本研究主要目的在探討有戀愛經驗大學生的「成人依附」、「社交能力」、「社會支持」、「寂寞」與「憂鬱」之間的關係。依此目的，本研究先探討不同背景變項的大學生在「逃避依附」、「焦慮依附」、「社交自我效能」、「困擾的自我揭露」、「社會支持」、「寂寞」與「憂鬱」上的差異情形，並以兩個模式探究有戀愛經驗大學生的「成人依附」、「社交能力」、「社會支持」、「寂寞」與「憂鬱」之間的關係。模式一探討「逃避依附、焦慮依附、社交自我效能、困擾的自我揭露、寂寞與憂鬱之關係」；模式二探討「逃避依附、焦慮依附、社交自我效能、困擾的自我揭露、社會支持與憂鬱之關係」。研究採問卷調查法，以台灣十一所大學805位有戀愛經驗的大學生為對象，並以隨機方式將之分為兩組，第一組399人，用來刪題與發展模式；第二組406人，用來驗證模式及探究不同背景變項的受試在各研究變項上的差異。研究工具包括成人依附量表、社交自我效能量表、困擾的自我揭露量表、寂寞量表、社會支持量表和憂鬱量表。資料分析方法為因素分析、信度分析、t考驗、單因子變異數分析及結構方程模式。 本研究以第一組樣本探討初始模式發現，初始模式與觀察資料間適配不理想，故根據修正指標進行修正，修正後的模式與資料適配後，再以第二組樣本驗證模式的穩定性，經驗證模式一及模式二具備相當穩定性，且模式一及模式二都與觀察資料適配，並能有效解釋變項間的關係。 主要研究結果如下： 第一，在背景變項方面：（一）女性受試的「困擾的自我揭露」與「社會支持」顯著高於男性；（二）一年級受試的「焦慮依附」顯著高於三年級及四年級的受試；（三）不在戀愛中受試的「逃避依附」、「焦慮依附」與「寂寞」顯著高於在戀愛中的受試，而在戀愛中受試的「社會支持」顯著高於不在戀愛中的受試；（四）無晤談經驗受試的「逃避依附」顯著高於有晤談經驗的受試，有晤談經驗受試的「寂寞」與「憂鬱」顯著高於無晤談經驗的受試。其他部分則無顯著差異。 第二，在模式方面：（一）逃避依附與焦慮依附會透過社交自我效能而間接影響寂寞，再間接影響憂鬱；（二）逃避依附與焦慮依附會透過社交自我效能而間接影響社會支持，再間接影響憂鬱；（三）逃避依附會透過困擾的自我揭露，間接影響社交自我效能，再透過社交自我效能間接影響寂寞，再間接影響憂鬱；（四）逃避依附會透過困擾的自我揭露，間接影響社交自我效能，再透過社交自我效能間接影響社會支持，再間接影響憂鬱；（五）焦慮依附會直接影響憂鬱；（六）焦慮依附會透過寂寞間接影響憂鬱；（七）焦慮依附對憂鬱的影響效果會大過逃避依附對憂鬱的影響效果；（八）困擾的自我揭露對寂寞不具直接影響力，困擾的自我揭露會透過社交自我效能間接影響寂寞；（九）逃避依附、焦慮依附、困擾的自我揭露對社會支持不具直接影響力，逃避依附、焦慮依附、困擾的自我揭露會透過社交自我效能間接影響社會支持。 最後，根據研究結果，針對個人、學校諮商與輔導實務及未來相關研究提出建議，以供參考。 / The Relationships Among Adult Attachment, Social Competencies, Social Support, Loneliness and Depression of College Students Ching-Ju Lee Abstract The main purpose of this research was to study the relationships among “adult attachment,” “social competencies,” “social support,” “loneliness” and “depression” of the college students who had romance. The researcher first investigated the differences in terms of “attachment avoidance,” “attachment anxiety,” “social self-efficacy,” “distress self-disclosure,” “social support,” “loneliness” and “depression” among the participants who had the different background variables. Also studied were the relationships among “adult attachment,” “social competencies,” “social support,” “loneliness” and “depression” of the college students who had romance by using two models. Model one was used to study “the relationships among attachment avoidance, attachment anxiety, social self-efficacy, distress self-disclosure, loneliness and depression.” Model two was used to study “the relationships among attachment avoidance, attachment anxiety, social self-efficacy, distress self-disclosure, social support and depression.” This study employed questionnaires to collect data. The participants of the study were 805 Taiwan college students who had romance from 11 universities, and were randomly divided into two groups. In group one, there were 399 participants whose data were used to cancel items and develop models, and in group two there were 406 participants whose data were used to test models and study the differences in regards of different research variables among the participants who had the different background variables. The participants were evaluated by Adult Attachment Scale, Social Self-efficacy Scale, Distress Self-disclosure Scale, Social Support Scale, Loneliness Scale and Depression Scale. The data were analyzed by factor analysis, Cronbach α analysis, t-test, one-way ANOVA and SEM. The initial models tested by group one did not fit well with the observed data. Therefore, applying the modification index, the researcher modified the models till the models fit the observed data, then tested the models’ stability by group two, and came to confirm the stability of model one and model two. The researcher found both model one and model two fit the observed data, and could effectively explain the relationships among the variables. The main results of this study were as follows: First, about the background variables: 1. The scores of “distress self-disclosure” and “social support” of girls were significantly higher than those of boys. 2. The scores of “attachment anxiety” of freshmen were significantly higher than those of juniors and seniors. 