EvalCompassEvalCompass
Analysis & synthesisCentralityNetwork densityCollective impactCollaboration evaluationVisualization

Social Network Analysis

Social Network Analysis
Also known as: SNA、Network analysis、Relationship analysis

Represents people or organizations as nodes and referrals, collaboration or information sharing as edges to examine connection structures. Define who is included and what counts as a tie first. [1]

Social network analysis design and implementation worksheet — openReferences and checked sections
Difficulty
Advanced
Cost
Medium
Duration
Planning example, not a universal standard duration: 2 weeks–2 months for survey design, collection, analysis and visualization
Content updated: 2026-09-13Source checked: 2026-09-13 (includes AI cross-checking)No documented human reviewIncludes fictional examples

How it works at a glance

Tie meanings depend on direction, strength and period. Examine information flows, concentration and isolation according to the question. [1]

Non-responding organizations create missing ties. Do not equate high centrality with organizational quality, and consider publication risks of identifying people or organizations. [1]

  1. Define boundaries and nodes
  2. Collect relationship data
  3. Create adjacency matrices or edge lists
  4. Calculate network indicators
  5. Visualize and interpret
  6. Compare over time or with benchmarks
Sequence of implementation steps. See “Implementation steps” for details of each stage.

When to use it and when it is unsuitable

When to use it

  • When visualizing and measuring stakeholder connections in programs strengthening collaboration
  • When identifying overall structure and key connectors in collective impact or multi-agency programs
  • When information, resources or support concentrate in certain organizations and bottlenecks need identifying
  • When visualizing collaboration outcomes through pre/post network comparisons

When it is unsuitable

  • When tie data are difficult to collect. Analysis requires systematic information on node relationships.
  • Evaluations mainly targeting individual attitudes or behavior rather than networks. Conventional surveys or intervention evaluations are appropriate.
  • When ties or network boundaries cannot be defined and collected coverage cannot be explained. Small networks can still be described.

Participants, roles, and deliverables

Participants and roles

Data collectors, analysts and participants. Check boundaries, tie definitions, sharing scope and consent; report effects of remaining missingness.

Main deliverables

  • Network diagram of centrality and edge weights
  • Network indicators: density, average distance and node centrality
  • Pre/post report interpreting intervention-related structural changes
  • Identified isolates and bridges

Implementation steps

  1. Define boundaries and nodes
    Decide who or what constitutes nodes and which edge types to analyze. Information exchange, funding and joint-program ties produce different results.
  2. Collect relationship data
    Use sociometric surveys asking all nodes about connections, existing documents, minutes or contracts, or observation. Whole-network surveys seek all nodes' responses; ego-network surveys examine ties from some nodes.
  3. Create adjacency matrices or edge lists
    Format data as node-by-node adjacency matrices or edge lists. Define how weights and direction are treated.
  4. Calculate network indicators
    Use Gephi, UCINET, R igraph or similar tools to calculate density, average distance and clustering, and degree, betweenness and eigenvector centrality. Identify isolates and bridging nodes.
  5. Visualize and interpret
    Create network diagrams with node size indicating centrality and edge thickness indicating strength. Interpret clusters, bridges and isolates against evaluation questions.
  6. Compare over time or with benchmarks
    Compare pre/post indicators such as density, reduced isolation or dispersed betweenness. Defining desired network structure beforehand supports benchmarking.

Strengths and limitations

Strengths

  • Provides a structural perspective different from aggregating individual characteristics.
  • Network diagrams intuitively communicate complex relationship patterns.
  • Quantitative and visual pre/post comparisons support accountability for collaboration programs.
  • Identifies isolated or overloaded intermediary nodes and suggests intervention points.

Limitations

  • Whole-network studies ideally include all nodes; non-response biases structure estimates.
  • Often relies on self-reports, which differ from actual interaction.
  • Commonly cross-sectional; dynamic network analysis over time is technically demanding.
  • Desirable density or centrality is contextual and not simply good or bad.

Common pitfalls

  • Producing diagrams without interpretation. Attractive visualization is a starting point, not the final output. Always explain how density or centrality changes relate to questions.
  • Do not code non-respondents' ties as absent. Check concentrated missingness, boundaries and collection coverage. Do not judge whole-network credibility solely by a universal response threshold.
  • Assuming denser is better. Closed clusters can trap information; weak ties may matter for external information exchange.

Conditions of use and evidence

  • Tie definitions and analytical boundaries determine indicator meanings because the method measures relationship structures. [1]
Fictional example; not actual results

Examine referral ties between four organizations

A fictional illustrative network of 4 organizations, with 1 undirected edge for a mutually confirmed referral relationship during the period.

ItemFictional value
Confirmed ties3 edges: A-B, B-C, C-D
Possible ties4×3÷2 = 6 edges
Density3÷6=0.5

Common errors in judgment

Common mistake

Judging non-responding organizations isolated

How to revise

Distinguish unobserved ties from absent ties.

ReasonMissingness may simply hide edges.

Common mistake

Combining referral, funding and information ties

How to revise

Record relationship types separately.

ReasonCombining different functions into one density prevents interpretation.

Organize your thinking by writing

Social network analysis design and implementation worksheet

A practical template is available to fill in and print on A4 paper.

Open worksheet

Related methods

Stakeholder AnalysisOutcome MappingCommunity MappingSocial Network AnalysisParticipatory Rural Appraisal / Participatory Learning and ActionDescriptive Statistics and Cross-tabulation

References and checked sections

  1. Social Network Analysis
    Olivier Serrat／Asian Development Bank · 2009
    Section consulted
    Social Network Analysis
    How it relates to the text
    Tie definitions and analytical boundaries determine indicator meanings because the method measures relationship structures.
    Source checked: (includes AI cross-checking)

Editorial and checking information

Content updated: Source checked: (includes AI cross-checking)No documented human reviewIncludes fictional examplesEditorial policy and scope of checkingCite this pageReport a correction

How to cite this page

EvalCompass Editorial Team. “Social Network Analysis”. EvalCompass, updated 2026-09-13. https://evalcompass.jun-nakatani.com/en/methods/social-network-analysis/
Add the date you accessed the page. Text created by this site is available under CC BY 4.0. Quotations, external materials, and bundled fonts have their own terms of use. When reusing content, credit the source, link to the license, and indicate any changes.

Suggest a correction or improvement to this page

Change history
  • — Checked sources for definitions and applicability conditions, and added fictional examples, reasons for judgments and examples for every worksheet field. Revised unverified claims about history and adoption. Also reviewed existing procedures and cautions, and adjusted fixed participant counts and thresholds and generalizations about institutions.