The present study employed a convenience sampling approach to recruit 251 elementary and secondary school teachers from Shandong province and Xinjiang Uygur Autonomous Region, China. Approximately 280 teachers were initially invited (via online invitations and school contacts), and 251 of them provided valid responses, yielding a response rate of about 90%. This sampling method was chosen for its practicality and feasibility in accessing the target population (i.e., school teachers in the targeted regions). As the study focused on married participants, 62 unmarried teachers were excluded, yielding a final analytic sample of 189 married teachers. Only married teachers were asked to complete the marital quality scale; accordingly, data from unmarried respondents were excluded from all analyses. Table 1 presents the demographic characteristics of the sample. A total of 189 teachers participated in the study, comprising 107 males (56.6%) and 82 females (43.4%). Most participants were between 31 and 50 years of age (71.4%), with 33.3% aged 31–40 years and 38.1% aged 41–50 years. In terms of educational attainment, the majority held a bachelor’s degree (83.6%), followed by vocational college (10.6%), a master’s degree (4.2%), a high school diploma or lower (1.1%), and doctoral degrees (0.5%). Regarding teaching level, 64.6% taught at the primary level, while 35.4% were secondary school teachers. (All participants were full-time teachers at the time of the survey.)

Before the commencement of the survey, the research protocol underwent a comprehensive ethical review by the first author’s academic institution. The review process considered aspects such as informed consent, confidentiality, and the potential risks and benefits of the study. With the assistance of local education bureaus, the Wenjuan Xing online questionnaire was distributed to teachers in primary and secondary schools across Shandong and Xinjiang provinces, who completed and submitted it anonymously. This platform offered several advantages, including ease of administration, efficient data collection, and the ability to reach a geographically dispersed sample. All participating teachers provided their voluntary informed consent to take part in the study. The questionnaire collected demographic information (e.g., teacher gender, age, marital status) and included a series of scales measuring teachers’ job stress, well-being, marital quality, and social support. The survey link remained open for approximately 1 week. All respondents completed the questionnaire with sections presented in a fixed order (demographic questions first, followed by the job stress, well-being, social support, and marital quality scales). Participants were assured that their responses would remain confidential and anonymous. Furthermore, all data may be shared in an anonymized format for research purposes only, ensuring both transparency and protection of participant privacy.

The data analysis proceeded in three stages. In the first stage, descriptive statistics and Pearson correlations were computed for the primary study variables, including teacher job stress, subjective well-being, social support, and marital quality. The distributional assumptions were also assessed by examining skewness and kurtosis values. Moreover, to examine the presence of common method bias, Harman’s single-factor test was conducted. In the second stage, multiple regression analysis was conducted in SPSS 26 to explore the direct effect of teacher job stress on subjective well-being. In the third stage, structural equation modeling (SEM) was employed to investigate whether marital quality and social support sequentially mediated the relationship between teacher job stress and subjective well-being (see Figure 1). Consistent with Hayes (2018), a bootstrap procedure was used to test the significance of indirect effects, as this method yields more accurate estimates than conventional significance tests. Specifically, 5,000 bootstrap samples were generated to calculate point estimates and 95% bias-corrected confidence intervals. Indirect effects were considered significant if the confidence interval did not include zero. As all models were saturated (i.e., just-identified), model fit indices were not reported (Kline, 2016). All descriptive and correlational analyses were performed in SPSS 26, whereas SEM was conducted using Mplus 8.3.

To reduce the likelihood of common method bias, this study utilized response tokens and ensured anonymity in participants’ answers (Zhou and Long, 2004). In addition, Harman’s single-factor test was employed as a diagnostic procedure (Podsakoff et al., 2003). The exploratory factor analysis revealed that the first factor explained 20.06% of the total variance, which is substantially lower than the recommended threshold of 40%. This result indicates that common method variance was unlikely to pose a significant threat in the present study. Furthermore, according to Hair et al. (2010), data can be considered approximately normal when skewness values range between −2 and +2 and kurtosis values between −7 and +7. Consistent with this guideline, all study variables exhibited an acceptable level of normality, with skewness and kurtosis both remaining within |2|.

Table 2 presents the descriptive statistics and zero-order correlations among all study variables. Additionally, to examine the relationships among the key variables, we generated scatter plots for all pairwise combinations of the four variables in this study (Supplementary Figures S1–S6). Job stress was negatively correlated with marital quality (r = −0.333, p < 0.001) and subjective well-being (r = −0.685, p < 0.001), indicating that higher levels of job stress were associated with poorer marital relationships and lower subjective well-being. Marital quality was positively related to social support (r = 0.228, p < 0.01) and subjective well-being (r = 0.531, p < 0.001), suggesting that teachers with higher marital quality tended to report greater social support and well-being. Social support also showed a significant positive association with subjective well-being (r = 0.201, p < 0.01).