3. The scores of “attachment avoidance”, “attachment anxiety”, and “loneliness” of the students who were not in love were significantly higher than those of the students who were in love, and the scores of “social support” of the students who were in love were significantly higher than those of the students who were not in love. 4. The scores of “attachment avoidance” of the students who had no counseling experience were significantly higher than those of the students who had counseling experience while the scores of “loneliness” and “depression” of the students who had counseling experience were significantly higher than those of the students who had no counseling experience. There were no significant differences among other variables. Second, about the models: 1. Attachment avoidance and attachment anxiety could affect social self-efficacy, and could thus in turn affect loneliness through social self-efficacy, and could indirectly continue to affect depression through loneliness. 2. Attachment avoidance and attachment anxiety could affect social self-efficacy, and could thus in turn affect social support through social self-efficacy, and could indirectly continue to affect depression through social support. 3. Attachment avoidance could affect distress self-disclosure, and could thus in turn affect social self-efficacy through distress self-disclosure, loneliness through social self-efficacy, and depression through loneliness. 4. Attachment avoidance could affect distress self-disclosure, and could thus in turn affect social self-efficacy through distress self-disclosure, social support through social self-efficacy, and depression through social support. 5. Attachment anxiety could directly affect depression. 6. Attachment anxiety could indirectly affect depression through loneliness. 7. The effect of attachment anxiety affecting depression was higher than that of attachment avoidance affecting depression. 8. Distress self-disclosure could not directly affect loneliness, while it could indirectly affect loneliness through social self-efficacy. 9. Attachment avoidance, attachment anxiety and distress self-disclosure could not directly affect social support, whereas they could indirectly affect social support through social self-efficacy. Finally, based on the results of the study, the researcher made some further suggestions for individuals, school counseling and future researchers.

本研究試圖整合成人依附理論與Benjamin的SASB人際行為環複模式，主要研究目的有二，研究目的一：探討伴侶間的人際行為是否會影響個人在愛情關係中依附特質的展現，研究目的二：探討伴侶間的人際行為對個人在愛情關係中依附風格、依附特質改變的影響情形。在本研究中，在伴侶間人際行為的內涵包括親和性（又分成滋潤性與破壞性）與互依性（又分成自主性與控制性）等人際行為，伴侶間互動層次共測量個體自陳對方提供行為、實際對方提供行為，個體的尋求行為、以及提供-尋求行為之間的互補性（亦分成個體自陳伴侶間的互補性、以及實際伴侶間的互補性）。依附特質的測量包括個體自陳在愛情中與伴侶間的焦慮與逃避依附特質等，並可組合出安全型、焦慮/矛盾型、逃避型、與錯亂型等四種依附風格。 本研究採取「追蹤研究法」，平均間隔約12.49週，進行兩個時間點的測量，邀請正在約會關係中的大學生伴侶雙方共234人接受單獨的成人依附特質量表(ECR)、伴侶間人際行為量表(SASB模式)的測量，並同時詢問其壓力事件調查表與測驗大學生身心健康調查表(GHQ)，以作為排除影響依附風格改變的干擾變項，兩個時間點皆完成測驗且未分手者共計199人。研究目的一的分析對象為第一個時間受測者，但必須是正在談戀愛而未分手的大學生，共計232人符合標準，其中男性佔35％，女性佔65%，，而完整的配對資料共76對，故有152人可進行實際伴侶間互動的分析。 研究目的二的受測者除了兩個時間點皆受測且未分手外，還必須排除在兩個時間點遭受重大壓力者與GHQ量表上表現出身心健康狀況不佳者，共計有效受試者為183人，伴侶雙方皆完成測驗的共計63對，故有126人可進行實際伴侶間互動的分析。本研究亦針對個體自陳對方提供行為、以及各人際行為的親和性與互依性指標、提供與尋求行為的互補性指標等，進行信效度的檢驗。 本研究結果可分成兩大部分。第一，在人際行為對愛情依附特質與風格改變的影響方面，可發現部分伴侶間人際行為的確可對依附特質造成影響：1. 「親和性」人際行為最能影響愛情依附特質，並與依附風格、依附特質改變最有關連；2.「滋潤性親和」不同於「破壞性親和」對依附特質的影響；3.個體在愛情關係中的依附特質，主要受到個體「主觀認知」到的伴侶間人際互動，而並不一定會受到「實際」伴侶間人際互動的影響；4. 逃避特質則較明顯受到伴侶間滋潤性親和人際行為的影響，而焦慮依附特質較受到伴侶間破壞性人際行為的影響，但焦慮依附特質的改變與伴侶人際行為關連性較低。