Estimation of Transitive Social Structures

 by Jukka Corander

 Research Report 1998:5

 Department of Statistics, Stockholm University, S-106 91 Stockholm, Sweden


An observation scheme where individuals report their perceptions of the cluster structure inherent in a social network or in a part of it is considered. A probability model generating such reported structures is introduced, where a mode parameter represents a latent transitive social structure. According to the model, the reported structures are spherically distributed around the mode, with exponentially decreasing probabilities as their distance from the mode increases. The rate of decrease in probabilities is determined by a concentration parameter. Inference concerning the mode and concentration parameter is made possible by applying simulation algorithms and a modified maximum likelihood approach. A resampling procedure is used in order to determine the effect of the sample size on the reliability of the estimated social structure.

 Key words: Classification, Latent Transitive Structure, Social Networks. 

Close this Window

Last update: 1998-05-19 / KH