Exploring pathways to academic engagement: a SEM analysis of self-compassion, social support, and stress management among Chinese university students

Abstract

This study investigated the relationships between self-compassion, social support, stress management, and academic engagement among Chinese university students using a cross-sectional design and structural equation modeling (SEM). A total of 389 students from various undergraduate programs at two universities in major Chinese cities participated in the study. The results revealed that self-compassion and social support had significant positive direct effects on academic engagement. Furthermore, self-compassion and social support also exerted significant indirect effects on academic engagement through stress management, highlighting the mediating role of stress management in this relationship. These findings suggest that interventions aimed at enhancing self-compassion and social support may be beneficial for promoting academic engagement among university students by fostering better stress management skills.

Introduction

Academic engagement, encompassing cognitive, emotional, and behavioral components, is pivotal for student success and is strongly linked to academic achievement, motivation, and overall well-being [1–3]. Despite its importance, many students struggle to maintain engagement due to the stress and setbacks inherent in the demanding academic environment [4]. Prior research has explored various influences on academic engagement, including emotions, sense of belonging, and self-directed learning [5–7]. However, there remains a critical gap in understanding how specific personal resources interact to buffer stress and enhance engagement within established theoretical frameworks. This study addresses this gap by focusing on three under-investigated yet potentially potent constructs: self-compassion, social support, and stress management.

Grounded in Conservation of Resources (COR) Theory [8] and Self-Determination Theory (SDT) [9], self-compassion, which involves treating oneself with kindness during challenges [10], has been identified as a crucial factor for student well-being and academic achievement [11, 12]. Research indicates that self-compassion fosters emotional resilience and helps buffer against academic stress [13, 14], enabling students to recover from setbacks more effectively. Social support, which includes instrumental, emotional, and companionship aspects [15–17], has also been linked to positive academic outcomes. It enhances engagement, psychological resources, and academic self-efficacy [18–21], providing students with a network of support that mitigates stress and promotes sustained academic effort. Effective stress management strategies, including problem-focused coping, emotion-focused coping, and mindfulness [22, 23], are essential for reducing the detrimental effects of stress on academic engagement [24, 25]. These strategies help students maintain focus and motivation despite the challenges they face.

Drawing upon Conservation of Resources (COR) Theory [8, 26] and Self-Determination Theory (SDT) [9, 27], this study investigates how self-compassion, social support, and stress management collectively influence academic engagement. COR Theory highlights acquiring and conserving resources to manage stress, while SDT emphasizes psychological needs for intrinsic motivation. We aim to bridge a critical gap by examining the combined effects of these factors on how students navigate academic challenges and maintain engagement. Our hypotheses propose that self-compassion buffers stress’s negative effects on engagement (H1), social support enhances the positive effects of self-compassion and stress management on engagement (H2), and stress management mediates the relationship between self-compassion, social support, and academic engagement (H3).

By situating these constructs within COR Theory and SDT, this research offers a theoretically grounded framework illuminating how personal and social resources contribute to academic engagement. Our study uniquely contributes to the literature by integrating these theories to explore the synergistic effects of self-compassion, social support, and stress management on academic engagement, particularly among Chinese university students. The findings hold potential to inform interventions that foster self-compassion, strengthen social support, and develop effective stress management skills. Such interventions could significantly enhance student academic engagement and overall well-being, offering practical strategies for educators and policymakers to support success in higher education.

Literature review

Theoretical framework

This study is grounded in Conservation of Resources (COR) Theory and Self-Determination Theory (SDT), providing a structured lens to understand how stress management, self-compassion, and social support influence academic engagement among university students. COR Theory posits that individuals strive to acquire, protect, and maintain valuable resources—including personal characteristics, conditions, energies, and social connections—to withstand stress and minimize resource loss in demanding environments like academia [8, 22, 26]. Within this framework, resources act as buffers against stress, preserving the energy and motivation essential for sustained academic engagement [22, 28].

Within this study’s context, stress management and self-compassion act as critical personal resources. Effective stress management strategies, such as problem-solving and time management, directly reduce stressors’ impact, preserving emotional energy and resilience, which in turn sustains academic engagement [28, 29]. Self-compassion, defined as self-kindness during challenges [10], serves as an internal coping resource, mitigating self-criticism and fostering emotional resilience [30, 31]. This internal reserve allows students to persist through academic difficulties without psychological depletion, supporting ongoing engagement [32, 33].

Social support, a crucial social resource within COR Theory, provides external assistance, empathy, and encouragement from peers, family, or educators. This support offers both emotional and instrumental resources that buffer stress, promoting resilience and preventing disengagement [15, 16, 34, 35]. Consequently, access to these personal (stress management, self-compassion) and social (social support) resources enhances students’ capacity to cope with academic stress, fostering sustained motivation and active participation.

Alongside COR Theory, Self-Determination Theory (SDT) offers a vital perspective on the motivational drives underpinning well-being and academic engagement. SDT posits three fundamental psychological needs—autonomy, competence, and relatedness—as essential for intrinsic motivation and engagement [6, 9, 27, 36]. Autonomy involves perceived control over actions, competence relates to feeling effective, and relatedness signifies connection with others [6]. Fulfillment of these needs increases intrinsic motivation and engagement, as students find personal value in their academic pursuits [6, 27, 36]. Self-compassion aligns with autonomy and competence by fostering a supportive inner attitude, enhancing self-efficacy and intrinsic motivation [10, 37, 38]. This approach encourages viewing challenges as growth opportunities, promoting internally driven motivation [31]. Similarly, social support fulfills the need for relatedness, reinforcing students’ connection to their academic community and fostering a sense of belonging that supports engagement [16, 37].

Together, COR Theory and SDT provide a robust framework for understanding how self-compassion, social support, and stress management facilitate academic engagement. By considering these personal and social resources through both theoretical lenses, this study aims to illuminate the pathways enabling students to maintain motivation and active engagement despite academic pressures. These theories collectively highlight self-compassion, social support, and stress management as critical resources that foster resilience and persistence, ultimately contributing to enhanced academic engagement in university settings.

