Appearance
Analyse an ego network when your evidence genuinely centres on one person and you cannot observe the wider population reliably — a single diary, a personal letterbook, or a biography-driven project. The ego network (one focal individual, their direct contacts, and the ties among those contacts) is the right scale precisely when a whole-network view would be mostly guesswork. When you instead want claims about a community or comparisons across many people, choose a bounded whole network instead.
What is an ego network, and what does it actually show?
An ego network has three parts: the ego (your focal figure), the alters (everyone directly tied to them), and the alter-alter ties (connections among the alters that do not pass through the ego). That last part is what makes it analytical rather than just a contact list — it reveals whether the figure sat among a tight clique or bridged otherwise separate groups.
The classic measures are size (number of alters), density of alter-alter ties, and the presence of structural holes — gaps the ego spans and could broker across.
When is an ego network the right choice?
Reach for it when these signals line up:
- Your sources are concentrated on one individual (a personal archive, a memoir, a single correspondent's outgoing letters).
- You can document that person's direct ties with reasonable confidence but not the wider field.
- Your research question is biographical or about an individual's social position and brokerage.
In these cases an ego network turns scattered relational mentions into a tractable, defensible structure.
When should I avoid ego networks?
Skip the ego approach when your question is really about a population or a system. Warning signs:
| Your goal | Better than an ego network |
|---|---|
| Compare positions across many people | Bounded whole network |
| Detect communities/factions | Whole-network community detection |
| Measure overall cohesion of a group | Whole-network density and components |
| Trace diffusion through a population | Whole network over time |
Forcing a population question into one ego's view produces a flattering but misleading portrait of a single hub.
How many alters do you need?
There is no strict minimum, but structure becomes unstable below roughly ten alters — a single extra letter can swing density or brokerage scores wildly. Always report the alter count and treat tiny egos (a handful of alters) as illustrative rather than as evidence for structural claims.
python
import networkx as nx
ego = "Aphra_Behn"
# G is the full reconstructed graph; radius=1 gives the immediate ego network
egonet = nx.ego_graph(G, ego, radius=1)
print("alters:", egonet.number_of_nodes() - 1)
print("density:", round(nx.density(egonet), 3))Can you compare two figures' ego networks fairly?
This is where most ego-network studies go wrong. A figure with a surviving letterbook will appear vastly more connected than one known only from others' mentions — but that gap is source survival, not sociability. Only compare egos documented by comparable source types, and state that comparability explicitly. When in doubt, compare shapes (density, brokerage) rather than raw sizes.
How do you handle the boundary problem?
Partial sources force a decision: which mentioned people count as alters? Make the rule explicit and mechanical, for example: an alter is anyone named at least twice in dated correspondence between 1670 and 1689. Apply it identically to every figure and record it in your methods. An undocumented, ad-hoc boundary is the difference between a reproducible study and an anecdote.
Key Takeaways
- Use ego networks when evidence concentrates on one individual and the wider population is unobservable.
- Avoid them for population-level, comparative, or community-detection questions — use a bounded whole network.
- Include alter-alter ties; without them you have a contact list, not a network.
- Report alter counts and treat egos below ~10 alters as illustrative only.
- Compare ego networks only across figures with comparable source survival.
- Define an explicit, mechanical boundary rule and apply it consistently.
Frequently Asked Questions
What is an ego network of a historical figure?
It is the sub-network centred on one person (the ego), containing their direct contacts (alters) and the ties among those alters. It captures an individual's immediate social world rather than a whole community.
When is an ego network the right choice?
Use it when your sources cluster around one well-documented individual — a single archive, diary or letterbook — and you cannot reliably observe the wider population. It fits biography-driven and single-collection research.
When should I avoid ego networks?
Avoid them when you want population-level claims about a whole community or want to compare structural positions across many people. A boundary-specified whole network serves those questions far better.
How many alters do I need for a meaningful ego network?
There is no hard minimum, but below about 10 alters most structural measures become unstable and anecdotal. Report the alter count and treat very small egos as illustrative rather than analytical.
Can I compare ego networks of two figures fairly?
Only if both were documented by comparable sources. Comparing a figure with a surviving letterbook to one known only from others' mentions confounds structure with source survival.
What is the boundary problem in an ego network?
It is deciding which contacts count as alters when sources are partial. Define an explicit inclusion rule — for example, named at least twice in dated correspondence — and apply it consistently.