Beyond Triangles: The Importance Of Diamonds In Networks Katherine Stovel Christine Fountain Yen-Sheng Chiang University of Washington
Roadmap The Problem Measuring 4-cycles Generating Models Empirical Examples
The Problem Many observed social networks are distinctive because of their high degree of local clustering Local clustering is often explained as a by- product of tendencies toward balance However, local clustering may not always be the result of transitivity The Problem
i Triadic measures of clustering 3( ) C3 = The Problem Watts 1999, Dorogovtsev 2004, etc. C v = density of subgraph X containing i’s neighbors C = ∑C v /n { Require that k>=2
Triadic measures fail to capture clustering in the presence of local prohibitions The Problem
Heterosexual Nets Minimal Structure The Problem
Chains of Affection Bearman, Moody, Stovel Male Female 2 Empirical Analyses
Producer 2 Competitive or Stratified Worlds Producer 1 supplier consumer The Problem
Solution: Consider the relative frequency of diamonds Diamonds capture simultaneous preference for nearness and local prohibitions Classic Bi-partite graphs… The Problem
Measuring Diamonds Bernoulli expectation Census of observed diamonds Variants for directed graphs time-ordered data Measuring 4-cycles
Bernoulli expectation e+00 1 e+06 2 e+06 3 e+06 4 e+06 Network Density (p) Diamonds Expected Observed in simulated data N = 200; 5 nets per simulated point Measuring 4-cycles Undirected graphs
Bernoulli Expectation: Directed graphs Measuring 4-cycles
i l jk i l jk i l jk i l jk Symmetric Hierarchy (HI)Unique Hierarchy (HII) CycleIncoherence Measuring 4-cycles
Diamond Census Count number of complete, partial, and empty diamonds in network Variants for more complex graphs Directed graphs Time-ordered graphs Coded in both R and Matlab Measuring 4-cycles
Two Generating Models Attribute Sort Model (θ) Variable strength prohibition against in-group ties Basic assortative-disassortative mixing model Burt Model (Ω) Actors build networks that are rich in structural holes Modification of Watts’ α model Generating Models
Attribute Sort Model Generating Models θ controls strength of mixing θ = 0 in-group ties prohibited θ =.5 no preference for in- or out-group ties θ = 1 out-group ties prohibited N nodes Mean degree = k Create matrix R that indexes similarity of nodes i and j If R ij = 1, If R ij = 0,
Generating Models
The Burt Model Generating Models Ω controls strength of preference for structural holes 0 ≤ Ω ≥ ∞ No preference ↔ Strong preference X is current tie matrix M indexes shared alters p = small tie probability k = mean degree if k > Mij, if Mij ≥ k
Generating Models
Diamonds and Triangles in Omega Graphs Generating Models
The Upshot: Both generating models create far more diamonds than in comparable random graphs In the absence of any preference for social closeness, effects are somewhat density dependent Though density is an artificial means of imposing a closeness constraint Generating Models
Data Analysis Strategic alliances Academic Citation Patterns Empirical Analyses
Strategic Alliances Biotechnology, Justin Baer 2002 Empirical Analyses
Academic Citation Patterns Empirical Analyses Lowell Hargens 2000
Prevalence of s NpC3Exp (each type) Symmetric Hierarchy Unique Hierarchy Incoherent Celestial Masers Toni Morrison Criticism Empirical Analyses
Take Away Message Transitivity is obviously not the only systematic form of local structuring Local out-group preferences or strategic behavior may preclude triadic closure in real social networks Combined with propinquity or a preference for nearness, these prohibitions may create diamond-like clustered local structures Observing diamonds may be an indication of a normative prohibition against specific relations The End
Expected number of m-link cycles (total) in Bernoulli random graph Expected number of m-link chains in Bernoulli random graph
Diamond Census Measuring 4-cycles i jk l i jk l i jk l i jk l Empty Partial Complete