Social support refers to the material and psychosocial assistance individuals obtain from their social networks when facing stressful events, primarily including emotional support and informational exchange (Cohen et al., 2004; Siedlecki et al., 2014). Within the framework of the Job Demands–Resources (JD-R) model, social support is conceptualized as an important contextual resource that can enhance learning motivation and engagement through the motivational process.

Previous research suggests a close association between perceived pressure and social support. According to conservation of resources theory, sustained high levels of pressure consume individuals’ limited psychological and temporal resources, thereby weakening their capacity to maintain and access social support (Hobfoll et al., 1990). In the context of doctoral training, prolonged pressure and its accompanying negative emotions may lead doctoral students to display irritability, withdrawal, or social avoidance, which can impair the quality of interactions with supervisors and peers and reduce their subjective perception of available social support (Sarason et al., 1990). When social support is compromised, doctoral students may find it more difficult to obtain emotional reassurance and academic guidance from their environment, which in turn may undermine their sustained engagement in learning and research activities.

In contrast, adequate and effective social support helps individuals maintain a positive psychological state and further Predicts learning engagement by strengthening self-efficacy (Bakker and Demerouti, 2007). In higher education settings, institutional support from universities and a positive academic climate have been shown to significantly enhance doctoral students’ learning engagement (Pyhältö et al., 2023). Empirical studies further indicate that, under high-pressure conditions, insufficient perceived social support is associated with a marked decline in engagement willingness, making individuals more prone to reduced involvement, psychological withdrawal, or even disengagement (Alarcon et al., 2011; Siu et al., 2021; Peltonen et al., 2017).

Social support represents not only a crucial external resource for doctoral students coping with academic pressure but also a potential mediating mechanism linking perceived pressure to learning engagement. Variations in pressure levels may influence doctoral students’ engagement by shaping the availability and perception of social support, thereby providing a theoretical basis for the mediating hypothesis of social support proposed in the present study.

Self-efficacy refers to an individual’s judgment of their capabilities to successfully execute specific tasks (Bandura, 1997), while academic self-efficacy represents students’ subjective beliefs regarding their capabilities to achieve academic goals (Song and Hu, 2024). As a key personal resource, academic self-efficacy serves as a core driving force in the individual’s motivational process. Perceived pressure, as a significant contextual factor, exerts a notable influence on academic self-efficacy (Ma, 2025). Regarding this relationship, there are two distinct perspectives in the literature. One view argues that chronic or high-intensity perceived pressure depletes students’ cognitive and emotional resources, thereby diminishing their academic self-efficacy (Gao, 2023). For instance, Lu et al. (2025) verified a negative relationship between pressure and academic self-efficacy among Chinese elite university students and students preparing for the college entrance examination, respectively. Conversely, another view posits that increased perceived pressure can enhance students’ academic self-efficacy (Pintrich and Schunk, 1996).

Students’ learning engagement is largely contingent upon their academic self-efficacy (Pajares and Schunk, 2001). High levels of perceived pressure are often accompanied by negative emotions such as anxiety and fatigue, which directly diminish academic self-efficacy (Ma, 2025). Furthermore, academic self-efficacy determines the intensity of engagement individuals exhibit when completing academic tasks (Wang and Gao, 2021). Students with low academic self-efficacy tend to adopt avoidance strategies, struggle to exert effort in their studies, and exhibit greater passivity (Komarraju and Dial, 2014), or even experience academic burnout (Lu et al., 2025). In contrast, students with high academic self-efficacy demonstrate greater concentration, persistence, and initiative, thereby displaying superior levels of learning engagement (Wang and Gao, 2021).

