We now begin the second portion of the book, which, as mentioned in Chapter 1, focuses on the statistical analysis of social network data. Most of the methods discussed in Chapters 13 through 16 (Parts V and VI) are based on stochastic assumptions about the relational data contained in a social network data set. There are a variety of such stochastic assumptions, and we will introduce and describe each in depth as they arise in the next four chapters.

The statistical methods that we will present in the next six chapters are organized into two parts (Parts V and VI) to separate earlier models for subgraphs from more recent models for entire graphs and digraphs. The statistical ideas, methods, and concepts presented in these chapters are quite diverse, and were developed over a period of forty years. We will begin with Part V – Dyadic and Triadic Methods for the analysis of social network data. Statistical analyses of network data can be quite important, and can nicely complement analyses based on methods described in the first portion of the book.

These methods are different from the structural analyses discussed earlier in the book, where a theory was translated into a set of graph theoretic statements about a network. These statements were studied in a descriptive or deterministic manner. Since the methods described in Chapters 5–12 were predominantly descriptive or even exploratory, we did not need distributional assumptions about particular structural properties.