Academic engagement

Academic engagement, encompassing cognitive, emotional, and behavioral dimensions, is central to student success and strongly predicts academic achievement, motivation, and overall well-being [1, 39]. High engagement, evident in behaviors like regular attendance, timely assignment completion, and active participation, consistently correlates positively with academic performance and learning satisfaction [2, 39]. Cognitive engagement, involving deep processing and critical thinking, fosters meaningful learning and retention [36, 40]. Similarly, emotional engagement, reflecting students’ interest and enjoyment, promotes persistence and self-efficacy [3, 41].

Diverse factors influence academic engagement, ranging from emotional experiences to social relationships and personal coping mechanisms. Positive emotions, such as curiosity and enjoyment, are known to enhance motivation and subsequently, cognitive and behavioral engagement in learning tasks [5, 42]. Students’ sense of relatedness—their connection to the academic environment and peers—is another vital factor; those feeling supported and connected demonstrate higher engagement, underscoring the importance of cohesive and inclusive classroom settings [6, 41]. Furthermore, self-directed learning and intrinsic motivation are widely recognized drivers of academic engagement. Research indicates that students perceiving autonomy and control over their learning tend to exhibit greater engagement, driven by personal learning goals and ownership of their academic journey [6, 7]. The broader school context also significantly shapes engagement, with institutions that prioritize and nurture student involvement positively influencing motivation and active participation [6].

Personal attributes, including adaptive coping strategies, are also instrumental in shaping academic engagement. Problem-solving and seeking support, for instance, positively impact engagement by enabling more effective management of academic demands [43]. Engagement dimensions, such as vigor and dedication, mediate the link between coping strategies and academic performance, suggesting engagement can enhance the effectiveness of adaptive coping. Moreover, intrinsic motivation serves as a foundational element, significantly influencing sustained engagement and achievement [44]. While engagement alone does not directly predict academic achievement, it operates alongside motivation to foster academic involvement, highlighting the need for motivational interventions [44].

Social factors, particularly belonging and social support, play a pivotal role in sustaining academic engagement. Students feeling a strong connection to their classroom community exhibit both emotional and behavioral engagement, mitigating isolation [45]. Research also emphasizes the importance of social resources—like peer and family support—in cultivating relatedness and motivating sustained study engagement [46]. For example, Soria and Stebleton [46] found lower social capital among working-class students often correlated with reduced engagement, suggesting institutions should build support structures for equitable engagement among underrepresented groups.

The classroom environment and psychological attributes also warrant consideration. A positive social climate and a growth-oriented mindset, for instance, have been shown to reduce boredom and encourage sustained engagement, particularly in language learning settings [47, 48]. Environments supporting both academic achievement and students’ social-emotional needs are essential for long-term engagement. Additionally, psychological constructs like harmonious passion contribute to academic engagement, as students demonstrate increased dedication and reduced cynicism toward their studies [49]. Despite extensive research, a critical gap remains in understanding how specific personal resources interact to influence engagement among college students. Few studies have examined the combined effects of self-compassion, social support, and stress management on academic engagement, particularly within the Chinese higher education context. This study directly addresses this gap by exploring these interrelationships.

Self-compassion

Self-compassion, rooted in Buddhist philosophy and increasingly recognized in contemporary psychology [10], is defined as treating oneself with kindness and understanding during times of difficulty [10]. It comprises three core components: self-kindness, common humanity, and mindfulness [10]. Self-kindness involves offering oneself warmth and support, fostering acceptance and emotional resilience in the face of challenges [30]. Common humanity acknowledges that suffering and imperfection are universal experiences, reducing feelings of isolation and promoting a sense of connection [10, 30]. Mindfulness entails maintaining present-moment awareness of thoughts and emotions without judgment, allowing individuals to respond to difficulties with clarity and better emotional regulation [10, 30].

Research demonstrates that self-compassion offers multifaceted benefits across mental health, relationships, and resilience domains [50]. Individuals with high levels of self-compassion tend to experience lower anxiety, depression, and psychological distress, alongside greater life satisfaction and overall well-being [38]. Furthermore, self-compassion fosters adaptive coping mechanisms during adversity; self-compassionate individuals are more likely to employ problem-solving strategies, seek social support, and maintain a positive outlook under stress, leading to better psychological adjustment [31].

In the academic context, self-compassion directly enhances academic engagement by fostering resilience and motivation [11, 12, 14, 51, 52]. It serves as a buffer against academic stress and burnout, enabling students to maintain motivation and focus despite pressures [13, 14]. For example, students who practice self-compassion experience less emotional strain when faced with academic challenges, allowing them to engage more fully in their studies [13, 14]. Moreover, self-compassion is positively linked to academic engagement and active participation in learning activities, fostering a sense of accomplishment and satisfaction [51]. Self-compassion also interacts with social support and stress management to bolster academic engagement. It strengthens the positive impact of general resourcefulness and social support networks, amplifying the benefits of these resources on student engagement [52]. Additionally, self-compassion mediates the relationship between emotional intelligence and academic buoyancy, suggesting that it can enhance the benefits of emotional intelligence for student well-being and sustained engagement [53].

Studies consistently associate self-compassion with greater resilience, intrinsic motivation, and mental well-being across various disciplines [12, 54]. Nursing students with higher self-compassion, for example, reported increased resilience and engagement, indicating its protective role in demanding educational settings [12]. Similarly, social work students showed links between self-compassion, resilience, intrinsic motivation, and academic engagement, highlighting its broad utility [14]. Notably, self-compassion has been identified as a strong predictor of mental health, often surpassing other positive psychological constructs like engagement and motivation [54]. This underscores its critical role in well-being and points to a need for interventions supporting student mental health and engagement, particularly in populations less inclined to seek help [54].

Self-compassion also appears to reduce students’ fear of evaluation, encouraging open communication with peers and instructors [55]. Students with higher self-compassion are more willing to engage in discussions, ask questions, and seek help, as it provides emotional resilience against fear and anxiety, promoting active participation [55]. Incorporating self-compassion training into academic programs can cultivate supportive environments where students are better equipped to manage academic stress, engage fully, and maintain mental health [11, 56]. Integrating self-compassion and mindfulness into life skills training can thus build resilient graduates capable of sustaining academic engagement [56].

