研究生与导师关系恶化导致的研究生的自伤自杀恶性事件，引起了社会舆论的强烈关注。近些年研究生以“老板”称呼导师，自称为“打工人”是高校里常见的现象。研究生们协助导师完成其所主持的课题或者科研项目，在此过程中也兼顾自己学位论文的完成；在协助导师的同时，多数学生能定期获得一定的经济补贴，导学关系可被看作一种特殊的上下级关系。当下研究生的培养和管理中，导师掌握着对研究生而言至关重要的事件的决策权力。导师的权力日益集中，导学关系就越容易变得不对等，研究生与导师关系逐渐向利益关系转化，容易诱发导师负性指导行为——辱虐指导行为的出现。与导师的人际关系是研究生求学阶段关键的人际关系，导师也是研究生学术能力的主要评价者，当研究生面临辱虐指导时可能集中表现为人际受挫和能力受挫，进而降低研究生心理健康水平。

本研究以素质-压力模型和自杀的人际理论为基础，面向在读研究生展开系列研究旨在解决以下问题：（1）研究生感知到辱虐指导是否会影响其自杀意念？（2）辱虐指导对自杀意念的影响机制是怎样的？人际需要缺失在两者关系间是否存在中介作用？（3）从积极心理学角度看，研究生希望感水平能否缓解辱虐指导对自杀意念的影响？本研究从风险因素和保护性因素双重视角出发，验证辱虐指导对自杀意念的危害以及希望感的缓冲作用，以期能为研究生自杀预防和心理危机干预提供实证支持。本论文包含四个研究，分别是质性研究、横断面研究、日记法研究、干预研究，从不同角度充分探索研究生感知到的辱虐指导对自杀意念的影响及作用机制。

研究一采用质性研究探索辱虐指导对研究生心理健康的影响。研究通过对21名经历过辱虐指导的研究生（男生11人，M _age = 25.14 ± 2.46）的半结构化访谈，数据采用主题分析的方法。结果显示：（1）导学关系中辱虐指导表现形式包括贬低打压学生、高压工作要求、指导投入度低、侵害学生权益四种；（2）研究生对辱虐指导的反应包括情绪反应、认知反应、行为反应和生理反应，根据四类反应类别和强烈程度不同，可分为自我反思阶段、关系调整阶段和习得无助阶段三个阶段；（3）保护因素因为分为内在保护性因素和外在保护性因素两大类，而希望感是个体内部重要的保护因素。

研究二以1593名在读研究生（男生785人，M _age = 23.38 ± 2.32）为研究对象，以导师辱虐指导问卷、人际需要问卷、贝克自杀意念量表、自编希望感（含未来期待维度和目标应对维度）、以病人健康问卷（PHQ-9）最后一道题等问卷/题项作为测量工具，分别检验辱虐指导对自杀意念出现与否（二分变量）的预测作用，辱虐指导对自杀意念强度的预测作用，人际需要缺失（受挫归属感、自我累赘感）的中介效应以及希望感对人际需要的中介效应的调节效应。首先，逻辑回归与ROC曲线结果显示辱虐指导对研究生自杀意念的出现（二分变量）具有显著预测作用，当感知到的辱虐指导均值的高于1.950时，研究生出现自杀意念这一判定的准确率为76.2%。其次，有调节的中介模型的结果显示：（1）辱虐指导显著预测自杀意念；（2）受挫归属感、自我累赘感中介辱虐指导对自杀意念的影响；（3）希望感/未来期待维度/目标应对维度负向调节自我累赘感与自杀意念的关系，只有未来期待维度负向调节辱虐指导通过自我累赘感对自杀意念的中介效应。

