大學生性別組成的社會脈絡、性別與友誼網絡機制：ERGMs (Exponential Random Graph Models) 的探討

不同社會脈絡對於個人行為所造成的影響是社會學一直以來都十分關懷的議題。早期，Peter Blau (1974)已經強調過社會脈絡的差異是造成人們產生不同人際互動的原因。Ridgeway (2006)更強調了個體的行為表現之所以有所差異，是因為個體所處的環境不一致，所以社會對個體的期待也有所不同所致，其中，性別就是一項自社會學萌芽以來倍受討論的關鍵議題。性別自生命誕生的那一刻起，甚至還沒呱呱墜地，就深深的影響著個體往後所受到的期望與待遇，進而成為薪資、勞動力、教育階層化等領域所熱切探討的議題。不同性別比例組成的社會結構，其內部差異更是眾多社會學家、人口學家共相爭鳴的主要戰場。但在社會網絡的領域裡，受到資料蒐集較為困難等限制，至今尚未看到有人較為深入的去研究在大學脈絡下，不同性別組成的班級間人際互動的差異。本研究使用ERGMs為主要的分析方法，對熊瑞梅（2015）所主持的科技部整合型研究計畫：「性別與信任行為的機制：從社會網絡、信任賽局和神經科學的觀點來探討」的班級網絡資料進行分析，證實了不同性別組成的社會脈絡會使得內部不同性別的成員做出不一樣的交友行為；另外，本研究同時也發現了不同的行為認知在這之中扮演著舉足輕重的角色。

關鍵字

社會脈絡；性別組成；大學生；社會網絡； ERGMs

並列摘要

The impact of different social contexts on individual behavior is an important issue that sociology has always been concerned about. In the early time, Peter Blau (1974) had already emphasized that different social contexts are the causes of different interpersonal interactions. Ridgeway (2006) also emphasized that individual behavior varies with different expectations of the positions in society. Among one of them, Gender is a key issue which has been discussed since the beginning of sociology. Since the birth of a new life, even it hasn’t started to breath, gender has already affected human behavior deeply and determined human future expectation and performance. Salary, labor, and education…, etc. are the common outcome affected by gender. The differences between different social structures due to different gender composition, is also the main battlefield for many demographers and sociologists. Owing to the difficulty of collecting data on social networks in classes of college students, almost no study discusses the contexts of gender composition and mechanisms of social networks in the class under the university context exquisitely. This study selects ERGMs as the main analytical method and uses the network data that collected by Hsung’s (2015) project, “Gender and Trust Behavior: Exploration from the Perspectives of Social Networks, Trust Games and Neurosciences”, sponsored by MOST. This study verified that social contexts with different gender composition lead members inside different networks have the different interpersonal interaction. Moreover, we also find that different cognition of behavior also plays an important role in these relationships.