第二，對個體在關係中單向的人際行為而言，親和性比互依性人際行為更能影響愛情依附特質，甚至與依附風格、依附特質的改變更有關連；但互依性指標（自主性）對依附特質並非沒有影響力，而是會顯現在雙向人際行為的互補性當中。 最後乃說明本研究的討論與限制，並提出未來在研究上與伴侶諮商實務上的建議。 / This study tried to integrate adult attachment theory and Benjamin’s Structural Analysis of Social Behavior Models (SASB). There were two major purposes of the study: The first one was to, discussing whether the impact of dating couples’ interpersonal behaviors on romantic attachment characteristics for college students. The Second was to, discussing the impact of dating couples’ interpersonal behaviors on the change of attachment styles and characteristics. In this study, the content of dating couples’ interpersonal behaviors includes affiliation (best and worst) and interdependence (autonomy and control) behaviors. The interactions of dating couples were measured by self reporting partner’s providing behaviors, actual partner’s providing behaviors, subjects’ seeking behaviors, the complements of self reporting dating relationship with providing-seeking behaviors, and the complements of actual dating relationship with providing-seeking behaviors. The measurement of attachment characteristics included anxiety and avoidance trait, which could be combined to four attachment styles: secure, anxious-ambivalent, avoidant, disoriented / disorganized. The study took ‘panel study’, and the average interval of test-retest time points was about 12.49 weeks. Total subjects were 234, who were measured by ECR and couples’ interpersonal behaviors (SASB model). Questionnaires of stress events, and GHQ were interference variables for the use of the purpose of the second study. There were 232 subjects composed of 35% male and 65% female in the purpose of first study. These subjects who were all measured in the first timing were falling in love and not yet breaking-up. There were 76 pairs complete matching data, so 152 subjects could process the analysis of actual dating relationship. There were 183 subjects in the purpose of the second study. Besides subjects who were measured in the two time points were falling in love and not yet breaking-up, it had to eliminate two interference variables. There were 63 pairs complete matching data, so 126 subjects could process the analysis of actual dating relationship. There were two major results in this study. First, couples’ interpersonal behaviors had the impact on attachment characteristics: 1. affiliation interpersonal behaviors could affect romantic attachment characteristics, and was related to the change of attachment styles and attachment characteristics. 2. The impact on attachment characteristics was different between ‘best affiliation’ and ‘worst affiliation’. 3. Self reporting couples’ interpersonal behaviors had impact on attachment characteristics. On the contrary, actual couples’ interpersonal behaviors had no impact. 4. ‘Best affiliation’ had impact on avoidance attachment characteristics, and ‘worst affiliation’ had impact on anxiety attachment characteristics. The association between the change of anxiety attachment characteristics and couples’ interpersonal behaviors was low. Second, for one-way interpersonal behavior in dating relationship, affiliation behaviors affected romantic attachment characteristics more than interdependence. Even affiliation behaviors were more related to the change of attachment styles and attachment characteristics than interdependence. The complement of couples’ autonomy interpersonal behaviors also affected attachment characteristics. In addition, implication for adult attachment research and couple therapy in dating relationship was discussed. Research limitation was also explicated.