研究三采用短时程、高密度采集数据的日记研究法，对105名研究生（男生58人，M _age = 23.42 ± 1.59）进行连续四周，每周三次的状态评估，即每名被试要求完成12次问卷作答，包括对日常状态下的辱虐指导、自杀意念、抑郁状态、受挫归属感、自我累赘感和希望感的水平的评估，以检验日常辱虐指对实时自杀意念的影响和作用机制。研究采用潜变量增长模型分析检测变量发展趋势间的关系，采用多水平线性模型检验自杀意念的近端风险因素、个体内水平辱虐指导与自杀意念的关系及其作用机制。首先，平行潜变量增长模型结果显示：辱虐指导截距因子正向预测自杀意念的截距因子，辱虐指导的斜率因子正向预测自杀意念斜率因子；潜变量增长中介模型显示受挫归属感的截距因子、自我累赘感的截距因子均能中介辱虐指导的截距因子到自杀意念的截距因子的影响，但仅发现自我累赘感的斜率因子中介辱虐指导的斜率因子到自杀意念斜率因子的影响。其次，基于多水平线性模型结果显示：（1）受挫归属感、自我累赘感、抑郁水平显著正向预测同一时间自杀意念，仅发现了自我累赘感显著正向预测下一时刻的自杀意念；（2）在个体间水平中，辱虐指导通过受挫归属感、自我累赘感的中介作用影响自杀意念；在个体内水平中，仅发现了自我累赘感的中介作用；（3）希望感/目标应对维度同时负向调节个体间水平、个体内水平辱虐指导对自杀意念的影响作用；而未来期待维度仅负向调节个体间水平辱虐指导对自杀意念的影响路径。

研究四采用随机对照实验法，对存在有自杀意念的研究生进行希望感写作干预，验证干预方案对提升希望感水平、降低自杀意念的效果。研究共招募了81名研究生（男生42人，M _age = 23.74 ± 2.24）参与研究，将其随机分配至干预组和对照组。结果显示：（1）写作干预对希望感的改变效果上，时间主效应显著、时间与组别的交互作用显著。即干预组、对照组的希望感水平均随时间呈现上升趋势，且干预组的提升更为显著；（2）写作干预对自杀意念的影响显示，时间主效应显著、时间与组别的交互作用显著。即干预组、对照组均随时间呈现下降趋势，且干预组自杀意念下降更为显著；（3）写作干预对辱虐指导与自杀意念的关系具有延后调节效应，在两个月追踪中发现写作干预负向调节自我累赘感与自杀意念的关系。

综合以上结果，本研究得出以下结论：

（1）导学关系中辱虐指导主要表现形式有贬低打压学生、高压工作要求、指导投入度低、侵害学生权益四种；研究生对辱虐指导的反应经历自我反思阶段、关系调整阶段和习得无助阶段三阶段；希望感是个体内部重要的保护因素。

（2）辱虐指导显著预测研究生自杀意念，在横断面研究、日记研究和干预研究中均得以验证。

（3）在自杀意念短时波动探索中，受挫归属感、自我累赘感具有共时性预测作用，自我累赘感具有前瞻性预测作用。

（4）受挫归属感、自我累赘感中介辱虐指导与自杀意念的关系，在横断面研究、日记研究个体间水平中，两种人际需要的中介效应同时存在，在个体内水平仅发现了自我累赘感的中介作用。

（5）希望感缓冲了辱虐指导对自杀意念的关系。在横断面研究中，希望感/未来期待维度/目标应对维度负向调节自我累赘感与自杀意念的关系，仅有未来期待维度调节了自我累赘感的中介效应。在日记研究的个体间水平希望感/未来期待维度/目标应对维度的负向调节作用，在个体内水平仅发现了希望感和目标应对维度的负向调节作用。

（6）希望感写作干预方案可以显著提升研究生希望感水平，显著降低自杀意念水平。

The malignant incidents of self-injury and suicide of graduate students caused by the deterioration of the relationship between graduate supervisors have aroused strong attention of the public opinion. In recent years, it is common for graduate students to call their supervisors "boss" and call themselves "workers" in colleges and universities. Graduate students assist their supervisors to complete their research projects, and also take the completion of their own dissertations into account in the process. In the return, most students can obtain certain stipend regularly for their assistance. The supervisior-student relationship can be regarded as a special relationship between superiors and subordinates. In the process of graduate education training, supervisors hold the decision-making power of the events that are crucial to the graduate students. The more the power of the supervisiors becomes concentrated, the easier the supervisor-student relationship becomes unequal. The relationship between graduate students and supervisors has gradually transformed into a benefit relationship, resulting in the emergence of negative guidance behavior of supervisors - abusive supervision in graduate training. The supervisor-student relationship is the key interpersonal relationship in the graduate stage, and the supervisor is also the main evaluator of the academic ability of the graduate students. When facing the abusive supervision, the main feelings of the students are interpersonal frustration and ability frustration, which will reduce the mental health level of the graduate students.

