Supervisor–student relationship and anxiety among graduate students: a moderated mediation model

Abstract

Background

The psychological crises associated with poor supervisor–student relationship among graduate students have raised growing public concerns. In the Chinese graduate education context, supervisors play a central and highly influential role in students’ academic development, which makes the supervisor–student relationship particularly critical for their psychological well-being. This study examined the association between supervisor–student relationship and anxiety, the mediating role of general self-efficacy, and the moderating role of attitudes toward seeking professional psychological help among Chinese graduate students.

Methods

A total of 1233 Chinese graduate students (36.6% female; M_age = 24.19, SD = 2.21, range = 20–40 years) participated in the study and completed measures of Teacher–Student Relationship Quality Questionnaire, General Self-Efficacy Scale, Attitudes Toward Seeking Professional Psychological Help Scale-Short Form, and the Generalized Anxiety Disorder-7. All measurement instruments demonstrated good reliability and validity in the present sample. The data were analyzed using a moderated model with SPSS 25.0 and the supplemental PROCESS macro 4.2.

Results

Supervisor–student relationship was significantly negatively associated with anxiety (β = −0.364, p < 0.001), and mediation analyses indicated that general self-efficacy partially mediated this association (indirect effect = − 0.152, 95% CI = [− 0.192, − 0.115]). Moderated mediation analysis further revealed that attitudes toward seeking professional psychological help buffered the adverse effects of supervisor–student relationship on anxiety (β = 0.106, p < 0.001), as well as the correlation between supervisor–student relationship and general self-efficacy (β = −0.119, p < 0.001).

Conclusions

This study demonstrates that the quality of the supervisor–student relationship is closely associated with graduate students’ anxiety and that this link operates partly through general self-efficacy. Positive attitudes toward seeking professional psychological help weakened the association between poor supervisor–student relationship and anxiety and mitigated the decline in general self-efficacy. These findings clarify how interpersonal and psychological factors jointly shape anxiety and suggest that strengthening supportive supervisor–student relationship and fostering favorable attitudes toward seeking professional psychological help may be beneficial for safeguarding the mental health of graduate students.

Introduction

Over the past decade, the scale of graduate education in China has achieved breakthrough growth, with the number of graduate students in 2024 reaching 4,095,500, an increase of 121.65% compared with 2014 [1, 2]. While this rapid growth has provided new opportunities for talent cultivation, it has also raised widespread concerns about the psychological well-being of graduate students. Anxiety, in particular, has been frequently identified as a major mental health challenge that undermines academic performance [3], reduces life satisfaction [4], and increases the risk of maladaptive outcomes [5]. Understanding the antecedents and mechanisms of anxiety within this population has therefore become a pressing issue in higher education and public health research. The supervisor–student relationship constitutes a defining feature of graduate education in China, where the tutor responsibility system grants supervisors extensive authority over students’ academic and professional trajectories [6, 7]. Supportive supervisor–student relationship can promote motivation, resilience, and personal growth, whereas strained or conflictual relationship may serve as significant interpersonal stressors that heighten graduate students’ vulnerability to anxiety [8–10]. Nevertheless, the psychological mechanisms through which supervisor–student relationship influences anxiety remain insufficiently clarified. General self-efficacy, defined as an individual’s overarching belief in their ability to manage challenges, represents a central personal resource that may account for these processes [11]. Supportive relationships may enhance self-efficacy by providing affirmation and constructive feedback, while poor relationships may erode it and thereby exacerbate anxiety. In addition, attitudes toward seeking professional psychological help have been emphasized as a crucial contextual factor shaping stress responses [12]. Graduate students who endorse positive attitudes are more likely to mobilize external resources when encountering difficulties, potentially buffering the adverse effects of poor supervisor–student relationship on both self-efficacy and anxiety.

Despite the theoretical relevance of these factors, empirical studies focusing on graduate students remain scarce, with the majority of prior work addressing teacher–student relationship at the primary and secondary levels. However, compared with the teacher–student relationship in other education stages, the relationship between supervisors and graduate students is closer and more complex [13, 14], and its interaction mode and influence mechanism have unique characteristics and research value. Hence, the current study addresses this gap by jointly examining supervisor–student relationship, general self-efficacy, and attitudes toward seeking professional psychological help in predicting anxiety, thereby providing a scientific basis and empirical reference for educational management departments and institutions to develop effective psychological education policies, ultimately promoting the physical and mental well-being of graduate students and enhancing the overall quality of graduate education.

Supervisor–student relationship and anxiety

The supervisor–student relationship, generally defined as the dynamic interaction between supervisors and their graduate students, constitutes a central element of graduate education [13]. Unlike relationships at earlier educational stages, this relationship is characterized by intensive academic dependence, hierarchical authority, and long-term personal involvement [15]. Consequently, the quality of the supervisor–student relationship exerts profound effects not only on academic outcomes but also on psychological health. Prior studies have demonstrated that supportive and constructive supervisor–student relationship is associated with higher levels of subjective well-being [7], greater learning adaptability [16], and enhanced self-efficacy among graduate students [13]. In contrast, strained or conflictual relationship has been linked to a range of psychological and behavioral difficulties, including anxiety [17], depression [18], sleep disturbances [19], and academic procrastination [20]. Among these outcomes, anxiety has emerged as particularly prevalent and detrimental. Anxiety is typically conceptualized as a complex emotional state involving tension, worry, and apprehension in response to perceived threats, and it has been recognized as one of the most common psychological problems among students in higher education [21]. According to the China National Mental Health Development Report (2021–2022), the detection rate of anxiety risk among Chinese college students is 45.28% [22], while the risk of anxiety disorder among graduate students is six times higher than that of the general population [23]. These statistics indicate that graduate students represent a particularly vulnerable group, and highlight the necessity of investigating interpersonal and psychological factors that contribute to their elevated anxiety.

