0.053, below the commonly used 0.08 threshold, while the normed fit index was 0.893, slightly below the ideal 0.90 cutoff, leading the author to interpret fit cautiously rather than describe it as unequivocally acceptable. The study also assessed potential common method bias, given that all data came from a single-wave self-report questionnaire. Full collinearity variance inflation factors ranged from 2.841 to 4.794, all below 5 but several exceeding the more conservative 3.3 cutoff, particularly online learning at 4.794 and training at 4.733. This leaves open the possibility of method-related covariance or partial conceptual overlap between the two professional-development constructs, and the author acknowledges common method bias as a limitation of the design.

The practical implications are framed cautiously given the cross-sectional design. The author suggests that organizations should treat professional development as a resource rather than an additional demand: scheduling mandatory training within paid working hours, offering short modular online courses with realistic completion windows, capping concurrent courses, and communicating clearly that flexible learning is not intended to extend the workday. Managers are encouraged to discuss individual learning goals and workload conditions rather than assigning uniform schedules, and to monitor work-life-balance indicators such as perceived training workload and time spent on learning outside working hours. The author stresses that because the data were collected at a single point in time, the results should be read as statistical associations within the specified model, not as proof that any resource causally improves work-life balance or job satisfaction; longitudinal, time-lagged, or experimental designs would be needed to establish causal effects, and future work should also test the framework across countries, cultures, and institutional settings.

Subject of Research: Associations among professional-development resources, lifestyle behaviors, work-life balance, and job satisfaction in a multi-sector employee sample

Article Title: Professional development, lifestyle behaviors, and job satisfaction: the mediating role of work-life balance in a multi-sector sample

Article References: Yang, Z. (2026). Professional development, lifestyle behaviors, and job satisfaction: the mediating role of work-life balance in a multi-sector sample. Frontiers in Psychology, 17, Article 1923493. https://doi.org/10.3389/fpsyg.2026.1923493

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1923493

Keywords: work-life balance, job satisfaction, online learning, employee training, structural equation modeling, physical activity, healthy diet, social media engagement, organizational psychology, cross-sectional survey, Job Demands-Resources model, employee wellbeing