When examining the joint roles of doctoral students’ perceived pressure, social support, and academic self-efficacy in shaping learning engagement, Self-Determination Theory (SDT) provides a systematic framework for understanding how external support can be transformed into internal motivation and psychological resources. According to SDT, individuals’ motivation and engagement are not determined solely by external conditions or personal traits, but are jointly influenced by social support and internal psychological resources such as self-efficacy (Ryan and Deci, 2020). Within this framework, social support functions as a key external protective resource that satisfies individuals’ emotional and relational needs, facilitates the development of positive self-perceptions, and strengthens self-efficacy. In turn, higher levels of self-efficacy constitute an important psychological foundation for sustained learning engagement.

Empirical research further suggests that, in high-pressure contexts, social support not only exerts a direct motivational influence but may also operate indirectly by shaping academic self-efficacy. When doctoral students face elevated academic demands and pressure, support from supervisors, peers, and family can help buffer the erosion of self-beliefs caused by stress and enable students to maintain a positive evaluation of their academic capabilities. For example, a study of doctoral students in public administration found that social support from peers and family facilitated work–life balance, thereby enhancing academic self-efficacy and subsequently promoting research engagement (Schwoerer et al., 2021).

Based on the above theoretical and empirical evidence, it can be inferred that, within the high-demand context of doctoral training, perceived pressure may not influence learning engagement directly. Instead, perceived pressure may shape the availability and perception of social support, which subsequently enhances academic self-efficacy and, in turn, affects learning engagement. This process reflects a pathway through which external environmental resources are transformed into internal psychological resources, providing a clear rationale for the proposed chain mediation involving social support and academic self-efficacy.

Existing research indicates that bivariate relationships exist among perceived pressure, learning engagement, social support, and academic self-efficacy. However, few studies have investigated their systematic impact on doctoral students’ learning engagement, and the potential mediating roles of social support and academic self-efficacy remain insufficiently explored. Therefore, grounded in the Job Demands–Resources model, this study investigates the roles of social support and academic self-efficacy in the process of transforming doctoral students’ perceived pressure into sustained learning engagement (see Figure 1).

We propose the following directionally specified hypotheses (the hypothetical model is shown in Figure 1):

To obtain the data, the author searched core journals across disciplines in the Chinese Social Science Citation Index (CSSCI)—the authoritative journal evaluation system in China—and collected email addresses of doctoral students who had published papers in these journals over the past 5 years. Data collection spanned from April to August 2025. In total, 700 doctoral students were identified and invited to participate; 500 questionnaires were returned, of which 432 were valid, yielding a validity response rate of 86.4%. Regarding sample composition, the respondents originally spanned all 13 official disciplinary categories defined by the Chinese Ministry of Education. To ensure conciseness and theoretical clarity, these disciplines were grouped into four distinct categories based on Becher's (1989) typology. Furthermore, the sample demonstrates strong structural diversity, covering doctoral students from both research-oriented and applied universities across more than 20 provinces and municipalities in China. Detailed sample characteristics are presented in Table 1.

Studies indicate that doctoral students’ gender (Xin et al., 2024), grade level (Moreira et al., 2023), age (Nja et al., 2023), and family economic status (Zhu, 2023) significantly influence perceived pressure or learning engagement. To mitigate potential confounding effects, these demographic factors were included as control variables in the analysis.

First, data screening was performed to exclude questionnaires exhibiting response sets or logical inconsistencies. Next, descriptive statistics were computed for demographic characteristics and core variables. Pearson correlation analysis was then conducted to examine the bivariate associations among these variables. Finally, for mediation analyzes, the regression-based approach recommended by Hayes and Rockwood (2017) was adopted. Specifically, Models 4 and 6 of the PROCESS macro for SPSS (Version 24.0) were utilized, with gender, age, grade, and family income entered as covariates to control for their potential confounding effects. Model 4 was used to examine the independent mediating effects of social support and academic self-efficacy between perceived pressure and learning engagement, Model 6 was employed to test the chain mediation effect. The significance of indirect effects was assessed using the bias-corrected percentile bootstrap method with 5,000 resamples to generate 95% confidence intervals (CIs).