As presented in Table 3, the results of hierarchical regression analysis found that teachers’ job stress significantly and negatively predicted subjective well-being (β = −0.685, p < 0.001) in Step 1. This result was consistent with our hypothesis and indicated that higher levels of job stress were associated with lower levels of subjective well-being. In Step 2, marital quality (β = 0.309, p < 0.001) and social support (β = 0.125, p < 0.05) were added to the model, both of which significantly and positively predicted teachers’ subjective well-being. These findings suggest that teachers with higher marital quality and stronger social support tend to report greater well-being, even after controlling for job stress.

Moreover, adding marital quality and social support in Step 2 explained an additional 11.8% of the variance in teachers’ subjective well-being, resulting in a total of 58.7% of the variance being explained by the full model. In addition, all variance inflation factor (VIF) values ranged from 1.000 to 1.193, well below the recommended threshold of 5 (Kim, 2019), indicating no multicollinearity issues among the predictors.

As illustrated in Figure 2, we further explored a sequential mediation model in which marital quality predicted social support. The results showed that job stress was significantly associated with marital quality (β = −0.33, p < 0.001), which in turn positively predicted social support (β = 0.25, p < 0.01), and social support significantly predicted teachers’ subjective well-being (β = 0.13, p < 0.05). Also, the marital quality was significantly and positively linked to teachers’ subjective well-being (β = 0.31, p < 0.001). Moreover, job stress was significantly and negatively associated with teachers’ subjective well-being (β = −0.58, p < 0.001). Notably, teachers’ job stress was not associated with social support, which indicated that social support did not mediate the relationship between teachers’ job stress and subjective well-being.

As presented in Model 1 of Table 4 and in line with Hypothesis 2, the bootstrap-corrected confidence intervals for the indirect role of marital quality in the relationship between teachers’ job stress and subjective well-being were significant (indirect effect = −0.103, SE = 0.030, 95% CI [−0.169, −0.053]). The confidence interval excluded zero, indicating that marital quality plays a partial mediating role in the association between the teachers’ job stress and subjective well-being. Similarly, as shown in Model 3 of Table 4 and in line with Hypothesis 4, marital quality and social support in the relationship between teachers’ job stress and subjective well-being were significant (indirect effect = −0.011, SE = 0.007, 95% CI [−0.026, −0.001]). The confidence interval excluded zero, indicating that marital quality and social support play a partial sequential mediating role in the association between teachers’ job stress and subjective well-being.

This study found a significant negative association between teachers’ job stress and their subjective well-being. In other words, teachers experiencing higher levels of job stress reported lower levels of subjective well-being, thus supporting Hypothesis 1. This result is consistent with the findings of Shao et al. (2025) and Mérida-López et al. (2022), indicating that within the Chinese cultural and educational context, high job stress similarly reduces teachers’ life satisfaction and overall subjective well-being.

These findings provide empirical support for the ecological systems theory (Bronfenbrenner, 1979, 2005), which posits that proximal processes within the microsystem exert a direct influence on individual development. The school microsystem constitutes a critical setting where teachers’ stress is generated and accumulated. When teaching demands, administrative requirements, and emotional burdens persist within this system, they transform into chronic stressors that undermine teachers’ subjective well-being. This mechanism suggests that teachers’ subjective well-being is not determined solely by personal traits but is also profoundly shaped by their daily work environment. Therefore, this study demonstrates the direct role of the school microsystem related to teachers’ subjective well-being and highlights the importance of contextual factors in understanding variations in subjective well-being.

This study found that marital quality served as a significant mediator in the relationship between teachers’ job stress and subjective well-being. Specifically, higher job stress was associated with lower marital quality, which in turn led to reduced subjective well-being, thereby supporting Hypothesis 2. This finding provides empirical evidence for understanding how occupational stress is linked to subjective well-being. In contrast, social support did not mediate the relationship between job stress and subjective well-being, and Hypothesis 3 was not supported. This suggests that social support may not be the primary mechanism, whereas marital quality plays a more central role in linking teachers’ stress and well-being.

