13  Fitting ERGMs to Faux Mesa High with statnetWeb (Lab)”

13.1 statnetWeb Introduction

statnetWeb is a web-based interface that provides access to cross-sectional network modeling with the statnet packages network, sna and ergm. Today we are using this app for the morning sessions on ERGMs.

13.2 Getting Started

Open Rstudio and load statnetWeb:

library(statnetWeb)

Launch the Shiny app:

run_sw()

13.3 The Lab Assignment

You are going to execute a typical ERGM workflow: fit and assess three models sequentially, using the faux.mesa.high built in network data.

13.3.1 Start with the edges only model

  • on the Data Tab: select the faux.mesa.high network data

  • on the Fit Model tab:

    • Add the edges term
    • Fit the model
    • Save the model
    • Interpret the estimated coefficient: calculate the density of the network as a function of the coefficient.

  • on the MCMC Diagnostics tab

    • can you run the dx for this model?
    • why or why not?

  • on the Goodness of Fit tab:

    • run the default goodness of fit dx
    • interpret the results

13.3.2 Then: repeat these steps, for the following models:

  • Add Terms: nodefactor("Grade") + nodefactor("Race") + nodefactor("Sex")
  • Add Terms: triangle

13.3.3 Questions

  • For the model with the nodefactor terms added
    • interpret each of the coefficients: Significance? Direction? Size? Compared to the other terms?
    • compare the goodness of fit to the edges only model: which higher order stats look better? which do not?

  • For the model with the triangle term added
    • use your breakout group Slack channel to report your findings
    • discuss