Theoretical perspectives provide further insight into the mechanisms linking poor supervisor–student relationship to anxiety. The diathesis-stress model posits that negative life events and chronic stressors act as risk factors for the onset of emotional disorders such as depression and anxiety [24]. In this context, a poor supervisor–student relationship functions as a salient interpersonal stressor: persistent criticism, excessive control, belittling, or neglect from supervisors may subject students to prolonged stress and foster anxiety symptoms. Psychodynamic theory further emphasizes the centrality of interpersonal relationships in shaping psychological states [25]. A positive and secure relationship with a supervisor can provide a sense of belonging and safety, which helps graduate students cope with academic pressures. Conversely, experiences of rejection or exclusion may heighten insecurity and evoke anxiety. Finally, emotion theories such as the James–Lange theory suggest that emotional experiences arise from physiological arousal [26]. Within negative supervisory contexts, students may display physiological stress responses, including accelerated heartbeat and sweating, when confronted with criticism or perceived injustice, which can intensify anxious experiences.

Taken together, these perspectives highlight that the supervisor–student relationship is not merely an academic interaction but a critical psychosocial context for graduate students. Poor supervisor–student relationship may represent a chronic interpersonal stressor that, through both psychological and physiological mechanisms, substantially elevates the risk of anxiety. Nevertheless, despite the theoretical plausibility and empirical indications, systematic research on the association between supervisor–student relationship and anxiety in graduate students remains limited, underscoring the need for further investigation in this area.

The mediating effect of general self-efficacy

General self-efficacy refers to an individual’s broad and stable belief in their capacity to cope effectively with a variety of challenging situations [27]. Rooted in Bandura’s social cognitive theory [28], general self-efficacy has been conceptualized as a crucial personal resource that influences motivation, learning, and emotional regulation. Individuans with high general self-efficacy tend to approach difficulties with greater confidence, persistence, and adaptive coping strategies, whereas those with lower general self-efficacy are more likely to perceive stressors as insurmountable and to experience heightened levels of distress. Empirical studies have consistently demonstrated that general self-efficacy is negatively associated with symptoms of anxiety and depression [29, 30], highlighting its role as a psychological buffer against stress and maladjustment. In the graduate education context, the supervisor–student relationship may be a significant antecedent of students’ general self-efficacy. Supportive supervisors can foster general self-efficacy by providing constructive feedback, encouraging autonomy, and validating students’ competencies [13]. Such interactions may reinforce students’ confidence in their ability to manage both academic and personal challenges. Conversely, poor supervisor–student relationship characterized by criticism, neglect, or excessive control may undermine students’ sense of competence, thereby eroding their general self-efficacy [16]. The erosion of this personal resource is likely to increase vulnerability to anxiety [31], as students may doubt their capacity to meet the demands of graduate training and perceive academic tasks as overwhelming.

Theoretically, this mediating pathway can be understood through the lens of social cognitive theory [28] and the conservation of resources theory [32]. According to social cognitive theory, self-efficacy beliefs are shaped by social interactions and feedback from significant others, such as supervisors. A lack of supportive relational experiences may therefore diminish efficacy beliefs, which in turn intensifies stress responses. In parallel, the conservation of resources theory posits that the depletion of personal resources, such as self-efficacy, contributes to psychological strain. Within this framework, poor supervisor–student relationship can be regarded as stressors that deplete efficacy beliefs, eventually manifesting as anxiety.

Although the associations among supervisor–student relationship, general self-efficacy, and anxiety are theoretically plausible and supported by preliminary findings, empirical evidence in graduate populations remains limited. Most studies have examined these constructs separately or within younger student cohorts, leaving a critical gap in understanding the mediating role of general self-efficacy among graduate students. Clarifying this mechanism is essential for both advancing theoretical models of student mental health and informing interventions aimed at strengthening personal resources to reduce anxiety.

The moderating effect of attitudes toward seeking professional psychological help

Attitudes toward seeking professional psychological help refer to individuals’ beliefs, perceptions, and predispositions regarding the utilization of mental health services. Such attitudes encompass the perceived value of professional assistance, willingness to disclose psychological difficulties, and readiness to engage with formal support systems [33, 34]. A large body of research has shown that positive attitudes toward seeking professional psychological help are positively associated with service utilization, earlier intervention, and better mental health outcomes [35]. Conversely, negative attitudes toward seeking professional psychological help, often influenced by stigma or lack of awareness, serve as barriers to help-seeking, leaving individuals more vulnerable to the detrimental consequences of psychological stress [36]. This view is consistent with the theory of planned behavior [37], which posits that attitudes are proximal determinants of intentions and behavior, including professional help-seeking.

According to the stress-buffering model [38], the availability and utilization of supportive resources can weaken the impact of stressful experiences on mental health. In this way, attitudes toward seeking professional psychological help may serve as a crucial protective function in the face of adverse relational experiences [35]. When confronted with a poor supervisor–student relationship, students who maintain positive attitudes toward seeking professional psychological help may be more likely to reach out for counseling or other support services, thereby alleviating emotional strain and reducing the risk of developing anxiety. By contrast, students with negative attitudes towardseeking professional psychological help may internalize stress and avoid external assistance, amplifying the negative psychological impact of a poor supervisory relationship. Thus, attitudes toward seeking professional psychological help may moderate the association between supervisor–student relationship and anxiety, buffering the adverse consequences of relational stress.

