Degree Centrality
- Introduced by
- Freeman, L. C.
- Year
- 1978
- Type
- Node-level centrality measure
- Family
- Network science
What does it do?
Degree centrality is the simplest and most intuitive measure of a node's importance in a network, defined as the number of direct ties a node has to other nodes.
Normalized by dividing by the maximum possible ties, it allows comparison across networks of different sizes and is the starting point of almost every network analysis.
The intuition behind the method
Imagine a conference room where everyone who knows a lot of people is the most socially active. Degree centrality captures exactly that: a node's raw connection count. A node with high degree sits at the center of immediate activity — it can reach many neighbors in one step. Because it is simple to compute and easy to explain, degree centrality is often the first metric computed in a network study, serving as a baseline against which more sophisticated measures are compared.
History
Degree centrality was formally defined and systematized by Linton C. Freeman in his landmark 1978 paper in Social Networks, which also introduced betweenness and closeness centrality under a unified conceptual framework. The measure itself traces to earlier sociometric work in the 1940s–1950s. Freeman's paper remains the canonical citation and established social network analysis as a rigorous quantitative discipline.
When to use it
Use degree centrality when you need a fast, interpretable summary of which nodes have the most direct connections — for identifying hubs, key actors, or highly active nodes in any binary or weighted network. It suits small and large networks alike and requires only an edge list. Do not rely on degree centrality alone when the research question concerns indirect influence, information-flow efficiency, or gatekeeping roles; for those purposes, eigenvector centrality, closeness centrality, or betweenness centrality are more appropriate. It is also uninformative in complete or near-complete graphs where almost all nodes share the same degree.
Limitations and considerations
- Counts only direct neighbors; ignores position in the global network structure.
- A node with high degree may be entirely peripheral if all its neighbors are isolated from each other.
- Uninformative in dense or complete graphs where degree variance is near zero.
- Does not distinguish whether connections are to influential or peripheral nodes (use eigenvector centrality for that).
- Sensitive to network boundary decisions — who is included in the network affects scores.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Computationally trivial — O(n) per node — and scales to very large networks.
- Immediately interpretable: raw connection count requires no network-analysis expertise to explain.
- Applicable to directed, undirected, binary, and weighted (as weighted degree) graphs.
- Serves as an indispensable baseline for comparing more complex centrality measures.
- Degree distribution provides a fingerprint of overall network topology (scale-free, random, etc.).
- Easily visualized: node size proportional to degree conveys structure at a glance.
Applications
- Identifying hubs and influential actors in social and organizational networks.
- Epidemiological contact tracing to prioritize highly connected individuals for intervention.
- Web and citation network analysis to rank highly linked pages or papers.
- Infrastructure resilience analysis to find critical nodes whose removal fragments the network.
- Marketing and diffusion research to locate seed nodes for viral spread campaigns.
- Co-authorship and collaboration network studies in bibliometrics.
How it works
How it works
The network is represented as a graph G = (V, E), where V is the set of nodes and E is the set of edges. An adjacency matrix A encodes which node pairs are connected. In an undirected network each edge is counted once; in a directed network in-degree and out-degree are tracked separately, reflecting incoming and outgoing ties respectively.
For each node v, degree centrality is computed by summing all entries in its row (or column) of the adjacency matrix — that is, counting how many other nodes it is directly connected to. The raw degree is C_D(v) = deg(v). In a directed graph, in-degree counts incoming ties while out-degree counts outgoing ties.
Formula
C_D(v) = \deg(v) = \sum_{u \in V} a_{vu}To compare across networks of different sizes the raw degree is divided by (n − 1), the maximum possible number of ties in a network of n nodes. The normalized score C_D'(v) falls between 0 and 1, where 1 indicates a node connected to every other node in the network.
Formula
C_D'(v) = \frac{\deg(v)}{n - 1}Nodes are ranked by their normalized degree centrality score. The distribution of scores can be visualized as a histogram or plotted on the network graph as node size or color. High-degree nodes are candidates for hubs or influencers; the shape of the degree distribution (e.g., heavy-tailed vs. uniform) reveals broader structural properties of the network.
Common pitfalls
- Reporting raw degree without normalizing, making results incomparable across networks of different sizes.
- Treating degree centrality as sufficient evidence of influence without checking betweenness or eigenvector centrality.
- Ignoring directionality in directed networks: in-degree and out-degree carry different substantive meanings.
- Conflating degree centrality with weighted degree centrality without acknowledging the difference in binary vs. valued networks.
- Overlooking isolates (nodes with degree 0) that may be substantively important missing actors.
Frequently asked questions
What is the difference between degree centrality and eigenvector centrality?
Degree centrality counts the number of direct connections regardless of who those neighbors are. Eigenvector centrality weights each connection by the importance of the neighbor, so a node connected to a few highly central nodes can score higher than a node with many peripheral neighbors.
Should I use in-degree or out-degree in a directed network?
The choice depends on the research question. In-degree measures how many actors point to a node (e.g., popularity, prestige, citation count). Out-degree measures how many actors a node points to (e.g., activity level, information sending). Reporting both is often informative.
Does degree centrality work on weighted networks?
The standard formulation uses binary edges. For weighted networks, the analogous measure is weighted degree centrality (also called node strength), which sums edge weights rather than counting edges. Most implementations in igraph, NetworkX, and Gephi support this option.
How large must my network be?
Degree centrality has no minimum size requirement. It is meaningful even in networks of 10 nodes. However, in very small networks the normalized score loses resolution — almost all nodes may appear equally central. For networks with fewer than about 20 nodes, examining the raw degree distribution is usually more informative than the normalized score.
When is degree centrality not enough?
When your research question concerns broker roles, gatekeeping, or information flow across the network rather than local popularity. A node with modest degree can control access between clusters (high betweenness) or reach every other node quickly (high closeness) without having many direct ties.
Related methods
Sources
Freeman, L. C. (1978). Centrality in social networks: Conceptual clarification. Social Networks, 1(3), 215–239.
Open DOI referenceCopied.Wasserman, S. & Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge University Press.
ISBN: 978-0-521-38707-1Copied.