Examine Cohesion, Components, and Communities
Move from one network picture to a defensible account of reachability, density, reciprocity, structural cohesion, and possible community divisions.
Method sources
By the end of this tutorial
- 1Distinguish whole-network density from reachability and component structure.
- 2Use reciprocity and cohesive-subgroup measures only when they match the tie definition.
- 3Treat community partitions as model-dependent summaries rather than natural labels.
SNA
Network specification
Analysis scenario
A professional learning network appears to contain several clusters. Leaders want to know whether knowledge can travel across the whole group, where fragile bridges sit, and whether a community algorithm reveals meaningful working groups.
Nodes
Teachers in one cross-school professional learning program.
Ties
A directed tie records at least one substantive exchange of teaching resources during the term.
Network type
One-mode, directed, weighted network with exchange counts and school membership attributes.
Step-by-step tutorial
Check reachability first
Identify weak and strong components before summarizing the whole network. In a directed network, weak reachability ignores arrow direction while strong reachability requires a directed path both ways.
Checkpoint
You can state how many actors are outside the main component and whether direction changes the reachability conclusion.
Measure density and reciprocity
Calculate density using the correct denominator for a directed network and calculate reciprocity only for ties where mutual exchange is meaningful. Compare results with a thresholded binary version if the original network is weighted.
Checkpoint
The denominator, loop rule, direction, weight treatment, and observation window are written beside every reported value.
Inspect cohesive structure
Look for articulation points, bridges, or node-connectivity patterns that reveal how easily a group separates. Use these diagnostics to describe structural vulnerability rather than to label people as problems.
Checkpoint
Any claimed bridge or cohesive subgroup is checked against the actual paths and the substantive meaning of the tie.
Compare community partitions
Run a documented community method on an appropriate undirected or directed representation, then vary resolution, threshold, or random seed. Compare partitions with known context without forcing school labels to validate the algorithm.
Checkpoint
The report names the algorithm and settings, shows stability checks, and uses community membership as an analytical summary rather than a fixed identity.
Interpret with care
A dense network has many observed ties relative to possible ties, but density does not guarantee that every actor can reach every other actor or that exchange is equitable. Components reveal reachability, while connectivity and bridge diagnostics reveal how robust that reachability is to the loss of particular nodes or ties.
Community detection optimizes a chosen criterion and may return different divisions under different methods or settings. A partition is most useful when it is stable enough to inspect, substantively interpretable, and reported alongside cross-community ties that prevent the visualization from exaggerating separation.