Beyond its role in shaping anxiety outcomes, attitudes toward seeking professional psychological help may also influence the extent to which supervisor–student relationship affects general self-efficacy. Supportive supervisor–student relationship typically reinforces efficacy beliefs through affirmation and constructive feedback, while negative relationship undermines them [13]. However, graduate students with positive attitudes toward professional help may compensate for low relational support by seeking external guidance, skills training, or psychological counseling, which can help preserve their sense of competence. This process aligns with Bandura’s social cognitive theory [28], which emphasizes that self-efficacy beliefs are shaped not only by direct experiences but also by vicarious learning, verbal persuasion, and psychological support from external sources. In this way, positive attitudes toward seeking professional psychological help may attenuate the erosion of general self-efficacy caused by poor supervisor–student relationship, thereby weakening the indirect pathway from supervisor–student relationship to anxiety.

Despite its importance, empirical research on the moderating role of attitudes toward seeking professional psychological help in graduate students remains sparse. Most prior studies have concentrated on undergraduates or general community samples, often neglecting the unique pressures of graduate education. Investigating this moderating mechanism among graduate students is therefore essential for clarifying how interpersonal and individual factors interact to shape mental health outcomes. Such evidence can inform the design of targeted interventions that not only improve supervisor–student relationship but also promote positive attitudes toward seeking professional psychological help as complementary strategies to safeguard graduate students’ psychological well-being.

The present study

Drawing on previous studies and theories, this current study attempted to establish how general self-efficacy helped explain the process by which supervisor–student relationship is correlated with anxiety and the specific relationships among the factors. Besides, the role of attitudes toward seeking professional psychological help was assessed to test whether attitudes toward seeking professional psychological help would moderate the association between supervisor–student relationship and general self-efficacy as well as supervisor–student relationship and anxiety. Consequently, we constructed a moderated mediation model (shown in Fig. 1) to test the following hypotheses:

• Hypothesis 1: Supervisor–student relationship will be negatively correlated with anxiety.

• Hypothesis 2: General self-efficacy will mediate the relationship between supervisor–student relationship and anxiety.

• Hypothesis 3: Attitudes toward seeking professional psychological help will moderate the association between supervisor–student relationship and general self-efficacy.

• Hypothesis 4: Attitudes toward seeking professional psychological help will moderate the association between supervisor–student relationship and anxiety.

Fig. 1.

The proposed moderated model. Note: ATSPPH = Attitudes Toward Seeking Professional Psychological Help

Methods

Participants

A cross-sectional survey was conducted using convenience sampling from March to April 2025. With the assistance and support of teachers from the student affairs offices of the relevant colleges, a web-based questionnaire was administered to graduate students from 15 Chinese colleges. A total of 1350 questionnaires were distributed, and 1233 valid questionnaires were returned, representing a valid response rate of 91.33%. Of the 1233 participants, 782 (63.4%) were male and 451 (36.6%) were female. Their ages ranged from 20 to 40 years old (M = 24.19 years, SD = 2.21 years). Table 1 presents the demographic information of the study participants.

Table 1.

Demographic information of the participants (N = 1233)

Characteristics	Categories	Frequency (N)	Percentage (%)
Gender	Male	782	63.4
	Female	451	36.6
Types of students	Master student	1083	87.8
	Doctoral student	150	12.2
Types of degrees	Academic degree	814	66
	Professional degree	419	34
Region	Rural	677	54.9
	Urban	556	45.1
Marital status	unmarried	1180	95.7
	married	53	4.3
Whether as student cadres	Student cadres	892	72.3
	Non student cadres	341	27.7

Measures

Demographic questionnaire

A self-designed questionnaire was used to collect basic information about the graduate students, including gender, age, types of students, types of degrees, region, marital status, whether as student cadres. Pertinent studies [39–41] have demonstrated that these demographic variables are significant factors influencing students’ supervisor–student relationship, general self-efficacy, attitudes toward seeking professional psychological help and anxiety. Given their potential impact on the subjects within the context of this study, we regarded these variables as control variables in our analyses.

Supervisor–student relationship

This study used the Teacher–Student Relationship Quality Questionnaire (T-SRQQ) developed by Yu et al. [15] to assess the supervisor–student relationship quality of graduate students. This questionnaire was compiled on the basis of the leader member exchange relationship questionnaire. The questionnaire comprised 7 items (e.g., "My supervisor has a good understanding of my research interests, needs and expertise"). Graduate students rated each item on a 5-point scale ranging from 1 (completely inconsistent) to 5 (completely consistent) with higher scores showing indicating a higher quality of supervisor–student relationship. The sum of the scores of each item is the total score of the scale, and the total score is 7 to 35. The scale has good reliability, validity and applicability in the study of Chinese graduate students [13, 15]. For the current sample, the Cronbach’s α coefficient of the T-SRQQ was 0.925. Confirmatory factor analysis (CFA) in this study suggested that the one-factor model fit the data well: RMSEA = 0.045, CFI = 0.941, TLI = 0.912, SRMR = 0.035.