Social support

Social support, defined as the assistance, care, and empathy provided by one’s social network [15], plays a crucial role in individual well-being and adaptation to stressors. It encompasses instrumental support (tangible aid) and emotional support (empathy and encouragement) [16, 34, 35], alongside social companionship, which mitigates isolation and enhances quality of life [17, 57].

Research indicates that individuals perceiving higher social support generally experience lower stress and improved psychological functioning [15, 16]. They show greater emotional resilience, reduced depression and anxiety, and better overall health compared to those with weaker networks [16]. In the academic context, social support profoundly influences students’ engagement and commitment [18–21, 37, 58, 59]. Students with strong support from peers, family, or educators are more likely to actively participate, attend classes regularly, and invest effort [18–20, 37, 58]. This heightened engagement leads to deeper understanding, improved performance, and greater learning satisfaction. Social support also cultivates psychological capital (confidence, optimism) and promotes problem-focused coping, empowering students to effectively navigate academic challenges and sustain engagement [19, 21].

Moreover, social support bolsters academic self-efficacy—the belief in one’s ability to succeed academically [20]. Increased self-efficacy, in turn, translates into higher motivation and a willingness to tackle challenging tasks, contributing to positive learning outcomes and sustained engagement. For students facing adverse experiences or those who are first-generation college students, social support uniquely impacts academic engagement and persistence [21, 60]. A sense of belonging, mediated by family support, for example, enhances academic persistence, highlighting how inclusive environments foster retention [60].

Different sources of social support (parents, teachers, peers) influence motivation and achievement distinctly. Parental emotional support is particularly influential, fostering mastery goals and reducing performance-avoidance behaviors that can hinder engagement [61]. Teacher and peer support also promote adaptive motivation and minimize academic stress, with their impact often mediated by students’ achievement goals. This underscores the importance of a balanced support network encompassing both emotional and academic dimensions. Furthermore, social support positively predicts academic engagement through mediators like life satisfaction and academic motivation [62], indirectly boosting engagement by enhancing well-being. Peer support, notably, closely links with life satisfaction and academic engagement in a reciprocal relationship [63].

Self-efficacy serves as another mediator between social support and engagement. Support from friends and family cultivates self-efficacy, which then increases academic engagement among adolescents [64]. This highlights the value of programs that simultaneously enhance social support and self-belief to improve academic outcomes. For students at risk of early educational departure, such as those in vocational tracks, social support from teachers, peers, and parents is vital for school engagement [65]. Targeted educator support can foster pro-school attitudes and mitigate early school leaving, underscoring the role of robust support systems in retention policies.

Stress management

Stress management is essential for student well-being and academic engagement. It involves adopting strategies to cope effectively with stressors, thereby reducing their negative impact [22]. Effective stress management enhances resilience, improves mental health, and fosters a more engaging learning experience. Common approaches include problem-focused coping, which actively addresses the source of stress through techniques like problem-solving and time management [28, 29]. Students employing these strategies often experience lower distress and better adaptation to academic challenges. Emotion-focused coping, conversely, focuses on regulating emotional responses to stressors [22]. Techniques such as seeking social support, practicing relaxation, or engaging in enjoyable activities can help manage emotions and reduce stress’s impact on well-being [28].

Stress management is closely linked to self-compassion and social support, both of which can enhance its effectiveness in promoting academic engagement. Mindfulness-based stress reduction (MBSR), which incorporates self-compassion practices, has gained prominence for student stress management. MBSR programs, including mindful breathing, meditation, and yoga, reduce stress, anxiety, and depressive symptoms while enhancing overall well-being and fostering academic engagement [23, 66]. For example, medical students participating in MBSR programs reported higher satisfaction and increased engagement [67], highlighting the potential of integrating self-compassion into academic stress management.

Social support plays a significant role in stress management by providing emotional and instrumental resources that help students cope with academic pressures. Strong social support networks can buffer the effects of stress, leading to better psychological functioning and higher engagement [15, 16]. Students receiving support from peers, family, or educators are better equipped to handle stress effectively, which in turn fosters greater academic engagement [18, 19, 37]. Additionally, lifestyle factors like regular physical activity, healthy eating, adequate sleep, and strong social support networks are crucial for building resilience against daily academic pressures [68].

The relationship between stress management and academic engagement is well-documented. High levels of unmanaged stress often lead to decreased motivation, concentration difficulties, and academic disconnection [4, 69]. Students who develop effective coping strategies enhance their resilience, leading to sustained engagement and academic success [70]. Moreover, integrating self-compassion into stress management can further enhance resilience and engagement by promoting a kinder self-approach during challenges [13, 14]. Emotional regulation is also vital; positive emotions foster engagement, while negative ones, like anxiety, can undermine it [71]. Thus, interventions supporting emotional regulation and stress management are essential for sustained academic engagement.

Self-regulation skills, including time management and study planning, enable students to handle academic demands effectively [72]. Social support and self-compassion can enhance these skills by offering encouragement and fostering a positive mindset, thereby contributing to sustained academic engagement. Effort regulation is another key mediator between stress symptoms and academic outcomes; students struggling with it due to high stress symptomatology often show negative impacts on engagement [73]. This underscores the need for adaptive coping mechanisms. Additionally, emotional intelligence and a sense of coherence link to better stress management and engagement [74]. Students with high emotional intelligence and active coping strategies report better health and greater academic engagement [74], suggesting that incorporating self-compassion and leveraging social support can further promote academic engagement.

The present study

University life presents significant demands, intricately linking factors like self-compassion, social support, stress management, and academic engagement. This study aims to examine their interactions, focusing on their potential interplay in shaping student engagement. Drawing upon COR Theory and SDT, we propose a model linking self-compassion, social support, stress management, and academic engagement, seeking to clarify both direct and indirect effects among these constructs.

Consistent with COR Theory [8, 22, 26, 28], which emphasizes acquiring and maintaining vital resources, we posit that self-compassion and social support function as crucial personal and social resources. These resources are hypothesized to promote academic engagement by helping students manage stress effectively [15, 22, 68]. Self-compassion, for instance, may directly enhance academic engagement by fostering emotional resilience; students approaching challenges with kindness and self-acceptance are less likely to be discouraged [50]. Through self-compassion, students maintain motivation by viewing difficulties as part of learning and responding constructively [10, 12]. Additionally, self-compassion cultivates mindfulness, enabling concentration without rumination, thereby supporting deeper engagement [10].