Since the data were derived from self-reported measures, the potential for common method bias (CMB) was addressed through both procedural and statistical means. Procedurally, measures such as respondent anonymity and randomized item ordering were implemented during data collection to minimize bias. Statistically, we first rigorously verified the construct validity of the measurement model using Confirmatory Factor Analysis (CFA) in AMOS 24.0. The results indicated that the four-factor measurement model (Perceived Pressure, Social Support, Academic Self-Efficacy, and Learning Engagement) fitted the data satisfactorily (χ²/df = 1.106, RMSEA = 0.016, RMR = 0.025, IFI = 0.997, TLI = 0.996, CFI = 0.997).

Furthermore, convergent validity was assessed (see Table 2). While the Average Variance Extracted (AVE) values for some constructs were slightly below the 0.50 threshold, the Composite Reliability (CR) values for all constructs ranged from 0.796 to 0.939, well above the recommended cutoff of 0.60. According to Fornell and Larcker (1981), if AVE is less than 0.5 but CR is higher than 0.6, the convergent validity of the construct is still considered adequate.

Following this validation, Harman’s single-factor test was conducted. The results extracted nine factors with eigenvalues exceeding 1, with the first factor accounting for only 28.343% of the total variance. This is well below the 40% threshold (Podsakoff et al., 2003), suggesting that common method bias is not a pervasive issue in this study.

Table 3 presents the descriptive statistics and Pearson correlation coefficients for all research variables. As the data indicate, perceived pressure exhibited significant positive correlations with social support (r = 0.379, p < 0.01), academic self-efficacy (r = 0.394, p < 0.01), and learning engagement (r = 0.414, p < 0.01). Furthermore, social support was positively associated with both academic self-efficacy (r = 0.36, p < 0.01) and learning engagement (r = 0.395, p < 0.01). Finally, a robust positive relationship was observed between academic self-efficacy and learning engagement (r = 0.424, p < 0.01).

Based on the descriptive statistics, the doctoral students in this sample generally exhibit a profile characterized by “moderate pressure–high resources–high engagement.” As shown in Table 2, the mean score for perceived pressure was 4.660, aligning with the theoretical midpoint of 4, which indicates a moderate level of pressure. The mean score for social support was 2.712 (slightly above the theoretical midpoint of 2.5), suggesting that participants generally perceive a certain level of supportive resources. The mean score for academic self-efficacy was 3.409 (notably higher than the theoretical midpoint of 3), demonstrating a strong sense of academic competence and self-belief. Furthermore, the mean score for learning engagement was 3.464 (significantly exceeding the theoretical midpoint of 2.5), indicating that doctoral students in this study generally maintain high levels of academic participation and engagement.

Correlation analysis further revealed the interplay among these variables. Perceived pressure showed significant positive correlations with social support, academic self-efficacy, and learning engagement (p < 0.01). Interpreted through the lens of the Job Demands–Resources (JD-R) model, this suggests that while doctoral students endure prominent pressure, they are simultaneously able to mobilize relatively abundant social support and internal psychological resources. Consequently, for many doctoral students, this pressure functions more as a “challenge stressor,” thereby stimulating learning engagement.

The study found that perceived pressure serves as a significant positive predictor of learning engagement (β = 0.414, p < 0.001). Specifically, for every one standard deviation increase in perceived pressure, learning engagement increases by 0.414 standard deviations. This result suggests that within the current sample, perceived pressure does not merely manifest as a ‘risk factor’ suppressing learning engagement. This finding aligns with existing research indicating that moderate stress contributes to enhanced cognitive function and learning performance. For instance, studies by Vogel and Schwabe (2016) and Zoladz et al. (2011) have demonstrated that moderate levels of stress can, under certain conditions, enhance memory processing and task engagement. Focusing on the doctoral cohort, this study reaffirms the positive role of perceived pressure within high-demand academic contexts.