General self-efficacy

This study used the General Self-Efficacy Scale (GSE) developed by Schwarzer and Jerusalem [42] to assess the general self-efficacy of graduate students. The scale consists of 10 items (e.g., "I can solve my problems if I do my best"). Graduate students rated each item on a 4-point scale ranging from 1 (completely incorrect) to 4 (completely correct) with higher scores showing higher general self-efficacy. The sum of the scores of each item is the total score of the scale, and the total score is 10 to 40. The scale has demonstrated good reliability and validity across different populations and is widely applied in psychological research [27, 43]. In this study, the Cronbach’s α coefficient was 0.951. Confirmatory factor analysis (CFA) in this study suggested that the one-factor model fit the data well: RMSEA = 0.068, CFI = 0.947, TLI = 0.932, SRMR = 0.030.

Attitudes toward seeking professional psychological help

This study used the Attitudes Toward Seeking Professional Psychological Help Scale-Short Form (ATSPPH-SF) compiled by Fischer et al. [44] and revised by Fang et al. [45] to measure the attitudes toward seeking professional psychological help of graduate students. This scale contains two dimensions: openness to seeking treatment for emotional problems, value and need in seeking treatment. Each dimension has 5 items, straight items (e.g., items 1, 3, 5, 6 and 7) are scored 3-2-1-0, and reversal items (e.g., items 2, 4, 8, 9 and 10) 0-1-2-3, respectively, for the response alternatives agree, partly agree, partly disagree, and disagree. The total score ranges from 0 to 30, with higher scores indicating more positive attitudes toward seeking professional psychological help. Previous studies have shown that the ATSPPH-SF has good reliability and validity [40, 45]. For the current sample, the Cronbach’s α coefficient was 0.833 for the ATSPPH-SF scores overall, 0.854 for openness to seeking treatment for emotional problems, 0.829 for value and need in seeking treatment. Confirmatory factor analysis (CFA) in this study suggested that the two-factor model fit the data well: RMSEA = 0.059, CFI = 0.931, TLI = 0.908, SRMR = 0.065.

Anxiety

This study used the Generalized Anxiety Disorder-7 (GAD-7) compiled by Spitzer et al. [46] to measure the anxiety level of graduate students. The scale consists of 7 items and measures the indicators of anxiety symptomatology in the last 2 weeks. Each item is rated on a 4-point Likert-type scale (0 = not at all; 1 = several days; 2 = more than half the day; 3 = nearly every day). The total score, ranging from 0 to 21, indicates levels of anxiety. A score of 10 or greater on the GAD-7 represents a reasonable cut point for identifying cases of GAD. Cut points of 5, 10, and 15 might be interpreted as representing mild, moderate, and severe levels of anxiety on the GAD-7. Although the GAD-7 was originally developed for clinical screening [47], it has become one of the most widely used instruments for assessing general anxiety in non-clinical populations [48, 49], demonstrating good performance and adequate diagnostic [50]. In this study, the Cronbach’s α coefficient of the GAD-7 was 0.927. Confirmatory factor analysis (CFA) in this study suggested that the one-factor model fit the data well: RMSEA = 0.076, CFI = 0.955, TLI = 0.933, SRMR = 0.032.

Procedure

The study was approved by the ethics committee of the first author’s institution. All participants provided individual informed consent prior to taking part in the study. The survey was conducted using Questionnaire Star (https://www.wjx.cn), a commonly used tool in China for data collection, survey administration, and various research purposes. We generated a QR code for distribution, which reflects the content of the questionnaire, and entrusted the teachers of the Student Affairs Office of the target colleges to implement it specifically. Prior to the survey, teachers responsible for its implementation were trained through online meetings to ensure they understood the entire measurement process. During the survey phase, standardized instructions were provided, emphasizing the confidentiality of the questionnaire and that there were no right or wrong answers. Additionally, participants were informed that they could withdraw from the survey at any time without penalty, and that the data collected would be used solely for research purposes. The researchers collected all completed questionnaires, screened out invalid questionnaires (e.g., with missing data), and entered participants’ responses into SPSS for data analysis.

Data processing

This study used SPSS 25.0 and the supplemental PROCESS macro 4.2 for data analysis [51]. Firstly, because all the data were self-reported by participants, common method bias is a possibility. Harman’s single-factor test was used to test for common method bias [52]. Secondly, descriptive statistics and correlation analysis were conducted for the study variables. Finally, the PROCESS macro for SPSS (Model 4) was applied to examine the mediating role of general self-efficacy, and the PROCESS macro for SPSS (Model 8) was used to investigate the moderating role of attitudes toward seeking professional psychological help [51]. Demographic variables (gender, age, types of students, types of degrees, region, marital status and whether as student cadres) were controlled for in the correlation and moderated mediation analysis. The bootstrap confidence intervals (CIs) based on 5,000 resamples were used to determine whether the effects in Model 4 and Model 8 were statistically significant [51]. The effects were regarded as significant if the 95% bootstrap CIs did not include zero [51]. All continuous variables were standardized in model 4 and model 8 before data analysis.

Results

Common method bias test

All the data were self-reported by participants, so common method bias is a possibility. Harman’s single-factor test was used to test for common method bias [52]. The results of non-rotating principal component factor analysis identified five factors with eigenvalues greater than 1. The first factor explained 32.42% of the total variance, which is below the critical value of 40% [52, 53], indicating that common method bias is not a serious issue in this study.