Social support, also aligning with COR Theory, acts as a valuable social resource, providing emotional and instrumental assistance through relationships with peers, family, or educators [10, 17, 34]. This support fulfills the SDT-related need for relatedness, fostering a sense of connection and value within the academic environment [16, 37]. Research suggests supportive relationships alleviate isolation, promoting a more engaged and motivated learning approach [58].

Given that the academic environment often presents stressors that can inhibit engagement, we propose stress management as a mediating factor between self-compassion, social support, and academic engagement. From a COR Theory perspective, stress management serves as a resource-conserving strategy. Students with high self-compassion or strong social support may utilize effective coping mechanisms like problem-solving, relaxation, and time management to buffer against academic stress [22, 28, 31, 68]. Similarly, social support offers external resources and stability against academic challenges [19, 21]. These stress management strategies, facilitated by self-compassion and social support, can create a more conducive learning environment, thereby indirectly promoting sustained academic engagement.

By examining these interconnections, this study offers a theoretically grounded framework for understanding factors that promote student engagement. Our hypothesized model posits that self-compassion and social support will exert direct effects on academic engagement, with stress management mediating both relationships.

Hypotheses:

• H1: Self-compassion is directly associated with increased academic engagement among university students. Guided by SDT, this relationship is expected as self-compassion fosters autonomy and competence, contributing to intrinsic motivation and academic commitment [6, 38].

• H2: Social support is directly associated with increased academic engagement among university students. COR Theory and SDT suggest that social support fulfills students’ need for relatedness, thereby fostering a sense of belonging that enhances engagement [8, 15, 27, 37].

• H3: Stress management mediates the relationship between self-compassion and academic engagement. Students with higher self-compassion are expected to employ more effective stress management strategies, which aligns with COR Theory’s emphasis on resource conservation, ultimately leading to higher levels of engagement [22, 31, 68].

• H4: Stress management mediates the relationship between social support and academic engagement. Students with robust social support networks are anticipated to experience lower stress levels, enabling greater academic engagement. This mediation aligns with COR Theory’s view of social support as a protective resource that promotes resilience and sustained engagement [19, 21].

Methods and materials

Participants

This study employed a cross-sectional design to investigate the relationships between self-compassion, social support, stress management, and academic engagement among Chinese university students. Convenience sampling was utilized, targeting students enrolled in various undergraduate programs at two universities in major Chinese cities: Tianjin and Shanghai. University administrators were contacted to secure permission for the study, and departmental emails were used to invite students to participate. Inclusion criteria specified that participants must be currently enrolled in an undergraduate program at one of the participating universities and at least 18 years of age.

A total of 389 students (approximately 37% male) volunteered to participate in the study. The average participant age was 20.2 years (SD = 1.5 years). The sample reflected the diversity of academic programs offered at the universities, with students from social sciences (e.g., psychology, sociology), humanities (e.g., history, literature), engineering (e.g., mechanical, computer), and natural sciences (e.g., biology, chemistry) represented.

Data collection occurred online over a three-month period using Qualtrics, a secure online survey platform. The survey consisted of four self-report measures, originally developed in English and then meticulously translated into Mandarin Chinese to ensure clarity for participants. The translated measures were then back-translated into English to verify their accuracy. To ensure participant autonomy and privacy, this study adhered to strict ethical principles. Participation was completely voluntary and anonymous. Before commencing the survey, participants received a detailed informed consent form explaining the study’s purpose, data collection methods, and their right to withdraw at any point without repercussions. The research protocol gained ethical approval from the Institutional Review Board at Tiangong University (Tianjin 300387, China), upholding the highest ethical standards for research involving human subjects.

To determine an adequate sample size for our Structural Equation Modeling (SEM) analyses, we conducted an a priori power analysis using the semPower R package [75]. This analysis was conducted for our specific hypothesized model, which was specified with 19 parceled indicators (see Sect. 3.3 for the rationale for parceling). The full structural model, including both measurement and structural components, involved an estimated 49 parameters (comprising 19 factor loadings, 19 error variances, 6 latent correlations among exogenous variables, and 5 structural paths linking the latent variables). Using semPower, we targeted 80% power at α = 0.05. For detecting a good model fit (null hypothesis: RMSEA ≤ 0.05), a minimum of approximately 180 participants was required. For detecting individual path coefficients with a small-to-medium effect size (β = 0.20), approximately 250 participants were needed. Our final sample of 389 participants comfortably exceeded these requirements for the specified model, ensuring sufficient statistical power for robust parameter estimation [47, 87].

Additionally, our sample size aligns with established SEM guidelines. With 83 items across all measures, the sample-to-item ratio was 4.7:1. The sample-to-parameter ratio was approximately 7.9:1 (389/49), exceeding common recommendations [76, 77]. Acceptable fit indices from the confirmatory factor analysis (CFI = 0.94, TLI = 0.92, RMSEA = 0.07) further supported the sample’s adequacy for reliable parameter estimation. These considerations confirm the robustness of our SEM analyses.

Instruments

We used several well-established, psychometrically sound self-report scales to assess the key constructs. As these instruments were originally developed and validated in Western contexts, a rigorous cross-cultural adaptation process was undertaken to ensure their psychometric integrity and appropriateness for our Chinese university student sample.

To establish linguistic and conceptual equivalence, the measures underwent a comprehensive translation and back-translation procedure. First, two bilingual researchers independently translated the items into Mandarin Chinese, followed by a reconciliation session to resolve discrepancies. A separate team of blind bilingual researchers then back-translated the consensus Mandarin version into English. Finally, a panel of three subject-matter experts, who are fluent in both languages and possess expertise in the relevant psychological constructs, reviewed all versions to verify semantic equivalence and conceptual fidelity.

Following the translation process, we conducted a pilot study with 25 Chinese university students to evaluate item clarity, comprehensibility, and cultural relevance. Feedback from this pilot informed minor wording adjustments, which enhanced the face validity of the measures. This systematic adaptation process was a critical prerequisite for ensuring our measures were sound before proceeding with Confirmatory Factor Analysis (CFA) to formally evaluate the construct validity of the adapted scales.

