Measuring Ego Networks

Ego Network Metrics: Size, Effective Size, Composition, Homophily, Diversity, and Clustering

Ego Networks: The Basic Ingredients

  • Ego: the central node whose network we are describing.
  • Alters: the nodes surrounding ego.
  • Ego-alter ties are obligatory: every alter is connected to ego.
  • Alter-alter ties are optional: alters may or may not know one another.

The Two Subgraphs of an Ego Network

  • The star subgraph contains all the ego-to-alter ties (\(E_{ea}\)).
  • The alter-alter subgraph is the local network among the people who are connected to ego.
  • Alter set: \(V = \{a, b, c, d, e, f, g, h\}\)
  • Alter-alter ties: \(E = \{ab, ah, ag, bh, bg, de, dg, ef, fg, gh\}\)

Ego Network Data: What You Need

There are four pieces of information we need to compute the basic ego-network properties:

  1. The number of ego-alter ties — the same as the number of alters!
  1. The number of alter-alter ties.
  1. Sociodemographic characteristics of ego: age, gender, racial identity, class, education.
  1. Sociodemographic characteristics of alters: the same attributes, for each alter.

Ego Network Properties: The Menu

Measure Ego-alter ties Alter-alter ties Attributes
Size ✓
Effective size ✓ ✓
Composition ✓
Homophily ✓ ✓
Diversity ✓
Clustering ✓
  • Every metric is a different way of summarizing the same small pile of information.

Size

  • Tells us whether your ego network is large or small — the number of alters (equivalently, ego-alter ties).

Small: Size = 2

Medium: Size = 5

Large: Size = 8

Effective Size: The Idea

  • Still tells us whether your ego network is large or small.
  • But discounts alters who are connected to one another as redundant.
  • Given by ego-network size minus the average degree of the alter-alter subgraph:

\[ES = S - \frac{2m}{n}\]

Effective Size: A Dense Alter-Alter Network

a b c d e f g h
a 0 1 0 0 0 0 1 1
b 1 0 1 0 1 0 1 1
c 0 1 0 1 0 0 0 0
d 0 0 1 0 1 0 1 0
e 0 1 0 1 0 1 1 0
f 0 0 0 0 1 0 1 0
g 1 1 0 1 1 1 0 1
h 1 1 0 0 0 0 1 0
  • \(n = 8\) alters; \(m = 14\) ties. Each tie appears twice in the matrix (above and below the diagonal); the shaded 1s count each tie once: \(m = 28/2 = 14\).

\[ES = 8 - \frac{28}{8} = 8 - 3.5 = 4.5\]

Effective Size: A Sparse Alter-Alter Network

a b c d e f g h
a 0 1 0 0 0 0 1 1
b 1 0 0 0 0 0 0 1
c 0 0 0 0 0 0 0 0
d 0 0 0 0 1 0 0 0
e 0 0 0 1 0 1 0 0
f 0 0 0 0 1 0 0 0
g 1 0 0 0 0 0 0 1
h 1 1 0 0 0 0 1 0
  • The shaded 1s give \(m = 14/2 = 7\). Same size (\(S = 8\)), but less redundancy: the sparse network has the larger effective size.

\[ES = 8 - \frac{14}{8} = 8 - 1.75 = 6.25\]

Effective Size: The Maximum

a b c d e f g h
a 0 0 0 0 0 0 0 0
b 0 0 0 0 0 0 0 0
c 0 0 0 0 0 0 0 0
d 0 0 0 0 0 0 0 0
e 0 0 0 0 0 0 0 0
f 0 0 0 0 0 0 0 0
g 0 0 0 0 0 0 0 0
h 0 0 0 0 0 0 0 0
  • No alter-alter ties at all (\(m = 0\)): a pure star. No alter is redundant, so nothing gets discounted.

\[ES = 8 - \frac{0}{8} = 8 = S\]

Effective Size: The Minimum

a b c d e f g h
a 0 1 1 1 1 1 1 1
b 1 0 1 1 1 1 1 1
c 1 1 0 1 1 1 1 1
d 1 1 1 0 1 1 1 1
e 1 1 1 1 0 1 1 1
f 1 1 1 1 1 0 1 1
g 1 1 1 1 1 1 0 1
h 1 1 1 1 1 1 1 0
  • Every alter knows every other (\(m = 28 = \binom{8}{2}\)): all alters are mutually redundant, so the whole circle acts like a single effective contact.

\[ES = 8 - \frac{56}{8} = 8 - 7 = 1\]

Composition

  • Tells us whether your ego network is dominated by one category of alters.
  • Given by the proportion of alters who belong to a category: kin, friends, neighbors, co-workers; gender, race, education.

\[Prop(Kin) = \frac{\text{number of kin alters}}{S}\]

  • Here: \(Prop(Kin) = 5/8 = 0.63\) — a kin-heavy ego network.

Composition: Another Example

  • Composition can be computed for any category the alters fall into.
  • The same network by gender identity:

    • \(Prop(Men) = 5/8 = 0.63\)
    • \(Prop(Women) = 2/8 = 0.25\)
    • \(Prop(Nonbinary) = 1/8 \approx 0.13\)
  • Same take: the network is predominantly male.