Descriptive statistics and correlation analysis

Means, standard deviations, and partial correlations for the study variables are shown in Table 2. Supervisor–student relationship is positively correlated with general self-efficacy (r = 0.503, p < 0.01) and attitudes toward seeking professional psychological help (r = 0.215, p < 0.01), but negatively correlated with anxiety (r = − 0.362, p < 0.01). General self-efficacy is positively correlated with attitudes toward seeking professional psychological help (r = 0.212, p < 0.01), but negatively correlated with anxiety (r = − 0.406, p < 0.01). In addition, Attitudes toward seeking professional psychological help is negatively correlated with anxiety (r = − 0.226, p < 0.01). Thus, the correlation results provide preliminary support for the subsequent mediation analysis.

Table 2.

Means, standard deviations, and correlation coefficients of the study variables

Variables	M	SD	1	2	3	4
1.Supervisor–student relationship	3.615	0.799	—
2.General self-efficacy	2.673	0.655	0.503**	—
3.ATSPPH	1.761	0.383	0.215**	0.212**	—
4.Anxiety	0.556	0.579	-0.362**	-0.406**	-0.226**	—

n = 1233, *p < 0.05, **p < 0.01, ***p < 0.001, ATSPPH Attitudes Toward Seeking Professional Psychological Help

Testing for mediation effect

Hayes’ PROCESS macro Model 4 for SPSS was applied to examine the mediating role of general self-efficacy in the association between supervisor–student relationship and anxiety. As shown in Table 3, supervisor–student relationship significantly negatively predicted anxiety (β = −0.364, t = − 13.587, p < 0.001) but positively predicted general self-efficacy (β = 0.504, t = 20.538, p < 0.001). When general self-efficacy was included in the model, it remained a significant negative predictor of anxiety (β = −0.302, t = − 10.138, p < 0.001), while the association between supervisor–student relationship and anxiety was reduced yet still significant (β = −0.212, t = − 7.123, p < 0.001), indicating partial mediation.

Table 3.

Mediating role of general self-efficacy between supervisor–student relationship and anxiety

Predictors	Anxiety			General self-efficacy			Anxiety
	β	t	95% CI	β	t	95% CI	β	t	95% CI
Gender	-0.168	-2.98**	[-0.278,-0.057]	0.122	2.357*	[0.021,0.224]	-0.131	-2.412*	[-0.237,-0.024]
Age	0.017	0.563	[-0.043,0.077]	-0.005	-0.161	[-0.06,0.051]	0.016	0.539	[-0.042,0.074]
Types of students	-0.105	-0.732	[-0.387,0.177]	-0.024	-0.184	[-0.284,0.236]	-0.113	-0.815	[-0.383,0.158]
Types of degrees	-0.045	-0.78	[-0.157,0.067]	0.022	0.414	[-0.082,0.125]	-0.038	-0.692	[-0.146,0.07]
Region	0.069	1.297	[-0.036,0.174]	0.006	0.115	[-0.091,0.103]	0.071	1.383	[-0.03,0.172]
Marital status	-0.024	-0.163	[-0.314,0.265]	0.209	1.536	[-0.058,0.476]	0.039	0.275	[-0.239,0.318]
Whether as student cadres	0.006	0.099	[-0.111,0.123]	0.173	3.149**	[0.065,0.281]	0.058	1.011	[-0.055,0.171]
Supervisor–student relationship	-0.364	−13.587****	[-0.417,-0.312]	0.504	20.358***	[0.455,0.552]	-0.212	-7.123***	[-0.271,-0.154]
General self-efficacy	—	—	—	—	—	—	-0.302	-10.138***	[-0.36,-0.243]
R ²		0.147			0.274			0.214
F		26.471***			57.744***			36.907***

n = 1233, *p < 0.05, **p < 0 .01, ***p < 0 .001, ATSPPH Attitudes Toward Seeking Professional Psychological Help; Gender, types of students, types of degrees, region, marital status and whether as student cadres were dummy variables, with female = 0 and male = 1, master student = 0 and doctoral student = 1, academic degree = 0 and professional degree = 1, rural = 0 and urban = 1,unmarried = 0 and married = 1, student cadres = 0 and non student cadres = 1, these dummy variables coefficients were the unstandardized coefficients B, the same below

The total effect of supervisor–student relationship on anxiety was − 0.364, 95% CI = [− 0.417, − 0.312], the direct effect was − 0.212, 95% CI = [− 0.271, − 0.154], and the indirect (mediation) effect was − 0.152, 95% CI = [− 0.192, − 0.115]. None of these confidence intervals includes zero, confirming a significant partial mediation effect. The mediation effect accounted for 41.76% of the total effect. Therefore, Hypotheses 1 and 2 were supported.

Moderated mediation effect analysis

Hayes’ PROCESS macro Model 8 for SPSS was used to test the moderating effect of attitudes toward seeking professional psychological help. Results are presented in Table 4. In Model 1, the interaction between supervisor–student relationship and attitudes toward seeking professional psychological help was significant (β = −0.119, t = − 6.399, p < 0.001), indicating that attitudes toward seeking professional psychological help moderated the relationship between supervisor–student relationship and general self-efficacy, the interaction effect was 0.018. In addition, in model 2, the interaction between supervisor–student relationship and attitudes toward seeking professional psychological help was also significant (β = 0.106, t = 5.351, p < 0.001), indicating that attitudes toward seeking professional psychological help moderated the association between supervisor–student relationship and anxiety, the interaction effect was 0.031.

Table 4.