Stress management

We utilized a shortened version of the Coping Orientation for Everyday Problems (Brief-COPE) developed by Carver et al. [78]. This 28-item inventory measures various coping strategies individuals employ when facing challenges. Conceptually, the Brief-COPE is designed to capture 14 theoretically distinct coping strategies (e.g., active coping, planning, emotional support, self-blame), which generally group into higher-order problem-focused and emotion-focused dimensions. For the purpose of this study, and consistent with previous research that has treated adaptive coping strategies as a unified construct in broader models, we conceptualized Stress Management as a single latent construct reflecting individuals’ overall repertoire of effective coping behaviors. Participants rated the frequency of each coping behavior on a 4-point Likert scale, ranging from “I haven’t been doing this at all” to “I’ve been doing this a lot.” Construct validity was confirmed via CFA of this unidimensional model to assess how well the items aligned with this unified conceptualization (χ²/df = 2.8, CFI = 0.93, TLI = 0.91, RMSEA = 0.05 [0.04, 0.06], RMSR = 0.04), and the scale demonstrated good reliability with a Cronbach’s alpha of 0.81.

Perceived social support

The Multidimensional Scale of Perceived Social Support (MSPSS) by Zimet et al. [79] was used to gauge participants’ perception of social support received from family, friends, and significant others during their academic journey. This 12-item scale is theoretically structured to assess perceived support across three distinct dimensions: Support from Family, Support from Friends, and Support from a Significant Other [79]. It utilized a 6-point Likert scale, with responses ranging from “strongly disagree” to “strongly agree,” indicating the level of perceived support. Construct validity was confirmed via CFA to verify this hypothesized three-factor structure (χ²/df = 3.0, CFI = 0.92, TLI = 0.90, RMSEA = 0.06 [0.05, 0.07], RMSR = 0.05), and the scale demonstrated good reliability with a Cronbach’s alpha of 0.88.

Self-compassion

The Self-Compassion Scale (SCS) developed by Neff [10] assessed participants’ self-compassionate attitudes when encountering difficulties. This 26-item scale is theoretically comprised of six distinct components: Self-Kindness, Self-Judgment (reverse-scored), Common Humanity, Isolation (reverse-scored), Mindfulness, and Over-Identification (reverse-scored) [10, 30]. It explores the extent to which individuals treat themselves with kindness and understanding rather than harsh self-criticism, recognize suffering as a common human experience, and manage negative emotions constructively. Participants responded on a 5-point Likert scale, ranging from “almost never” to “almost always.” For descriptive purposes, a composite self-compassion score was calculated by averaging scores from all six subscales, with self-judgment, isolation, and over-identification reverse-scored. The validity was approved via CFA to confirm this underlying six-factor structure (χ²/df = 2.5, CFI = 0.95, TLI = 0.94, RMSEA = 0.05 [0.04, 0.06], RMSR = 0.04), and the scale demonstrated excellent reliability with a Cronbach’s alpha of 0.91.

Academic engagement

The Chinese version of the Utrecht Work Engagement Scale for Students (UWES-S) by Gan et al. [80], based on the original UWES-S by Schaufeli et al. [81], measured participants’ academic engagement. This 17-item scale comprises three theoretically distinct subscales: Vigor (energy and mental resilience), Dedication (a strong sense of significance and enthusiasm towards studies), and Absorption (full concentration and engrossment in academic tasks). Participants rated their level of agreement with each statement on a 7-point Likert scale ranging from “never” (0) to “every day” (6), with higher scores reflecting greater levels of engagement. Construct validity was confirmed via CFA to verify this hypothesized three-factor structure (χ²/df = 2.4, CFI = 0.96, TLI = 0.95, RMSEA = 0.04 [0.03, 0.05], RMSR = 0.03), and the scale demonstrated good reliability with a Cronbach’s alpha of 0.85.

Statistical analysis

Data analysis employed SPSS and AMOS software. Following common practices [76], a two-step structural equation modeling (SEM) approach was used. To operationalize our latent variables, we employed item parceling. This methodological approach is recommended for complex models with constructs measured by numerous items, as it enhances model parsimony, stability, and the psychometric properties of indicators. Specifically, parceling reduces item-level measurement error and creates more reliable, normally distributed indicators, thereby providing a clearer test of the theoretical relationships between latent constructs [82]. This decision was based on these methodological grounds and was independent of statistical power considerations, for which our sample size was already confirmed as adequate (see Sect. 3.1).

The 19 parcels were created using a subscale-based approach: Self-Compassion (26 items) was represented by 6 parcels, formed by averaging items within each of its six theoretical subscales. Perceived Social Support (12 items) was reduced to 3 parcels, one for each subscale. Stress Management (28 items) formed 6 parcels based on a conceptual grouping of the 14 coping strategies into adaptive and maladaptive categories (three parcels each). Academic Engagement (17 items) produced 4 parcels: two for the Vigor subscale and one each for Dedication and Absorption.

First, confirmatory factor analysis (CFA) assessed the measurement model using these parceled indicators. Then, the structural model was tested. Multiple fit indices evaluated model fit, including Root Mean Square Error of Approximation (RMSEA) (values below 0.05 indicated a good fit and those between 0.05 and 0.08 indicated an acceptable fit; [77]), Standardized Root Mean Square Residual (SRMR) (considered acceptable if ≤ 0.08; [76]), chi-square/degrees of freedom ratio maintained between 1 and 3, Comparative Fit Index (CFI) expected to be above 0.90 (with values above 0.95 indicating excellent fit), and other criteria encompassing a Goodness-of-Fit Index (GFI) greater than 0.90 and a Tucker-Lewis Index (TLI) exceeding 0.90 [77].

Results

Confirmatory factor analysis (CFA)

Following the assessment of common method bias, we conducted a Confirmatory Factor Analysis (CFA) using AMOS software (version 26.0) to evaluate the adequacy of our hypothesized measurement model. This CFA was performed using the 19 parceled indicators for the latent constructs (self-compassion, social support, stress management, and academic engagement). While the χ2 statistic was statistically significant (χ² = 287.12, df = 160, χ2/ df = 2.45, p < .001), its known sensitivity to sample size in larger datasets led us to prioritize alternative fit indices for a more nuanced evaluation of model fit [77].