This stands in contrast to previous conclusions, which highlighted that the prevalence of mental health issues like depression and anxiety among doctoral students exerts detrimental effects on academic performance and persistence (Levecque et al., 2017). Our findings suggest that the impact of perceived pressure is not simply ‘the higher, the worse.’ Its function depends largely on the stress level and the resources available to the individual (Chen et al., 2024). In contexts where resources are relatively abundant, moderate perceived pressure can foster doctoral students’ learning engagement by stimulating effort and a sense of responsibility.

It is important to acknowledge that the Graduate Stress Scale (GSS) used in this study primarily measures the intensity of stress exposure rather than the students’ cognitive appraisal of stress (i.e., whether they view it as a challenge or a hindrance). However, our findings show a positive association between perceived pressure and engagement, suggesting that in the context of high resource availability (as observed in our sample), these high-intensity demands are functionally operating as ‘challenge demands’ that stimulate effort, rather than ‘hindrance demands’ that deplete it. Future research should explicitly measure stress appraisal to confirm this mechanism.

Regarding the pathway mediated by social support, learning engagement is positively related to perceived pressure by elevating social support levels. This indirect effect accounts for 24.986% of the total effect, underscoring the significant mediating role of social support in the pressure–engagement relationship. This finding aligns with prior research demonstrating the mediating function of social support in the link between stress and work engagement (Hakanen et al., 2008). Notably, this study reveals a distinctive pattern: doctoral students experiencing higher pressure did not succumb to withdrawal or psychological exhaustion. Instead, in the face of intensified pressure, they were more inclined to mobilize social support networks to compensate for personal resource deficits, thereby sustaining their learning engagement. In other words, heightened perceived pressure may serve to activate the individual’s support system. Concurrently, a stronger social support network bolsters confidence in coping with academic challenges, thereby increasing the likelihood of successful degree completion and sustained learning engagement. This corroborates the conclusion of García-Machado et al. (2024) that, under conditions of resource sufficiency, moderate pressure can enhance behavioral efficiency and goal persistence via social support, ultimately maintaining high levels of learning engagement within stressful contexts.

In the pathway mediated by academic self-efficacy, perceived pressure is positively related to learning engagement by bolstering academic self-efficacy levels. The indirect effect was estimated at 0.075, accounting for 28.712% of the total effect. This finding resonates with the core tenets of Job Demands–Resources (JD-R) theory: in high-demand contexts, provided individuals can leverage sufficient resources, perceived pressure can be transformed into a driver of engagement through a motivational process. As a critical personal resource, academic self-efficacy acts as a pivot in this process. On the one hand, the pressure context stimulates coping motivation, prompting doctoral students to accumulate mastery experiences through overcoming difficulties, thereby enhancing academic confidence. On the other hand, this fortified self-efficacy further strengthens goal commitment and perceived task control, enabling the maintenance of high levels of focus and effort even within high-pressure environments.

The chain mediation analysis further elucidates how perceived pressure influences doctoral students’ learning engagement through the interplay of external social resources and internal psychological resources. Specifically, perceived pressure prompts doctoral students to actively mobilize external social resources to cope with challenges. The utilization of this social support is then converted into internal academic self-efficacy, thereby enhancing students’ confidence in their research and problem-solving capabilities. Ultimately, this heightened self-efficacy drives students to maintain deep learning engagement—even under high-pressure conditions—by strengthening goal commitment and persistence. This forms a complete transmutation pathway: ‘Perceived Pressure→Social Support→Academic Self-Efficacy→Learning Engagement.’

It is worth noting that the effect size of the chain mediation path was significantly smaller than that of the single mediation paths. This phenomenon is primarily attributed to the statistical ‘cumulative attenuation effect’: as the mediation chain lengthens, the influence of the predictor variable is inevitably diluted during the multi-stage transmission process. Furthermore, this reflects the incompleteness of resource conversion. Although perceived pressure can motivate individuals to acquire external social support, this support does not automatically or fully internalize into academic self-efficacy. This process is constrained by individual cognitive processing and other factors, thereby limiting the overall strength of the chain mediation effect.