Network analysis glossary
Network centralization
Network centralization measures how much a whole network is organised around a single person. It compares the most central person with everyone else: a star, where one hub touches everyone, scores 1, and a network where everyone has the same number of ties scores 0.
Updated · Netgraf
| Measure | Value |
|---|---|
| Degree centralization | 0.400 |
| Reads as | moderately hub-driven |
| Top tenth hold | 35% of tie ends |
Freeman degree centralization: 0 when everyone has the same number of ties, 1 for a star. Computed live from the example map with the same code the Netgraf insights panel runs.
How it's calculated
centralization = sum over v of (max degree - degree(v)) / ((n - 1)(n - 2))
Linton Freeman defined centralization in 1978 as the network-level partner of each centrality measure: there is a betweenness centralization and a closeness centralization too. Degree centralization is the one most often reported.
Reading it
Netgraf reads the number out rather than leaving you with a decimal: below 0.2 is spread fairly evenly, up to 0.45 moderately hub-driven, up to 0.7 strongly hub-driven, and above that almost a single hub. A highly centralised team is fast when the hub is available and stuck when they are not.
In Netgraf
The Who carries it card in the Insights panel gives the centralization and the share of all ties held by the busiest tenth of people, and the one-line summary at the top of the panel opens with it.