The additional fit indices indicated an acceptable to good model fit: RMSEA = 0.07 (within the recommended range of 0.05–0.08) [77]; SRMR = 0.06 (suggesting good fit); CFI = 0.94; and TLI = 0.92. These values, meeting or exceeding conventional cut-offs (e.g., 0.90 for CFI/TLI), provided strong evidence that the hypothesized model adequately represented the data. This strengthened our confidence that the latent constructs were well-captured by their designated measurement items, establishing a suitable framework for testing our hypotheses regarding the relationships between the study variables.

Table 1 presents the descriptive statistics, reliability coefficients, and normality indices for the study variables. Mean scores indicated moderate levels across all constructs: self-compassion (M = 3.78, SD = 0.69), social support (M = 4.02, SD = 0.72), stress management (M = 3.45, SD = 0.84), and academic engagement (M = 3.21, SD = 0.78).

Table 1.

Descriptive statistics

Construct	M (SD)	Cronbach’s α	Skewness	Kurtosis
Self-Compassion	3.78 (0.69)	0.91	0.18	2.02
Social Support	4.02 (0.72)	0.88	0.09	1.78
Stress Management	3.45 (0.84)	0.81	0.37	2.81
Academic Engagement	3.21 (0.78)	0.85	0.42	3.15

All four constructs exhibited good internal consistency, with Cronbach’s alpha coefficients ranging from 0.81 to 0.91 (self-compassion: α = 0.91; social support: α = 0.88; stress management: α = 0.81; academic engagement: α = 0.85). These values indicate that items within each scale were highly correlated and reliably measured their respective constructs.

Normality of data distribution was assessed via skewness and kurtosis values. All values fell within acceptable thresholds for parametric analyses (skewness: -2 to + 2; kurtosis: -7 to + 7) [84]. Specifically, skewness values ranged from 0.09 to 0.42, and kurtosis values from 1.78 to 3.15. This suggests the data approximated a normal distribution, justifying the use of parametric statistical techniques in subsequent analyses.

Convergent and discriminant validity

To evaluate the quality of the measurement model, we assessed both convergent and discriminant validity. Convergent validity confirms that items within a construct measure the same underlying concept, while discriminant validity ensures constructs are distinct from each other [83].

Table 2 presents the average variance extracted (AVE) and the square root of AVE for each construct. Convergent validity was supported as all AVE values exceeded the 0.50 threshold [84], ranging from 0.72 to 0.80 (self-compassion = 0.80, social support = 0.78, stress management = 0.75, academic engagement = 0.72). This indicates that a substantial proportion of variance in indicators was captured by their respective constructs. Furthermore, the square root of each construct’s AVE was greater than its correlations with all other constructs in Table 2, providing additional evidence for convergent validity.

Table 2.

Convergent and discriminant validity

Construct	AVE	Square root of AVE	1	2	3	4
1. Self-Compassion	0.80	0.89	-	0.54**	0.62**	0.48**
2. Social Support	0.78	0.88		-	0.58**	0.51**
3. Stress Management	0.75	0.87			-	0.65**
4. Academic Engagement	0.72	0.85				-

Discriminant validity was assessed using two methods. First, comparing each construct’s AVE with the squared correlations between constructs [83] showed that all squared correlations were lower than the corresponding AVE values (e.g., self-compassion and social support squared correlation (0.54²) < self-compassion AVE (0.80)). This pattern suggests sufficient distinctiveness among constructs. Second, a more stringent Heterotrait-Monotrait (HTMT) analysis [85] was performed. All HTMT ratios fell below the 0.85 threshold [86], ranging from 0.62 (between self-compassion and social support) to 0.78 (between self-compassion and stress management). These consistent findings strongly support discriminant validity, confirming the constructs are unique.

Structural equation modeling

All constructs—self-compassion, social support, stress management, and academic engagement—were treated as latent variables, measured by their respective survey items. We tested the hypothesized model, examining the proposed relationships, using SEM with AMOS software. The overall model fit indices indicated a good to acceptable solution: χ2 = 312.47, df = 160, χ2/ df = 1.95, p < .001; RMSEA = 0.07 (within the acceptable range of 0.05–0.08) [77]; SRMR = 0.05 (indicating good fit); CFI = 0.95; and TLI = 0.93 (exceeding conventional cut-offs of 0.90 and 0.95, respectively) [77]. These results provide strong evidence that the hypothesized model fits the data well and that the proposed relationships are supported. Path coefficients are presented in Fig. 1.

Fig. 1.

The Final Model of Academic Engagement Among Chinese University Students

Table 3 summarizes the estimated path coefficients (β), their 95% confidence intervals (CIs), and p-values for the hypothesized relationships. All hypothesized direct effects were statistically significant (p < .001). Higher self-compassion showed a positive direct effect on academic engagement (β = 0.24), as did social support (β = 0.18). Stress management demonstrated the strongest positive direct effect on academic engagement (β = 0.42).

Table 3.

Path coefficients

Path	β	Bootstrap Confidence Interval [2.5%; 97.5%]	p-value
Direct Effects
Self-Compassion → Academic Engagement	0.24	[0.17; 0.31]	< 0.001
Social Support → Academic Engagement	0.18	[0.12; 0.24]	< 0.001
Stress Management → Academic Engagement	0.42	[0.35; 0.49]	< 0.001
Self-Compassion → Stress Management	0.38	[0.31; 0.45]	< 0.001
Social Support → Stress Management	0.29	[0.22; 0.36]	< 0.001
Indirect Effects
Self-Compassion → Stress Management → Academic Engagement	0.16	[0.10; 0.22]	< 0.001
Social Support → Stress Management → Academic Engagement	0.12	[0.08; 0.17]	< 0.001
Total Effects
Self-Compassion → Academic Engagement	0.40	[0.32; 0.48]	< 0.001
Social Support → Academic Engagement	0.30	[0.22; 0.38]	< 0.001

Regarding the independent variables’ influence on stress management, self-compassion exhibited a significant positive direct effect (β = 0.38). Social support also showed a significant positive direct effect on stress management (β = 0.29), indicating its benefit in managing academic stress.