Results of moderated mediation effect analysis of attitudes toward seeking professional psychological help

Predictor	Model 1 (General self- efficacy)			Model 2 (Anxiety)
	β	t	95% CI	β	t	95% CI
Gender	0.189	3.616***	[0.086,0.291]	-0.21	-3.833***	[-0.317,-0.102]
Age	-0.007	-0.251	[-0.061,0.047]	0.02	0.69	[-0.037,0.077]
Types of students	-0.052	-0.4	[-0.306,0.202]	-0.081	-0.597	[-0.347,0.185]
Types of degrees	0.043	0.832	[-0.058,0.144]	-0.059	-1.086	[-0.164,0.047]
Region	0.009	0.176	[-0.086,0.103]	0.071	1.398	[-0.028,0.17]
Marital status	0.219	1.648	[-0.042,0.48]	0.017	0.124	[-0.256,0.29]
Whether as student cadres	0.142	2.64**	[0.037,0.248]	0.079	1.392	[-0.032,0.189]
Supervisor–student relationship	0.498	19.99***	[0.449,0.547]	-0.224	-7.462****	[-0.282,-0.165]
ATSPPH	0.117	4.686***	[0.068,0.166]	-0.136	-5.144***	[-0.187,-0.084]
Supervisor–student relationship × ATSPPH	-0.119	-6.399***	[-0.156,-0.083]	0.106	5.351***	[0.067,0.145]
General self- efficacy	—	—	—	-0.255	-8.548***	[-0.314,-0.197]
R ²		0.308			0.246
F		54.453***			36.129***

n = 1233, *p < 0.05, **p < 0 .01, ***p < 0 .001, ATSPPH Attitudes Toward Seeking Professional Psychological Help

To visualize the interaction pattern, simple slope figures (Figs. 2 and 3) were plotted for general self-efficacy against supervisor–student relationship and anxiety against supervisor–student relationship under negative (i.e., one SD below the mean) and positive (i.e., one SD above the mean) attitudes toward seeking professional psychological help, respectively.

Fig. 2.

Moderation by attitudes toward seeking professional psychological help between supervisor–student relationship and general self-efficacy. Note: SSR = Supervisor–Student Relationship; ATSPPH = Attitudes Toward Seeking Professional Psychological Help

Fig. 3.

Moderation by attitudes toward seeking professional psychological help between supervisor–student relationship and anxiety. Note: SSR = Supervisor–Student Relationship; ATSPPH = Attitudes Toward Seeking Professional Psychological Help

As depicted in Fig. 2, simple slope tests showed that supervisor–student relationship was positively associated with general self-efficacy for students with both negative (β_simple = 0.617, t = 18.866, p < 0.001) and positive (β_simple = 0.379, t = 12.874, p < 0.001) attitudes toward seeking professional psychological help, but the correlation was notably weaker in the latter. In other words, graduate students with positive attitudes toward seeking professional psychological help would experience a slower decline in general self-efficacy when facing a poor supervisor–student relationship than those with negative attitudes toward seeking professional psychological help.

As Fig. 3 shows, for graduate students with negative attitudes toward seeking professional psychological help, the relationship between supervisor–student relationship and anxiety was negative and statistically significant (β_simple = − 0.33, t = − 8.487, p < 0.001), but the strength of the association was weaker among graduate students with positive attitudes toward seeking professional psychological help (β_simple = − 0.118, t = − 3.597, p < 0.001). Simple slope tests demonstrated that the more positive the attitudes toward seeking professional psychological help, the weaker the association between supervisor–student relationship and anxiety.

The bias-corrected percentile bootstrap analysis further demonstrated that the indirect effect of supervisor–student relationship on anxiety via general self-efficacy was moderated by attitudes toward seeking professional psychological help. Particularly, for graduate students with negative attitudes toward seeking professional psychological help, the indirect effect of supervisor–student relationship on anxiety via general self-efficacy was significant (Indirect effect = − 0.158, Bootstrap 95% CI = [− 0.217, − 0.105]). The indirect effect was also significant for those with positive attitudes toward seeking professional psychological help (Indirect effect = − 0.097, Bootstrap 95% CI = [− 0.134, − 0.066]), but weaker. In sum, these results indicated that attitudes toward seeking professional psychological help moderated the mediation effect of supervisor–student relationship on anxiety through general self-efficacy. Hence, Hypotheses 3 and 4 were supported (see Fig. 4).

Fig. 4.

Test values for the moderated mediation model. Note: n = 1233; *p < 0.05, **p < 0.01, ***p < 0.001; ATSPPH = Attitudes Toward Seeking Professional Psychological Help

Discussion

The rising prevalence of anxiety among graduate students has attracted increasing attention in recent years, with poor supervisor–student relationship often regarded as an important source of psychological strain. Understanding the mechanisms through which relational quality affects anxiety is therefore essential for designing effective interventions. Against this backdrop, the present study investigated whether supervisor–student relationship was negatively associated with anxiety among Chinese graduate students, whether general self-efficacy mediated this association, and whether attitudes toward seeking professional psychological help moderated both the direct and indirect pathways. The results supported our hypotheses, a poorer supervisor–student relationship was linked to higher anxiety, general self-efficacy partially accounted for this association, and positive attitudes toward seeking professional psychological help attenuated the negative impact of poor relational quality. By testing this moderated mediation model, our findings contribute to a deeper understanding of the mechanisms through which relational quality influences graduate students’ mental health and highlight potential avenues for intervention.

