Simple methods for simulating sociomatrices with given marginal totals

Abstract

Abstract For many purposes it is useful to simulate sociomatrices from the distribution in which all sociomatrices with the given marginal totals are equally likely. Procedures for this exist in the literature, including an approach based on Markov chain Monte Carlo (MCMC) techniques. This paper discusses this approach (with a small modification of existing methods), and shows how to apply it to distributions in which all sociomatrices with other given totals (as well as the row and column totals) are equally likely. The methods are applied to some examples, and substantive interpretations of results are discussed.