Week 6 - Centrality and Centralization

Dad Joke

Can you believe my wife told me to quit acting like a flamingo?

I had to put my foot down.

Housekeeping

  • Self Assessment Reflection Feedback is in
  • Posting homework questions on Discord:
```r

G |>
  activate(nodes) |>
  filter(age > 25)
```

Housekeeping

  • How did the AI assignment go?
    • What was it good at?
    • What was it bad at?
  • How do you think AI will be incorporated into teaching in the future?

Concept Review

  • What is centrality?
  • What is the difference between centrality and centralization?

Flash presentations

  • Choose one centrality measure
  • 4 minutes of prep
  • 1-2 minutes of presentation
  • What is it? How is it calculated? What does it tell us about power/influence?
  • What are its limitations?

Concept Review

  • What aspects of power/influence do centrality measures miss?
  • What can centralization tell us about a network?
  • What does density tell us about a network?
  • What is hierarchy?

Practical Review

What is the degree/closeness/betweenness centrality of node A? What is the density of the network?

Practical Review

What is the degree/closeness/betweenness centrality of node A?

G |>
  mutate(degree = centrality_degree(),
         closeness = centrality_closeness(),
         betweenness = centrality_betweenness()) |>
  filter(name == "A")
# A tbl_graph: 1 nodes and 0 edges
#
# An unrooted tree
#
# Node Data: 1 × 4 (active)
  name  degree closeness betweenness
  <chr>  <dbl>     <dbl>       <dbl>
1 A          2     0.125           0
#
# Edge Data: 0 × 2
# ℹ 2 variables: from <int>, to <int>

Discussion Questions

  • Could someone have high network / degree centrality but low actual power/influence?
    • Are there situations where degree centrality is negatively associated with power/influence?
  • Are follower counts less meaningful than they used to be, due to individualized algorithms?
  • What aspects of power do network measures miss?
  • Does network position cause influence, or vice versa?
  • Bonacich point about being connected to non-influential people can make you more influential

Discussion Questions

  • How do network structures reflect political systems?
  • How can we visualize centrality?
  • How do these approaches scale to huge networks?
  • How are centrality measures impacted by missing data?

R Review