Network analysis
Study relationships rather than isolated cases
Network analysis represents entities as nodes and defined relationships as edges, then examines structure through measures such as degree, centrality, paths and communities. Directed, weighted, temporal and bipartite networks require different specifications. The sampling boundary and edge definition are substantive decisions. Network visualisation supports interpretation but is not itself evidence for a structural or causal claim.
Choose it when the question concerns connections, influence opportunities, collaboration or system structure and relational data can be defined consistently, including missing ties and the population boundary.
Strengths
- Captures relational patterns ordinary case-level summaries miss
- Supports explicit analysis of clustering, brokerage and connectivity
Limitations
- Missing nodes or ties can distort structural measures
- Dependence and arbitrary boundary choices complicate inference
Know the boundary
A central node is not automatically causally influential, and a visually separated cluster need not be a substantive community.