Network analysis glossary

What is social network analysis?

Social network analysis (SNA) is the study of the ties between people rather than the people themselves. It treats a group as a graph of nodes and edges and asks structural questions: who is central, who connects separate groups, how tightly knit the group is, and where it would split.

Updated · Netgraf

The core idea

Most analysis looks at individuals and their attributes: age, role, income. Network analysis looks at the relationships between them, on the idea that where a person sits in a web of ties shapes what they can see and do. Two people with the same job title can hold very different amounts of influence because one of them is the only link between two departments.

The field grew out of Moreno's sociograms in the 1930s, took its mathematical form from graph theory in the 1950s to 1970s, and now runs through sociology, organisational research, epidemiology, journalism and intelligence work.

The measures, in plain words

MeasureThe question it answers
Degree centralityWho has the most ties?
Betweenness centralityWho sits on the paths between everyone else?
Closeness centralityWho can reach everyone in the fewest steps?
Eigenvector centrality and PageRankWho is tied to other well-tied people?
DensityHow many of the possible ties exist?
Clustering coefficientDo my contacts know each other?
CommunitiesWhich groups are tighter inside than out?
CentralizationIs the network run through one hub?
Brokers and cut pointsWhose removal would split the group?
AssortativityDo the well connected stick together?

Three classic studies

  • Zachary's karate club (1977): 34 members of a university club that split in two after a dispute. The network drawn before the split predicts, almost exactly, who went with which leader.
  • The Florentine families (Padgett and Ansell, 1993): marriage ties among Renaissance Florence's elite. The Medici were not the richest family, but they sat between families that had no ties to each other.
  • Krackhardt's kite (1990): a ten-person network built to show that "most central" has at least three right answers depending on the measure.

All three are open as live maps below. Open one and press Insights to see its measures.

Doing SNA without code

Serious network research usually runs in Gephi, R (igraph, statnet) or Python (NetworkX). For a group of tens or a few hundred people, most of the measures above are also one click away in Netgraf's insights panel, and exports carry them as columns in the CSV and GraphML so the work can continue in those tools. See Netgraf vs Gephi for where each fits.

Example maps

Open one to explore it, then press Edit a copy to make it yours.

Questions

What data do I need for social network analysis?
At minimum an edge list: two columns naming who is tied to whom. A second table of people with attributes such as role or team makes the results easier to read. Surveys, email metadata, co-authorship and public records are common sources.
Which social network analysis measure should I use?
It depends on the question. Use degree for popularity or activity, betweenness for brokerage and control over flow, closeness for speed of reach, and eigenvector or PageRank for influence through well-connected contacts. Reporting two or three together is usually more honest than picking one.

References

  1. Wasserman, S., & Faust, K. (1994). Social Network Analysis: Methods and Applications. Cambridge University Press.
  2. Zachary, W. W. (1977). An information flow model for conflict and fission in small groups. Journal of Anthropological Research, 33(4), 452-473. Link
  3. Padgett, J. F., & Ansell, C. K. (1993). Robust action and the rise of the Medici, 1400-1434. American Journal of Sociology, 98(6), 1259-1319. Link
  4. Krackhardt, D. (1990). Assessing the political landscape: Structure, cognition, and power in organizations. Administrative Science Quarterly, 35(2), 342-369. Link

Draw your own

Free, no account, and private by default. Start blank, import a spreadsheet, or describe the network in words.