The mediating effect of general self-efficacy

The mediation results of this study suggested that general self-efficacy was not only an outcome of supervisor–student relationship but also a partial mechanism linking this relationship to anxiety. For the first stage of the mediation process (i.e., supervisor–student relationship → general self-efficacy), our findings indicated that supportive relationships strengthened graduate students’ beliefs in their competence, whereas adverse relationships undermined them. This result is consistent with Bandura’s social cognitive theory [28], which emphasizes that mastery experiences, verbal persuasion, and social feedback from significant others shape efficacy beliefs. Supervisors, as primary academic figures, are in a position to provide these sources of efficacy, and their encouragement or criticism directly influences students’ self-appraisals. The present finding is also in line with earlier studies [13, 16] showing that positive mentoring relationships promote higher self-efficacy in student populations, while relational conflict erodes efficacy beliefs and impairs motivation.

For the second stage of the mediation process (i.e., general self-efficacy → anxiety), the results showed that lower levels of general self-efficacy were associated with higher levels of anxiety among graduate students. This is consistent with prior studies [29, 54] reporting that individuals with diminished efficacy beliefs are more likely to interpret stressors as overwhelming and to exhibit heightened worry and tension. According to the conservation of resources theory [32], the loss of a core personal resource such as self-efficacy increases vulnerability to stress outcomes, as students may feel less capable of mobilizing the psychological resources needed to meet academic demands. Previous empirical studies have similarly found that low self-efficacy predicts anxiety [55], depression [30], and other maladaptive outcomes [56], reinforcing the notion that efficacy beliefs are critical protective factors in student mental health.

It is also worth noting, however, that general self-efficacy only partially mediated the association between supervisor–student relationship and anxiety. Supervisor–student relationship remained a significant direct predictor of anxiety even after controlling for self-efficacy. This suggests that while general self-efficacy accounts for part of the mechanism, supervisor–student relationship may exert additional effects through other pathways, such as emotion regulation [57], perceived social support [58], or fairness perceptions in supervision [59]. Thus, each component of the mediation model is noteworthy, and the results highlight the multifaceted ways in which relational quality contributes to graduate students’ anxiety.

Beyond the indirect effects, the present study also confirmed a significant overall pathway between supervisor–student relationship and anxiety. A Poorer supervisor–student relationship was directly associated with higher anxiety, consistent with previous findings among elementary and secondary school students [17, 60]. This supports several theoretical perspectives. The diathesis-stress model [61] conceptualizes adverse supervisory relationships as salient stressors that heighten vulnerability to emotional disorders. Psychodynamic theory [25] likewise highlights that insecure or conflictual interpersonal bonds generate psychological distress, while the James-Lange theory [26] emphasizes that negative relational encounters can elicit physiological arousal that intensifies anxious experiences. From a biological perspective, strained supervisor–student relationship may also activate the hypothalamic–pituitary–adrenal (HPA) axis, leading to the release of stress-related hormones such as cortisol, thereby exacerbating anxiety [62]. Taken together, these perspectives underscore that poor supervisor–student relationship deprives graduate students of essential emotional and instrumental support, disrupt their sense of security, and directly contribute to heightened anxiety.

The moderating effect of attitudes toward seeking professional psychological help

Our findings indicated that attitudes toward seeking professional psychological help moderated both the indirect and direct pathways in the proposed model. Specifically, graduate students with positive attitudes toward seeking professional psychological help showed weaker adverse effects of poor supervisor–student relationship on both general self-efficacy and anxiety, whereas those with negative attitudes toward seeking professional psychological help were more strongly affected. This highlights that attitudes toward professional psychological help are not merely background characteristics but act as key contextual factors that condition how relational quality translates into psychological outcomes.

For the pathway linking supervisor–student relationship and general self-efficacy, graduate students with positive attitudes toward seeking professional psychological help experienced a slower decline in general self-efficacy even when confronted with negative supervisor–student relationship. By contrast, students with negative attitudes toward seeking professional psychological help were more likely to internalize supervisory conflict or neglect as personal shortcomings, leading to sharper reductions in their sense of competence. This result resonates with the stress-buffering model [38], which propose that adaptive coping resources can mitigate the psychological impact of stressors. Students who are open to seeking professional help may be more inclined to interpret relational difficulties as challenges that can be addressed through external support rather than as personal incompetence. Such openness provides an alternative efficacy-building channel, thereby dampening the negative influence of strained supervisory interactions [63]. Prior research has similarly shown that positive help-seeking attitudes enhance students’ resource utilization and protect their self-evaluations under academic stress [64], reinforcing the moderating role observed in this study.

For the pathway from supervisor–student relationship to anxiety, positive attitudes toward seeking professional help again served as a protective factor. Graduate students endorsing positive attitudes toward seeking professional help exhibited a weaker association between poor supervisor–student relationship and heightened anxiety, likely because they were more willing to reframe stressors and access external support [28, 38]. In contrast, those with negative attitudes toward seeking professional help lacked such coping strategies, intensifying their perception of relational strain as uncontrollable and directly increasing anxious symptoms. This aligns with social cognitive theory [28], which emphasizes the role of coping expectancies and cognitive appraisals in shaping emotional responses. Positive attitudes toward seeking professional help may foster more adaptive appraisals, reducing the tendency to perceive relational strain as uncontrollable or threatening. Empirical studies in college populations have likewise documented that students who are more open to professional counseling report lower levels of stress and anxiety when facing academic or interpersonal challenges, supporting the present findings [65, 66].