Based on the stress- diathesis model and interpersonal theory of suicide, this study conducted a series of studies targeting graduate students aimed at addressing the following questions: (1) Does the abusive supervision perceived by graduate students affect their suicidal ideation？(2) What is the influence mechanism of abusive supervision on suicidal ideation? Do the interpersonal needs (thwarted belongingness and perceived burdensomeness) mediate the relationship between perceived abusive supervision and suicidal ideation among graduate students? (3) From a positive psychological perspective, can the level of hope alleviate the relationship between abusive supervision and suicidal ideation? This study, from a dual perspective of risk factors and protective factors, examines the harm of abusive supervision on suicidal ideation and the buffering effect of hope, in order to provide empirical support for suicide prevention and crisis intervention for graduate students. This study includes four studies that fully explore the influence and mechanism of perceived abusive supervision on suicidal ideation among graduate students from different perspectives, including qualitative research, cross-sectional research, diary method research and intervention research.

In study 1 qualitative research was used to explore the impact of abusive supervision on the mental health of graduate students. Data was collected by semi-structured interviews among 21 graduate students (11 males, M _age = 25.14 ± 2.46) who had experienced abusive supervision, and analyzed by thematic analysis. The research results found that (1) in the supervisor-student relationship, there are four forms of abusive supervision: belittling and suppressing students, high-pressure work requirements, low investment of academic guidance and infringement of students' rights and interests; (2) The responses of graduate students to abusive guidance include emotional response, cognitive response, behavioral response, and physiological response. According to the different types and intensity of the four types of response, three different reaction stages, self-reflection stage, relationship regulation stage, and acquisition helplessness stage, are proposed; (3) Protective factors include two categories: internal protective factors and external protective factors. And hope is an important internal protective factor.

In study 2, 1593 graduate students (785 males, M _age = 23.38 ± 2.32) were selected as the subjects. The discriminant effect of abusive supervision on the occurrence of suicidal ideation (binary variable), the relationship between abusive supervision and suicidal ideation intensity, mediation role of interpersonal needs and the moderation role of hope were tested. First, the results of logistic regression and ROC showed that abusive supervision had a significant discriminative effect on the occurrence of suicidal ideation in graduate students (binary variable). When the mean score of abusive supervision scale was higher than 1.950, the accuracy of determining the occurrence of suicidal ideation was 76.2%. Second, the results of the moderated mediation model found that: (1) abusive supervision significantly predicted suicidal ideation; (2) The relationshiop between abusive supervision and suicidal ideation mediated by thwarted belongingness and perceived burdensomeness; (3) Hope, its sub-dimensions future expectation and goal coping negatively moderated the relationship between perceived burdensomeness and suicidal ideation, and only the sub-dimension of future expectation negatively moderated the mediating effect of abusive supervision on suicidal ideation through perceived burdensomeness.

In study 3, totle 105 graduate students (58 male students, M _age = 23.42 ± 1.59) were investigated three times a week for four consecutive weeks by using the diary method of short-term and high-density data collection, and the data including the level of daily abusive supervision, suicidal ideation, depressive state, thwarted belongingness and perceived burdensomeness and the hope were collected 12 times in total. In order to further examine the impact and mechanism of abusive supervision on daily suicidal ideation, the latent variable growth models were used to analyze the relationships among the development trends of the core variables, and multi-level linear models were used to test proximal risk factors of suicidal ideation, and the relationship between abusive supervision and suicidal ideation at the within level.

