本研究旨在編製團體領導者工作同盟量表，作為領導者在團體過程中評估團體工作同盟之工具，及了解本量表與成員團體工作同盟、團體凝聚力、團體氣氛之相關。量表編製主要參考相關同盟文獻、Marziali、Munroe-Blum和McCleary（1997）、與研究者訓練團體領導者的實務經驗來編製團體領導者工作同盟量表，形成本量表架構內涵，並邀請國內五位輔導與諮商領域專家進行專家效度考驗，形力成預試量表。以團體領導者進行預試，共取得31個團體，179份領導者資料，以預試資料進行鑑別分析、項目分析及探索性因素分析，決定保留所有題項，形成36題之正式量表，接著以預試樣本進行內部一致性、建構效度和效標關聯效度，以作為評量工具信效度考驗。量表分成兩個因素向度：內容向度和系統向度。研究結果顯示，本量表α係數為.96，內容向度α係數值介於.89至.90之間；系統向度值介於.84至.88之間，具有良好的信度。在效標關聯效度方面，本量表與成員團體工作同盟量表相關為.12（p ＜ .01），團體凝聚力相關為.13（p ＜ .01），團體氣氛之分量表投入、逃避和衝突間則皆達顯著水準，分別為.07（p ＜ .05）、-.10（p ＜ .01）及-.18（p ＜ .01），顯示具有可接受的效度。本研究結果符合文獻研究結果，也符合理論假設，高工作同盟的團體相對團體凝聚力亦高，團體氣氛之投入與團體工作同盟有正向的相關，亦即領導者與成員的團體工作同盟也越高，成員評估投入團體的程度也越高，而成員所評估的逃避與衝突則與領導者與成員團體工作同盟有負向的相關，亦即團體領導者工作同盟分數則越低，成員呈現逃避與衝突的現象越多。最後，本研究針對量表應用，以及後續研究進行討論並提出相關建議。

The therapist-client relationship or the working alliance plays an important role not only in individual counseling but also in group counseling. Group counseling differs from individual counseling in many ways. Group dynamics is particularly attended to in the process of group counseling. The relations among the group members are rather complex because there are a variety of members with multiple relations among them in a counseling group. Besides, the role of the group leader in a counseling group is different from that of the therapist in individual counseling. The group leader creates an atmosphere of interaction in the group process that cultivates the member's self-exposure and lower the pattern of one-way interaction between the leader and the individual. The group leader would only intervene in the mutual interaction among the members in a group process, instead of responding to the individual. The multiple ways of interactions such as group members' observation, reflection, explanation and feedback require a situation different from working alliance in individual counseling. Based on previous studies, assessing the working alliance used in individual counseling indeed has its limit. Developing the working alliance scale specifically used for group counseling would allow us directly and effectively to assess the working alliance between the leader and group members. There is no group counseling working alliance scale up to now in Taiwan. The purpose of this study is to develop leader-rated group counseling working alliance scale (LGCWAS), which helps the group leaders as a tool to assess groups working alliance development, and help to understand the correlation between the leaders group working alliance scale and members of group working alliance, cohesion, and group climate. The development of this scale mainly refers to the relevant alliance literature, the practical experience of training the leadership skills and strategies for leaders in group counseling by the first author, and the group therapy alliance Scale (GTAS) used by the group members, which is developed by Marziali, Munroe-Blum and McCleary (1997). An original scale is formed by inviting five domestic experts in guidance and counseling to test expert validity on the scale structure and items. The group counseling leaders are tested using this scale. Through evaluating the participants' attitude toward the group therapy working alliance, there are 179 effective questionnaires in total from 31 different groups. We apply confirmatory factor analysis to test the internal consistency and validity of scale construction and its criterion-related validity as a reference for testing the reliability and validity of this assessment tool. After the item analysis and confirmatory factor analysis of the original scale data are carried out, we keep all the items and develop the formal scale with 36 items in total. These results reveal that the three categories in the Content Dimension are Goal, Task, and Bond; the four categories in the Interpersonal Dimension are Self-Therapist, Other-Therapist, Group-Therapist, and within system. The result reveals that the Cronbach's α is .96, Content Dimension Cronbach's α between .89 and .90; Interpersonal Dimension Cronbach's α between .84 and .88. It can be readily seen that the leaders group working alliance scale has good internal consistency. For the criterion-related validity, the correlation between leader-rated group counseling working alliance scale and member-rated group counseling working alliance scale is .12(p ＜ .01). The correlation between leaders group working alliance scale and cohesion is .13(p ＜ .01). The correlation between leaders group working alliance scale and engagement, avoidance, and conflict is .07, -.10, and -.18 respectively (p ＜ .01). These results indicate that the leaders group working alliance scale has good validity as well. This research is consistent with the results reported in the literature as well as the theoretical hypotheses. High working alliance group has high group cohesion. The involvement of group members has positive correlation with the group working alliance. In other words, the high working alliance between the leader and the group members would lead to high involvement in the group process assessed by the group members. On the contrary, the avoidance and conflict assessed by the group members have negative correlation with the working alliance between the leader and the group members. In other words, the low score in the leaders group working alliance scale would lead to more avoidance and conflict of group members. Based on our findings, we make some suggestions for the application of group counseling working alliance scale, the group leaders, professional training in group counseling and future research.