Homophily: The E-I Index

  • Tells us whether the people in your ego network are the same or different from you on some sociodemographic characteristic: age, race, gender, education.
  • Example: ego identifies as Black. Alters shaded the same color as ego share ego’s identity (“internal”); all others are “external.”
  • \(E\) = number of “External” alters (different from you); \(I\) = number of “Internal” alters (same as you).

\[EI = \frac{E - I}{E + I}\]

Homophily: Worked Example and Interpretation

  • Here: \(E = 6\), \(I = 2\) (alters b and f are Black, like ego).

\[EI = \frac{6 - 2}{6 + 2} = \frac{4}{8} = +0.5\]

  • Negative EI: preference for “internal” ties. An EI of \(-1.0\) is maximum homophily.
  • Positive EI: preference for “external” ties. An EI of \(+1.0\) is maximum heterophily.
  • EI = 0: no preference — equal numbers of internal and external ties. Here, ego’s ties lean external.

Homophily: A Homophilous Example

  • Flip the pattern: ego (Black) now has 6 Black alters and 2 White alters (c and g).

\[EI = \frac{2 - 6}{2 + 6} = \frac{-4}{8} = -0.5\]

  • Same magnitude as before, but on the homophily side: ego’s ties lean internal.

Diversity: Blau’s Index

  • Tells us whether you tend to connect with a wide range of others on some sociodemographic characteristic.
  • Given by Blau’s index of heterogeneity (\(H\)):

\[H = 1 - \sum_{k} p_k^2\]

  • Steps:

    1. Compute the proportion \(p_k\) of alters in each group \(k\).
    2. Square each proportion and sum them.
    3. Subtract from 1.0.

Diversity: Worked Example

  • Eight alters: 3 White, 2 Black, 2 Asian, 1 Hispanic.
  • Step 1, proportions of alters in each group: \(p_{\text{W}} = \tfrac{3}{8} = 0.375\), \(p_{\text{B}} = \tfrac{2}{8} = 0.250\), \(p_{\text{A}} = \tfrac{2}{8} = 0.250\), \(p_{\text{H}} = \tfrac{1}{8} = 0.125\).
  • Steps 2–3, square each proportion, sum, and subtract from 1.0:

\[H = 1 - \left(0.375^2 + 0.250^2 + 0.250^2 + 0.125^2\right)\]

\[= 1 - 0.281 = 0.72\]

Diversity: Interpreting \(H\)

  • A completely homogeneous ego network has \(H = 0\) — no diversity among alters.
  • The more diverse the ego network, the closer \(H\) is to 1.0.
  • \(H\) is at a maximum (\(H_{max} = 1 - 1/k\)) when alters are spread evenly across \(k\) groups. With \(k = 4\): \(H_{max} = 0.75\), so \(H/H_{max} = 0.96\).

Diversity: A Low-Diversity Example

  • Reverse the pattern: ego (Black) now has 7 White alters and 1 Black alter (f). Only \(k = 2\) groups are present.
  • Step 1: \(p_{\text{W}} = \tfrac{7}{8} = 0.875\), \(p_{\text{B}} = \tfrac{1}{8} = 0.125\).

  • Steps 2–3:

\[H = 1 - \left(0.875^2 + 0.125^2\right) = 1 - 0.781 = 0.22\]

  • Far below the \(H = 0.72\) of the diverse network. With \(k = 2\): \(H_{max} = 0.5\), so \(H/H_{max} = 0.44\) — alters crowd into one group.

Clustering

  • Tells us the extent to which your alters are interconnected among themselves (know one another) versus strangers to one another.
  • Given by the clustering coefficient (\(CC_i\)). You need:

    • The number of alter-alter ties (\(m\))
    • The number of alters (\(n\))
  • Don’t care about the ego-alter links!

\[CC_i = \frac{2m}{n(n-1)}\]

Clustering: A Clustered Example

  • \(n = 8\) alters; \(m = 17\) alter-alter ties.

\[CC_i = \frac{2 \times 17}{8 \times 7} = \frac{34}{56} = 0.61\]

  • The ties are concentrated: two dense friend circles — everyone knows everyone within their circle — plus a handful of cross-circle ties.
  • The clustering coefficient is the density of the network formed among your alters — the probability that two randomly chosen people you know also know one another.

Clustering: A Less Clustered Example

  • Same number of alters (\(n = 8\)), but only \(m = 7\) alter-alter ties.

\[CC_i = \frac{2 \times 7}{8 \times 7} = \frac{14}{56} = 0.25\]

  • The ties are thinner: two tight triangles of mutual friends, joined by a single bridge.
  • Same-size ego networks can differ sharply in how concentrated their alter-alter ties are — and the clustering coefficient picks this up.

Clustering: The Extremes

CC = 0.0 (\(m = 0\)): none of your friends know one another.

CC = 1.0 (\(m = 28\)): every one of your friends knows one another.

Summary: Measuring Ego Networks

Measure Question
Size How many alters are in the network?
Effective size How large is it net of redundant alters?
Composition Is it dominated by one category of alters?
Homophily (E-I) How similar are alters to ego?
Diversity (H) How varied are the attributes of the alters?
Clustering (CC) To what extent do alters know one another?
  • Choosing the right metric depends entirely on your research question.
  • Homogeneity and diversity are distinct: an ego network can be high on one and low on the other, depending on how the groups are defined.
  • Ego clustering measures network density, showing whether an ego is embedded in a single tight-knit circle or bridges separate, unconstrained social worlds.