The parallel latent variable growth model results showed that the intercept factor of abusive supervision predicted the intercept factor of suicidal ideation, and the slope factor of abusive supervision significantly predicted the slope factor of suicidal ideation. The latent variable growth mediation models showed that both the intercept factor of thwarted belongingness and the intercept factor of perceived burdensomeness mediated the influence of the intercept factor of abusive supervision on the intercept factor of suicidal ideation, but only the slope factor of perceived burdensomeness mediated the relationship between the slope factor of abusive supervision and the slope factor of suicidal ideation. The results based on multi-level linear models showed that: (1) When exploring the risk factors for short-term fluctuation of suicidal ideation, it was found that thwarted belongingness, perceived burdensomeness and depression significantly predicted concurrent suicidal ideation, while only thwarted belongingness was found to significantly predict suicidal ideation at the next moment; (2) When constructing a multi-level 1-1-1 mediation model to test the relationship between the impact of daily abusive supervision and daily suicidal ideation. At the between-level, it was found that abusive supervision influenced suicidal ideation through mediating effects of thwarted belongingness and perceived burdensomeness, while at the within level, only the mediating effect of perceived burdensomeness was found; (3) A moderated 1-1-1 mediation models were used to test moderating roles of hope and its sub-dimensions. The hope and goal coping sub-dimension negatively moderated the relationship between abusive supervision and suicidal ideation at both between-level and within-level. The future expectation sub-dimension only negatively moderated the relationship between abusive supervision and suicidal ideation at within-level.

Study 4 used a randomized controlled quasi experimental research method to conduct writing intervention targeting on the improvement of hope among graduate students with suicidal ideation to verify the effectiveness of the intervention in improving hope and reducing suicidal ideation. A total of 81 graduate students (42 males, M _age = 23.74 ± 2.24) participated in the study and were randomly assigned to the intervention group and control group. The paticipatants in the intervention group completed seven writing tasks, and the core variable evaluations were conducted after each task and follow-up evaluations conducted at two weeks, one month, and two months after the intervention was completed. The control group received no intervention and just evaluated at the same time point as the intervention group.

The results showed that: (1) In the change of hope, both the main effect of time and the the interaction between time and intervention group were significant. The level of hope in both groups showed an improvement trend over time, but the improvement in the intervention group was more significant. (2) In the change of suicidal ideation, the main effect of time was significant, and the interaction between time and group was significant. That is, both groups showed a decreasing trend over time and the intensity of suicide ideation decreased significantly in the intervention group. (3) Writing intervention has a delayed moderating effect on the relationship between abusive supervision and suicidal ideation. In the two-month follow-up, it was found that writing intervention negatively moderated the relationship between perceived burdensomeness and suicidal ideation.

Based on the above research results, the following conclusions can be drawn from this study:

(1) In the supervisor-student relationship, there are four forms of abusive supervision: belittling and suppressing students, high-pressure work requirements, low investment of academic guidance and infringement of students' rights and interests. The reaction to abusive supervision among graduate students includes self-reflection stage, relationship regulation stage, and acquisition helplessness stage. Hope is an important internal protective factor.

(2) The abusive supervision significantly predicted the suicidal ideation, and the results were verified in cross-sectional studies, diary studies, and intervention studies.

(3) In the exploration of short-term fluctuations in suicidal ideation, thwarted belongingness and perceived burdensomeness had concurrent predictive effect, while only the perceived burdensomeness had a prospective predictive effect.

(4) The relationship between abusive supervision and suicidal ideation was mediated by thwarted belongingness and perceived burdensomeness. In cross-sectional study and at the between-level in diary study, the mediating effects of both interpersonal needs coexisted, but at within-level, only the mediating role of perceived burdensomeness was found.

(5) Hope cushioned the relationship between abusive supervision and suicidal ideation. In cross-sectional studies, the hope and its sub-dimensions negatively moderated the relationship between perceived burdensomeness and suicidal ideation, while only the sub-dimension of future expectation moderated the mediating effect of perceived burdensomeness. In the diary method studies, negative moderating effects of the hope, future expectation dimension and goal coping dimension were found at the between level. negative moderating effects of the hope, and goal coping dimension were found at within level.

(6) The writing intervention targeting on the hope significantly improved the level of hope among graduate students and significantly reduced the intensity of suicidal ideation.