These findings provide empirical confirmation of ecological systems theory (Bronfenbrenner, 1979, 2005), particularly its emphasis on the interconnections between systems. Teachers’ job stress is not confined to the school microsystem; it also spills over into the family microsystem through cross-system “spillover effects” (Grzywacz and Marks, 2000), thereby undermining marital quality and reducing well-being. This process reflects the role of the mesosystem, highlighting how interactions between school and family systems are jointly related to teachers’ well-being. The findings underscore the importance of family systems, compensate for the lack of attention to family dynamics in prior research, and extend the applicability of ecological systems theory within the Chinese educational context.

This study further revealed that teachers’ job stress is indirectly linked to well-being through the sequential mediating roles of marital quality and social support, thus supporting Hypothesis 4. Specifically, higher job stress reduced marital quality, which subsequently led to diminished social support, ultimately resulting in lower levels of well-being.

This finding validates and extends ecological systems theory (Bronfenbrenner, 1979, 2005) by illustrating the interdependence of multiple systems. Teachers’ stress within the school microsystem first undermines marital relationships in the family microsystem, which in turn weakens broader social support networks, thereby reducing subjective well-being. This sequential mediation pathway reveals a cross-system cascading effect, showing that a single system does not determine teachers’ well-being but is related to the dynamic interconnections among school, family, and social systems. This mechanism resonates with ecological systems theory’s emphasis on nested systems and interdependence (Neal and Neal, 2013) and enriches our understanding of how stress transmits across systems.

Furthermore, the findings indicate that stress experienced in one system not only depletes resources within that system but also triggers cross-system transmission effects that erode other relational resources, thereby producing cumulative negative impacts on well-being. Thus, the identified sequential mediation pathway extends the theoretical application of ecological systems theory in educational contexts and deepens our understanding of the mechanisms shaping teachers’ subjective well-being.

Although this study revealed important relationships among teachers’ job stress, marital quality, social support, and subjective well-being, several limitations should be noted. First, the use of a cross-sectional design limits the ability to draw causal inferences among variables. Future studies should employ longitudinal or experimental designs to further examine the dynamic processes linking stress, relational resources, and well-being. Second, the sample was drawn from teachers in Shandong and Xinjiang Provinces, which may limit the representativeness of the findings for teachers nationwide. Moreover, a non-probabilistic (convenience) sampling method was used, which may further restrict generalizability. Future studies should consider using larger and more diverse samples across multiple regions to enhance the generalizability of the results. Third, the study relied on self-report questionnaires. Although efforts were made to ensure the reliability and validity of the measures, the data may still be subject to social desirability effects. Fourth, the study primarily focused on the mediating roles of marital quality and social support. At the same time, other potential factors were not considered, such as teachers’ personality traits, organizational climate, or cultural variables. Future studies could also include relevant covariates (e.g., teachers’ years of teaching experience and school type) to better account for contextual influences on teachers’ subjective well-being. Fifth, although SEM was used to examine the hypothesized mediation model, all constructs were treated as observed variables rather than latent factors. Given the modest sample size (N = 189) and the large number of model parameters (over 20 dimensions across all variables), this approach ensured model stability but limited the control of measurement error. Prior research suggests that the ratio of sample size to estimated parameters should not fall below 10:1 for stable latent-variable estimation (Jackson, 2003). Future research with larger samples should employ full latent-variable SEM to yield more accurate and robust results.

Despite these limitations, the findings yield several important implications for educational practice in China. First, enhancing teachers’ well-being requires coordinated efforts across multiple levels. Schools and educational authorities should mitigate the negative effects of job stress on teachers’ well-being by redistributing teaching and administrative tasks more equitably, streamlining formalities, and offering professional development opportunities along with psychological support (Hartney, 2016; Keeley, 2023). Given the cultural context of China, where family and marital quality play a pivotal role in teachers’ well-being, policies should also promote work–family balance. Initiatives such as flexible scheduling, paid leave, and access to family counseling services (Cvenkel, 2020; Akhter and Mahmud, 2025) could help mitigate the “spillover effects” of occupational stress. At the same time, schools should go beyond superficial collegial support to build deeper support systems, such as mentoring programs, peer support groups, and stronger home–school collaboration, to enhance teachers’ sense of social support (Chelelwa et al., 2025; Putra et al., 2024).

More importantly, improving teachers’ well-being requires multi-level and integrated interventions. At the individual level, mindfulness training and stress management programs can strengthen teachers’ self-regulation abilities (Rana et al., 2022; Sehgal and Kaur, 2024). At the family level, fostering communication and relational support is crucial for maintaining family functioning. At the societal level, establishing community-based and union-based support networks can help prevent the erosion of relational resources. The coordinated implementation of these multidimensional measures may alleviate teachers’ job stress, enhance their subjective well-being, and ultimately contribute to greater professional engagement and improved educational quality.