Social network analysis is explicitly interested in the relationships among social actors. Focusing on structural variables, it opens up a field of data analysis and model building which is completely different from c...Social network analysis is explicitly interested in the relationships among social actors. Focusing on structural variables, it opens up a field of data analysis and model building which is completely different from conventional social statistical methods. Spanning nearly seventy years of research, statistical network analysis has witnessed three stages of models. Beginning from the late 1930s, the first generation of scholars (Moreno, Katz, Heider, etc.) studied the distribution of various network statistics. The second stage began from the 1970s and continued to the mid 1980s. It dealt primarily with exponential family of probability distributions for directed graphs (p 1 model) under the vital assumption of “dyad independence”. Relaxing this assumption, Frank and Strauss (1986), Strauss and Ikeda (1990), Wasserman and Pattison (1996) published their pathbreaking papers based on Markov’s random graphs models (p\+* model and its generalization: logit p\+*), which brought social network models to a new stage. It is an extremely flexible and complete model dealing with all sorts of structural aspects of social networks. This substantial “real” structural research should be employed to examine the relational essence of Chinese society.展开更多
文摘Social network analysis is explicitly interested in the relationships among social actors. Focusing on structural variables, it opens up a field of data analysis and model building which is completely different from conventional social statistical methods. Spanning nearly seventy years of research, statistical network analysis has witnessed three stages of models. Beginning from the late 1930s, the first generation of scholars (Moreno, Katz, Heider, etc.) studied the distribution of various network statistics. The second stage began from the 1970s and continued to the mid 1980s. It dealt primarily with exponential family of probability distributions for directed graphs (p 1 model) under the vital assumption of “dyad independence”. Relaxing this assumption, Frank and Strauss (1986), Strauss and Ikeda (1990), Wasserman and Pattison (1996) published their pathbreaking papers based on Markov’s random graphs models (p\+* model and its generalization: logit p\+*), which brought social network models to a new stage. It is an extremely flexible and complete model dealing with all sorts of structural aspects of social networks. This substantial “real” structural research should be employed to examine the relational essence of Chinese society.
