library(ergm)
library(sna)
library(coda)

Rural Arizona risk Networks

The Data

The data was aggregated by Martina Morris (University of Washington) and Richard Rothenberg (Emory University) and put online at ICPSR. The original data can be found here. In this exercise we are goint to investigate four networks derived from the Rural Arizona risk networks in Flagstaff, AZ. These networks were collected from May 1996 to Jan 1998 and originally had 95 respondents interviewed 5 times each. All participants are over 18 years old. This network was generated from random walk sampling strategy. Warning These networks are obviously from network sample and not true complete networks, but we are going to ignore that for this exercise and will treat them as complete.

  • Name generator
    • Sex, needle, other (illicit) drug contact, social contact in last 6 months
    • Sampling strategy
      • Six seeds chosen at random within same geographic area (Flagstaff) from persons presumed to be at elevated risk for HIV acquisition (through sex and/or drug behaviors)

Networks

Here, I have constructed 4 networks with attributes: race, ethnicity, sex, and age.

Load the data:

load("data/flagstaff_rural.rda")

Needle Network

flag_needle_net
 Network attributes:
  vertices = 42 
  directed = FALSE 
  hyper = FALSE 
  loops = FALSE 
  multiple = FALSE 
  bipartite = FALSE 
  total edges= 87 
    missing edges= 0 
    non-missing edges= 87 

 Vertex attribute names: 
    age ethn id race sex vertex.names 

No edge attributes
plot(flag_needle_net, vertex.col = "race", main = "Needle")

Sexual Contact Network

flag_sex_net
 Network attributes:
  vertices = 167 
  directed = FALSE 
  hyper = FALSE 
  loops = FALSE 
  multiple = FALSE 
  bipartite = FALSE 
  total edges= 282 
    missing edges= 0 
    non-missing edges= 282 

 Vertex attribute names: 
    age ethn id race sex vertex.names 

No edge attributes
plot(flag_sex_net, vertex.col = "race", main = "Sex")

Social Contact Network

flag_social_net
 Network attributes:
  vertices = 523 
  directed = FALSE 
  hyper = FALSE 
  loops = FALSE 
  multiple = FALSE 
  bipartite = FALSE 
  total edges= 2713 
    missing edges= 0 
    non-missing edges= 2713 

 Vertex attribute names: 
    age ethn id race sex vertex.names 

 Edge attribute names not shown 
plot(flag_social_net, vertex.col = "race", main = "Social")

Drug Contact Network

flag_drug_net
 Network attributes:
  vertices = 352 
  directed = FALSE 
  hyper = FALSE 
  loops = FALSE 
  multiple = FALSE 
  bipartite = FALSE 
  total edges= 1626 
    missing edges= 0 
    non-missing edges= 1626 

 Vertex attribute names: 
    age ethn id race sex vertex.names 

 Edge attribute names not shown 
plot(flag_drug_net, vertex.col = "race", main = "Drug")

Exercise

For each network explore the basic properties of the network (e.g., mixing matrix, density, etc.) and then attempt to develop a “good” ERG model for each network. After you have the best ERG model you can construct, use the tutorial to try to interpret the predicted probabilities and odds ratio of a given conditional tie. Further, attempt to explain what these results would mean substantively.