The analysis further revealed significant indirect effects on academic engagement. Self-compassion indirectly influenced academic engagement through stress management (β = 0.16), suggesting self-compassion partly supports engagement by fostering better stress management skills. Similarly, social support had a significant indirect effect on academic engagement mediated by stress management (β = 0.12), indicating its indirect enhancement of engagement through effective stress coping. The total effects of self-compassion (β = 0.40) and social support (β = 0.30) on academic engagement were both statistically significant (p < .001), reflecting their combined direct and indirect influences. These findings consistently indicate that students with higher self-compassion and stronger social networks tend to be more academically engaged.

Common method bias

Given the self-report nature of our data, we employed a two-pronged approach to address potential common method bias (CMB) [87]. First, we utilized distinct, well-established scales for each construct (self-compassion, social support, stress management, and academic engagement). This approach reduces the likelihood of a single method factor influencing all measures, thus helping to ensure that observed relationships reflect each construct’s unique characteristics.

Second, a post-hoc one-factor exploratory factor analysis (EFA) with principal component analysis (PCA) was conducted in SPSS 26.0, entering all measurement items into a single factor model. If substantial CMB were present, a dominant single factor would explain a large portion of the total variance. However, the first extracted factor accounted for only 40.67% of the total variance, falling well below the suggested threshold for significant CMB [87]. These results indicate that CMB is unlikely to heavily influence our data, strengthening confidence in the observed relationships.

Discussion

The current study aimed to investigate the relationships between self-compassion, social support, stress management, and academic engagement among university students. The findings provided strong support for our hypotheses, demonstrating that both self-compassion [H1] and social support [H2] are directly and positively associated with academic engagement, while stress management appears to play a mediating role in the relationships between self-compassion and academic engagement [H3] and between social support and academic engagement [H4]. These findings are consistent with the significance of both personal and social resources as outlined in Hobfoll’s COR Theory [8, 26], which posits that individuals strive to acquire and maintain resources to buffer against stress, and Deci and Ryan’s SDT, which emphasizes that intrinsic motivation is achieved when psychological needs for autonomy, competence, and relatedness are met [9, 27].

Our findings highlight that self-compassion and social support are linked to academic engagement, potentially functioning as internal and external resources, respectively. The direct association of self-compassion with academic engagement is in line with research suggesting that self-compassion may provide emotional resilience, potentially allowing students to maintain their motivation and persistence in academic tasks despite facing challenges [11, 12, 14, 51, 52]. This association could be attributed to the three core elements of self-compassion—self-kindness, common humanity, and mindfulness—which collectively may foster a positive, growth-oriented mindset essential for engagement. Self-kindness, or treating oneself with understanding during setbacks, is theorized to reduce self-criticism and negative emotional responses, thereby conserving psychological resources that can be redirected towards academic efforts [37]. Common humanity may encourage students to recognize their struggles as part of a shared human experience, potentially promoting a sense of connectedness and reducing feelings of isolation that might otherwise deter engagement [41]. Finally, mindfulness could enable students to approach academic stressors without judgment, which may support emotional regulation and allow for sustained focus on academic tasks [23, 30].

In terms of SDT, self-compassion appears to align with the need for competence by reinforcing students’ sense of efficacy through resilience and self-efficacy. When students respond to academic challenges with self-compassion, they may be more likely to perceive these challenges as opportunities for growth, potentially fostering a sense of competence and internal motivation that could fuel engagement [31].

The direct association of social support with academic engagement further reinforces the perceived importance of external resources for academic success. Social support may meet the SDT-related need for relatedness, potentially providing students with a sense of connection and community essential for sustained engagement [18, 19, 21, 38, 58, 59]. Instrumental support, such as assistance with coursework or academic guidance from professors and peers, could address practical needs, potentially allowing students to focus more effectively on their studies [16, 18]. Emotional support, characterized by empathy and validation, may create an environment of security and belonging, which research shows is crucial for maintaining engagement even during high-stress periods [34, 58]. This aligns with findings that a strong sense of social connectedness may buffer against feelings of isolation, which can otherwise hinder academic motivation [45, 61].

In addition to these direct pathways, our results indicate that stress management mediates the relationships between self-compassion, social support, and academic engagement, suggesting that these resources may promote engagement indirectly by being associated with students’ ability to manage academic stress effectively. COR Theory posits that resourceful individuals can draw on personal and social resources to protect against stressors, potentially enabling sustained engagement in academic activities [8, 26]. In this context, self-compassion appears to enhance stress management by being associated with adaptive coping mechanisms such as problem-solving, self-regulation, and mindfulness [12, 31, 50]. For instance, self-kindness could reduce self-criticism during stressful academic periods, potentially allowing students to preserve psychological energy and approach challenges with resilience [10]. Mindfulness further may empower students to observe stressors from a balanced perspective, potentially enabling them to manage stress without becoming overwhelmed by anxiety or negative emotions [23, 50].

Social support similarly appears to bolster stress management by providing students with both emotional and instrumental resources that may reduce stress in academic contexts. Instrumental support, like assistance with academic tasks, could alleviate direct sources of stress, potentially enabling students to focus on academic engagement without distraction [18]. Emotional support, characterized by care and empathy, may reduce the psychological toll of stress, potentially fostering resilience and encouraging students to persist in their studies [16, 34]. Social companionship, through shared enjoyable activities, could offer additional relief from stress, creating opportunities for emotional renewal that are crucial for sustained focus on academic responsibilities [17, 35]. By being associated with improved stress management, self-compassion and social support may help students conserve psychological resources and maintain the motivation and focus necessary for active engagement. This mediating effect is consistent with Conley et al.’s [24] findings that stress management interventions significantly improve student engagement and well-being.

