Social network analysis
Connection: Search, collapse, robustness
This page is part of the materials supporting Sociology
157, an undergraduate introductory course on social network analysis. The course is taught
by Robert A. Hanneman of the Department of Sociology at the University of California, Riverside. Feel free to use and reproduce this textbook (with citation). For more information, or to offer comments,
you can send me e-mail.
Sources:
Watts, chapters 5 and 6
Demonstration/exercise
- Directed search for a person named "J"
- Directed search for a person by name (gender and ethnicity)
Milgram's results
- Completion rate in a geographic partition (Boston) and in a context
(investing) were moderate
- Completion rate in truly random samples were 18%
- People use identity and affiliation to make robust searches of large
networks using only local information
Kleinberg's mathematical result
- Rewiring of lattices to make small worlds
- Rewiring of length 2 (make a new connection to a friend of a friend) is
optimal
Search in Social networks
- Who am I? game
- The idea of "social distance" and "Blau space"
- Empirical results suggest that most use two-dimensional social space maps
- often geography and occupation
Discussion: Why is AIDs a more significant problem in South Africa
than in the U.S.?
- Laumann on American sexual networks; versus the South African case
- Why has intervention worked more effectively in the U.S.
- Early intervention before epidemic
- Closing "choke points" in the network
Watts' "percolation models"
- SIR model assumes a random network
- Random re-wiring of a lattice leads to epidemic threshold instead of
logistic
- Percolation: occupied nodes and open pathways generate
"correlation lengths"
- Percolating cluster: creates epidemic threshold; also assures robust
transmission
- Scale-free networks are more robust against random failure, but less
robust to failure proportional to load or centrality
top of this page
index of lecture outlines
course home page