Much, if not most, social network data is derived from informant self-reports or participant observation; past research, however, has indicated that such reports are in fact highly inaccurate representations of social interaction. Although this problem has been known for over two decades, little consensus has developed concerning what might be done about it. Here, I demonstrate a simple family of hierarchical Bayesian models which allow for the simultaneous inference of informant accuracy and social structure in the presence of measurement error. Thoughts regarding the potential for further development in this area -- and some ruminations on important problems yet to be solved -- will also be discussed.
Joint work with Greg Ridgeway, RAND Corporation
