Network data containing attributes of both vertices and vertex pairs can be analysed together using multivariate entropies. This general multivariate technique is here illustrated with data previously analysed by Lazega and others. We focus here on the role of triad count statistics for exploring the dependency structure of the network attributes.