Network analysis glossary
Clustering coefficient
The clustering coefficient measures how many of a person's contacts are also tied to each other. A value of 1 means all of their contacts know one another; 0 means none do. Averaged or totalled across the network, it shows how cliquish the group is.
Updated · Netgraf
| Person | Local clustering |
|---|---|
| Carol | 1.00 |
| Ed | 1.00 |
| Andre | 0.67 |
| Beverly | 0.67 |
| Diane | 0.53 |
Global clustering (transitivity) for the whole map: 0.579. Computed live from the example map with the same code the Netgraf insights panel runs.
How it's calculated
local clustering(v) = ties among v's neighbours / (k (k - 1) / 2) transitivity = 3 * triangles / connected triples
What it tells you
High local clustering means a person sits inside a tight group where everyone knows everyone: good for trust and support, weak for new information. Low clustering with high degree is the signature of a broker. Watts and Strogatz showed in 1998 that most real networks combine high clustering with short paths, the "small world" pattern.
In Krackhardt's kite, Carol and Ed have the maximum local clustering of 1: all of their contacts are tied to each other. Ike, the only link between Jane and everyone else, has 0: his two contacts have never met.
In Netgraf
The Insights panel's How dense card reports the global clustering coefficient (transitivity) alongside density.