These findings contribute to the theoretical understanding of academic engagement by integrating COR Theory and SDT to explain both the direct associations of self-compassion and social support with engagement, and the indirect associations mediated through stress management. Practically, this research suggests potential avenues for universities to support student engagement by developing programs that focus on self-compassion, social support, and effective stress management. For example, incorporating self-compassion training into student support services and fostering a supportive campus culture could provide students with the personal and social resources that may be necessary to engage fully in their academic pursuits, even in the face of stress. Overall, this study highlights the vital roles of self-compassion and social support in their associations with academic engagement, both directly and indirectly through stress management. Future research could further examine these relationships across different cultural contexts and utilize longitudinal or experimental designs to capture the long-term effects of these resources on academic success and to establish causality.

Conclusion

The current study investigated the relationships between self-compassion, social support, stress management, and academic engagement among a sample of Chinese university students. The findings provided strong support for our hypotheses. Self-compassion and social support were directly associated with academic engagement, suggesting that students who perceive themselves with kindness and understanding, and who have access to supportive social networks, tend to be more actively engaged in their studies. Additionally, stress management emerged as a mediating factor in the relationships between both self-compassion and social support with academic engagement. These results highlight the potential interplay between these psychological well-being factors in influencing student success within the Chinese university context.

This study contributes to the growing body of research exploring the multifaceted nature of academic engagement. By demonstrating the positive associations between self-compassion, social support, stress management, and academic engagement, our findings offer valuable insights into the psychological mechanisms that may underlie student motivation and participation in learning activities. The mediating role of stress management sheds light on the potential pathways through which self-compassion and social support may indirectly be linked to greater academic engagement by fostering effective stress management strategies. These findings enrich our theoretical understanding of the complex interplay between these factors and academic engagement.

The present study holds significant practical implications for potentially promoting academic engagement in Chinese universities. The findings underscore the importance of self-compassion and social support in their association with academic engagement, suggesting several actionable recommendations for educators and policymakers. Firstly, the emphasis on self-compassion indicates that interventions designed to cultivate self-kindness and self-acceptance may be valuable tools for students navigating academic challenges. Universities could consider incorporating self-compassion training programs or workshops into their student support services. Such programs could include mindfulness exercises, self-compassion meditations, and cognitive-behavioral techniques aimed at reducing self-criticism and fostering a growth-oriented mindset. Research by Neff and Germer [50] has supported the effectiveness of these interventions in reducing stress and enhancing emotional resilience in other contexts. Additionally, integrating self-compassion modules into the existing curriculum of first-year seminars or orientation programs could help students develop these skills early in their academic journey.

Furthermore, the importance of social support networks highlights the need for fostering a strong sense of community within universities. Initiatives that promote connections between students, faculty, and staff can potentially cultivate a supportive environment that may bolster academic engagement. Peer mentoring programs, where upperclassmen provide guidance and support to new students, can foster a sense of belonging and may reduce feelings of isolation. Social events and group activities that encourage interaction and collaboration among students from diverse backgrounds can also strengthen community bonds. Studies by Wang and Eccles [38] have emphasized the role of social support in fostering a sense of belonging and enhancing academic engagement in other populations.

Additionally, universities should strengthen communication channels between students and support services, such as counseling centers or academic advisors, to ensure that students have access to resources and support during challenging times. Implementing a robust student advising system, where academic advisors regularly check in with students to discuss their progress and challenges, can help identify and address potential issues early. Enhancing the visibility and accessibility of support services on campus is also crucial. Providing information about these services during orientation, through online portals, and via campus-wide communications can encourage students to seek help when needed.

These practical implications are offered to inform the development of potential evidence-based interventions tailored to the specific needs of Chinese university students. Future research could explore the effectiveness of such interventions in their ability to enhance self-compassion, social support networks, and stress management skills, ultimately contributing to academic engagement. Crucially, longitudinal studies or experimental designs examining the long-term impact of these interventions on academic engagement would provide valuable insights into their sustainability and effectiveness and could establish causal links.

Limitations

This research has some inherent constraints that should be considered when interpreting the results. First, the cross-sectional design employed here limits our ability to definitively establish cause-and-effect relationships between the variables studied. Future investigations utilizing longitudinal designs could offer a clearer picture of the directional nature of these associations, while experimental designs would be necessary to infer causality. Second, while this study provides robust evidence for the construct validity of the measures, its scope did not extend to assessing external forms of validity. Future research should examine the predictive validity of these constructs by testing their ability to forecast objective, long-term outcomes such as student GPA, retention rates, and post-graduation success. Establishing such links would significantly bolster the practical utility of our findings.

Third, the reliance on self-reported data introduces the possibility of bias, such as participants responding in a socially desirable manner or having difficulties with accurate recall. The inclusion of objective measures for stress or academic performance could strengthen the validity of future findings. Furthermore, the convenience sampling method used in this study may limit the generalizability of the results to the entire population of Chinese university students. Future research efforts could benefit from employing more representative sampling strategies or replicating the study in a wider range of university settings.

Abbreviations

AVE

Average variance extracted

CFI

Comparative fit index

CMB

Common method bias

COR Theory

Conservation of resources theory

CFA

Confirmatory factor analysis

EFA

Exploratory factor analysis

GFI

Goodness-of-fit index

HTMT

Heterotrait-monotrait ratio

MBSR

Mindfulness-based stress reduction

MSPSS

Multidimensional scale of perceived social support

RMSEA

Root mean square error of approximation

SDT

Self-determination theory

SEM

Structural equation modeling

SRMR

Standardized root mean square residual

TLI

Tucker-lewis index

UWES-S

Utrecht work engagement scale for students

Author contributions

Each author contributed significantly to this study, including conception, design, data acquisition, analysis, interpretation, drafting, revising, and critically reviewing the article. All authors provided final approval for publication and agreed on the chosen journal. They also accept responsibility for all aspects of the work.

Funding

No external funding was received for this research project.

Data availability

For access to the data utilized in this study, please contact the corresponding author.

Declarations

Ethics approval and consent to participate

Approval for this study’s protocol was granted by the Ethics Approval Committee at the Foreign Languages College, Tianjin Normal University, Tianjin 300387, China. The committee ensured that the research posed no harm to participants and that all procedures adhered to relevant guidelines and regulations. Written informed consent was obtained from all participants, who were briefed on the study’s objectives, given the chance to seek clarification, and freely consented to participate with the understanding of their right to withdraw at any time without explanation or repercussions.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.