Social network analysis (SNA) has emerged as one of the most powerful methodological frameworks in the social sciences, enabling researchers to map, measure, and analyse the relationships and flows between people, groups, organisations, and other entities. By focusing on the structures of interaction rather than the attributes of individuals, SNA reveals patterns that are invisible to conventional analytical approaches. This comprehensive guide explores the methods, software tools, and applications of social network analysis in contemporary research.
What Is Social Network Analysis?
Social network analysis is a systematic approach to investigating social structures through the use of networks and graph theory. It characterises networked structures in terms of nodes (individual actors, people, or entities within the network) and the ties, edges, or links (relationships or interactions) that connect them. SNA provides both a visual and a mathematical analysis of human relationships, offering insights into how information flows, how influence spreads, and how communities form and evolve.
The origins of SNA can be traced to sociometry in the 1930s, when Jacob Moreno first began mapping interpersonal relationships using sociograms. Since then, the field has grown enormously, drawing on contributions from sociology, anthropology, mathematics, physics, and computer science. For a deeper exploration of foundational approaches, see our overview of research methods in social network analysis.
Key Concepts and Metrics in Social Network Analysis
Understanding SNA requires familiarity with several fundamental concepts and metrics that describe the properties of networks and the positions of actors within them.
Nodes and Ties
Nodes represent the actors in a network — these may be individuals, organisations, countries, websites, or any other entity of interest. Ties represent the relationships between nodes. Ties can be directed (flowing from one node to another, such as an email sent) or undirected (reciprocal, such as a friendship). They can also be weighted to reflect the strength or frequency of interaction.
Centrality Measures
Centrality is one of the most important concepts in SNA, as it identifies the most important or influential nodes within a network. Several centrality measures exist, each capturing a different aspect of importance:
- Degree centrality counts the number of direct connections a node has. Nodes with high degree centrality are well-connected and often serve as hubs.
- Betweenness centrality measures the extent to which a node lies on the shortest paths between other nodes. High betweenness centrality indicates a brokerage role — the node controls the flow of information between different parts of the network.
- Closeness centrality assesses how close a node is to all other nodes in the network. Nodes with high closeness centrality can reach all other nodes quickly and efficiently.
- Eigenvector centrality considers not only a node’s direct connections but also the importance of those connections. A node connected to other highly connected nodes receives a higher eigenvector centrality score.
Network Density
Density measures the proportion of actual ties in a network relative to the total number of possible ties. A highly dense network is one where most nodes are connected to most other nodes, while a sparse network has relatively few connections. Density affects information flow, social cohesion, and the speed at which innovations or behaviours spread through a network.
Clustering and Community Detection
Clustering refers to the tendency of nodes to form tightly connected groups or communities within a larger network. The clustering coefficient measures the degree to which nodes tend to cluster together. Community detection algorithms identify subgroups within networks where internal connections are denser than connections to the rest of the network.
Structural Holes
Ronald Burt’s concept of structural holes refers to gaps in a network where two clusters are not directly connected. Actors who bridge structural holes gain advantages through access to diverse information and the ability to broker connections between otherwise disconnected groups.
Social Network Analysis Software
The analysis of social networks requires specialised software capable of handling large datasets, computing network metrics, and producing visual representations of network structures. A detailed comparison of available tools can be found in our review of social network analysis software.
UCINET
UCINET (University of California at Irvine NETwork) is one of the most widely used software packages for the analysis of social network data. Developed by Steve Borgatti, Martin Everett, and Lin Freeman, UCINET provides a comprehensive suite of network analysis tools including centrality measures, subgroup identification, role analysis, and statistical testing. It includes NetDraw, a built-in visualisation tool for creating network diagrams. UCINET is particularly popular in academic research due to its extensive analytical capabilities and its ability to handle multiple types of network data.
Gephi
Gephi is an open-source network analysis and visualisation platform that has become increasingly popular among researchers and practitioners. It offers powerful visualisation capabilities, with support for large networks containing hundreds of thousands of nodes. Gephi includes various layout algorithms, filtering tools, and statistical functions. Its plugin architecture allows users to extend its functionality for specific research needs.
NodeXL
NodeXL is a free, open-source network analysis add-in for Microsoft Excel. It provides an accessible entry point for researchers who are familiar with spreadsheet software but new to network analysis. NodeXL supports data import from various social media platforms, making it particularly useful for analysing online social networks. It includes basic network metrics, visualisation tools, and community detection algorithms.
Pajek
Pajek is a free software package for the analysis and visualisation of large networks. It is designed specifically to handle very large networks efficiently, supporting networks with millions of nodes. Pajek provides tools for network partitioning, hierarchical decomposition, and the analysis of temporal networks. It is widely used in bibliometric analysis, genealogical research, and the study of large-scale social systems.
R Packages for Network Analysis
The R statistical programming environment offers several powerful packages for network analysis. The igraph package provides a comprehensive library of network analysis functions, including centrality measures, community detection, and network visualisation. The statnet suite of packages supports statistical modelling of network data, including exponential random graph models (ERGMs) and latent space models. The sna package provides tools for sociometric analysis and network visualisation.
Python Libraries
Python has emerged as a major platform for network analysis, with libraries such as NetworkX providing extensive tools for the creation, manipulation, and study of complex networks. NetworkX supports a wide range of graph types and includes algorithms for centrality, clustering, shortest paths, and community detection. For large-scale network analysis, the graph-tool library offers high-performance implementations of network algorithms using C++ with a Python interface.