Importantly, the moderating effects observed in both pathways suggest that attitudes toward seeking professional psychological help function as a resilience factor in graduate education. Rather than uniformly experiencing the detrimental consequences of poor supervisor–student relationship, students who endorse positive attitudes appear more capable of buffering its negative impact, both by preserving general self-efficacy and by directly alleviating anxiety. These results contribute to the literature by demonstrating that openness to professional help is not only beneficial in itself but also shapes the relational-psychological mechanisms that underlie graduate students’ mental health.

Implications for practice

The current study carries several important implications. From a theoretical perspective, the findings contribute to a more comprehensive understanding of how relational quality influences the psychological well-being of graduate students. By identifying general self-efficacy as a partial mediator, the study provides empirical support for social cognitive theory and the conservation of resources theory, extending their applicability to the graduate education context. Moreover, the evidence that attitudes toward seeking professional psychological help moderate both the supervisor–student relationship–general self-efficacy link and the supervisor–student relationship–anxiety link adds nuance to existing models of stress and coping. Rather than functioning as a uniform buffer, attitudes toward professional help operate selectively across different stages of the stress process, thereby enriching theoretical accounts of resilience in higher education settings.

From a practical perspective, the results underscore the need for multifaceted strategies to alleviate the negative impact of poor supervisor–student relationship on graduate students’ mental health. At the institutional level, educational authorities may strengthen supervisory systems for graduate tutors, standardize training processes, and implement clearer guidelines to foster constructive supervisor–student relationship. Such measures have been shown to reduce adverse supervisory interactions and protect students’ psychological well-being [7, 67]. At the same time, universities could establish psychological early-warning systems, such as routine mental health surveys [68], to detect vulnerable students, particularly those reporting poor supervisor–student relationship and negative attitudes toward seeking professional psychological help, at an early stage and provide timely interventions. Furthermore, initiatives aimed at reducing stigma and improving attitudes toward professional psychological help are essential [69]. These may include comprehensive mental health education, diverse formats of counseling services (e.g., individual, group, online, or telephone counseling), and awareness-raising campaigns designed to normalize help-seeking behaviors [70]. Finally, graduate students themselves should be encouraged to develop adaptive strategies, including open communication with supervisors, cognitive reframing, exercise [28, 71], and mindfulness practices [72], as well as greater willingness to engage with professional counseling services. Collectively, these initiatives highlight the practical avenues through which institutions and individuals can mitigate the adverse effects of strained supervisor–student relationship and promote graduate students’ mental well-being.

Limitations and future directions

This study has several limitations that future research could seek to address.

Firstly, the cross-sectional design restricts the ability to make strong causal inferences regarding the relationships among supervisor–student relationship, general self-efficacy, attitudes toward seeking professional psychological help, and anxiety. Although the findings are consistent with theoretical models, longitudinal or experimental designs would be necessary to establish temporal precedence and clarify the dynamic interplay among these variables.

Secondly, the data were collected through self-report questionnaires, which may be subject to response biases such as social desirability or common method variance. Although established and widely validated scales were used, future studies could incorporate multi-informant assessments or objective indicators of mental health (e.g., physiological measures of stress reactivity) to strengthen validity.

Thirdly, the sample was drawn exclusively from Chinese graduate students, which may limit the generalizability of the findings. Given the unique characteristics of graduate education and the cultural context in China, it is unclear whether the observed mechanisms would apply equally to graduate students in other countries or to different educational systems. Future research could address this limitation by employing cross-cultural designs to examine the universality or cultural specificity of the moderated mediation model.

Finally, the study focused on general self-efficacy and attitudes toward seeking professional psychological help as key explanatory variables. While the model provides important insights, other psychological constructs such as resilience, perceived social support, or coping styles may also play significant roles in shaping the association between supervisor–student relationship and anxiety [73, 74]. Future studies could expand the model to include these additional factors, thereby providing a more comprehensive picture of the mechanisms linking supervisory relationships to graduate students’ mental health.

Conclusion

In conclusion, this study examined the association between supervisor–student relationship and anxiety among Chinese graduate students, highlighting the mediating role of general self-efficacy and the moderating role of attitudes toward seeking professional psychological help. The results showed that a poorer supervisor–student relationship was directly associated with higher anxiety, and that general self-efficacy partially accounted for this link. Furthermore, positive attitudes toward seeking professional psychological help buffered the detrimental effects of poor supervisor–student relationship, both by preserving general self-efficacy and by attenuating anxiety.

These findings extend existing theoretical models by demonstrating how interpersonal and individual factors jointly shape mental health outcomes in graduate education. They also underscore the importance of considering both relational dynamics and students’ attitudes toward professional psychological help in understanding vulnerability to anxiety. Taken together, the results suggest that fostering supportive supervisor–student relationship and promoting positive attitudes toward seeking professional psychological help are key avenues for safeguarding graduate students’ psychological well-being.

Authors’ contributions

LL and RW designed the study and drafted the manuscript. XD and SL collected and analyzed the data. ZL supervised and revised the manuscript. All authors contributed to the article and approved the submitted version.

Funding

This study was funded by the Science and Technology Research Project of Jiangxi Department of Education (GJJ2200929) and the Key Project of Guangzhou Psychological Society (2023GZPS05).

Data availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. The study was approved by the Ethics Committee of Jiangxi University of Chinese Medicine.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.