Methods and Approaches in Social Network Analysis
Data Collection Methods
Collecting network data requires different approaches depending on the research context. Survey-based methods ask respondents to identify their contacts, collaborators, or interaction partners. Observational methods record interactions as they occur in natural settings. Archival methods extract network data from existing records such as email logs, publication databases, or organisational charts. Digital trace data from social media platforms, online forums, and communication systems provide increasingly rich sources of network data.
Ego Network Analysis
Ego network analysis focuses on the network surrounding a single focal actor (the ego). This approach examines the ego’s direct contacts (alters) and the relationships among those contacts. Ego network analysis is particularly useful when it is impractical to survey an entire network, as data collection focuses on a manageable subset of actors.
Whole Network Analysis
Whole network analysis examines the complete structure of relationships within a bounded network. This approach requires identifying the network boundary and collecting data on all actors and relationships within that boundary. Whole network analysis is essential for studying network-level properties such as density, centralisation, and community structure.
Two-Mode Networks
Two-mode (or bipartite) networks involve two different types of nodes — for example, individuals and organisations, or authors and publications. Two-mode network analysis examines the relationships between these different types of actors, providing insights into patterns of affiliation, collaboration, and participation.
Dynamic Network Analysis
Dynamic network analysis examines how networks change over time. This approach tracks the formation and dissolution of ties, the entry and exit of actors, and changes in network structure. Dynamic analysis reveals processes of network evolution, including preferential attachment, triadic closure, and the emergence of structural patterns.
Applications of Social Network Analysis
Public Health
SNA has been widely applied in public health to study the spread of infectious diseases, the diffusion of health behaviours, and the effectiveness of intervention programmes. Network-based approaches have informed strategies for HIV prevention, smoking cessation, and vaccination campaigns. During the COVID-19 pandemic, SNA techniques were used to model transmission chains and evaluate the effectiveness of contact tracing programmes.
Organisational Studies
In organisational contexts, SNA reveals informal communication patterns, knowledge-sharing networks, and power structures that may differ significantly from formal organisational hierarchies. Organisations use SNA to identify key knowledge brokers, improve collaboration across departments, and support change management initiatives.
Crime and Security
Law enforcement and intelligence agencies use SNA to map criminal networks, identify key players, and disrupt illegal operations. Network analysis has been applied to the study of terrorist networks, drug trafficking organisations, and money laundering schemes. By identifying network vulnerabilities, analysts can develop targeted intervention strategies.
Education and Learning
SNA is used in educational research to study peer learning networks, teacher collaboration, and knowledge sharing in academic communities. Learning analytics platforms increasingly incorporate network analysis to understand student interaction patterns and their relationship to academic performance.
Digital and Social Media
The proliferation of social media platforms has created vast datasets suitable for network analysis. Researchers use SNA to study information diffusion, opinion formation, community structure, and influence patterns on platforms such as Twitter, Facebook, and LinkedIn. Network analysis of digital communications reveals how online communities form, evolve, and interact.
Challenges and Limitations
Despite its analytical power, SNA faces several challenges. Boundary specification — determining which actors and relationships to include in the network — involves subjective decisions that can significantly affect results. Missing data, whether due to non-response in surveys or incomplete digital records, introduces bias. The computational demands of analysing large networks require specialised software and hardware resources.
Ethical considerations are particularly important in SNA, as network data often reveals information about individuals’ relationships and social positions. Researchers must carefully consider issues of consent, confidentiality, and the potential for harm when collecting, analysing, and reporting network data.
Getting Started with Social Network Analysis
For researchers and practitioners new to SNA, several steps can help build foundational skills. Begin by familiarising yourself with basic network concepts and terminology. Select a software tool appropriate to your needs — UCINET for comprehensive academic analysis, Gephi for visualisation-focused work, or R/Python for integration with broader analytical workflows. Start with a small dataset to practise basic operations before tackling larger, more complex networks.
Frequently Asked Questions
Social network analysis (SNA) is a methodological approach that examines the relationships and structures among social actors. Using concepts from graph theory and sociology, SNA maps networks of relationships — such as friendships, collaborations, or communications — to reveal patterns of connection, influence, and information flow that are not apparent through traditional analytical methods.
Popular SNA software includes UCINET (comprehensive academic analysis with NetDraw visualisation), Gephi (open-source visualisation platform), NodeXL (Excel-based, accessible for beginners), Pajek (designed for very large networks), and programming libraries such as igraph and NetworkX for R and Python respectively. The choice depends on the size of the network, analytical requirements, and the researcher’s technical skills.
UCINET (University of California at Irvine NETwork) is a comprehensive software package for social network analysis. Developed by Borgatti, Everett, and Freeman, it provides tools for centrality analysis, subgroup identification, role analysis, and statistical testing. It includes NetDraw for network visualisation and is one of the most widely used SNA platforms in academic research.
SNA is used across many disciplines including sociology, public health, organisational studies, criminology, and education. Researchers use SNA to study disease transmission, organisational communication patterns, criminal networks, online communities, and knowledge diffusion. It is applied in both qualitative studies (mapping relationships) and quantitative studies (statistical modelling of network processes).
Key SNA metrics include centrality measures (degree, betweenness, closeness, and eigenvector centrality), network density (proportion of actual to possible ties), clustering coefficient (tendency of nodes to form groups), path length (distance between nodes), and measures of structural equivalence (similarity of nodes’ network positions). These metrics help identify influential actors, assess network cohesion, and detect community structures.