
How Macrostructure, Social Foci, and Structural Features Shape Personal Networks

Social structure as distribution: Blau (1977) defines social structure as the distribution of people across social positions. These distributions determine who has the opportunity to meet whom.
Nominal parameters: categorical attributes with no rank order (gender, race, religion, nationality). Their distribution is heterogeneity.
Graduated parameters: rank-ordered attributes (age, education, income, wealth). Their distribution is inequality.

Broad demographic proportions determine the statistical probability of meeting ingroup versus outgroup members, heavily biasing ego networks.
Rule 1: Smaller Groups and Outgroup Ties
Rule 2: Larger Groups and Ingroup Ties
Rule 3: Correlated dimensions of association: macro structures rarely have independent parameters; dimensions of social differentiation are systematically correlated (race with religion, race with income, age with wealth).
Propagated bias: a network biased on one parameter is automatically biased on every correlated parameter — even if ego is entirely neutral about the second. Observed homogeneity often reflects a societal correlation, not a personal choice.
Left to right: categorical × categorical, categorical × graduated, graduated × graduated.


Foci differ in physical boundedness, entry barriers, and interaction costs (Hachen et al., 2024) — and that determines what they do to networks.
Concrete sites (student clubs): high face-to-face interaction, high entry barriers — excellent tie generators, but hard for friends to recruit each other into.
Broad cultural structures (musical genres): unbounded, low barriers — poor tie generators, but excellent taste diffusers through existing friends.
Hybrid foci (course areas, activities): concrete workspaces and shared cultural preferences — both roles at once.


Peter Marsden (1987) utilized the 1985 General Social Survey (GSS)—which used name generators to measure core discussion networks (“discuss matters important to you”)—to uncover fundamental structural rules of personal networks.
The Size-Diversity-Clustering Trade-off:
1. Network Size and Diversity (Marsden’s Rule):

As network size increases, diversity rises, homophily decreases, and density/clustering drops.
Setting: the Facebook networks of Dutch adolescents (Hofstra, Corten, van Tubergen, & Ellison, 2017; \(N = 2{,}810\), ~1.1 million ties), comparing small core (strong-tie) with larger extended (weak-tie) networks.
Question: do the classic offline rules of opportunity (Blau) and context (Feld) still govern friendship in a seemingly unbounded online world?
Findings: online networks are far more ethnically segregated than gender segregated — gender is a 50/50 population split, ethnic group sizes are not — and classroom and school composition (Feld’s foci) directly predicts online segregation.

Schematic networks: a dense five-alter core and a sparse nine-alter extended network. They illustrate the contrast, not the study’s data.

Pattern after Hofstra et al. (2017).
The composition of your personal network is heavily structured for you by forces beyond your direct control:
| Theorist | Core Source of Network Layout | Key Dynamic / Findings |
|---|---|---|
| Miller McPherson | Multidimensional social space and opportunity structures. | Distinguishes baseline (demographic pools) from inbreeding (choice) homophily. |
| Peter Blau | Macro-demographic proportions & societal parameter correlations. | Group sizes dictate outgroup probabilities. Biases propagate via parameter correlations. |
| Scott Feld | Organizational contexts, physical entities, and joint activities (foci). | Co-membership drives tie formation; focus constraint determines tie maintenance. |
| David Hachen et al. | Co-evolving coupled network ecology (one-mode & two-mode). | Classifies contexts as Tie Generators (clubs), Taste Diffusers (music), or hybrids. |
| Peter Marsden | Structural features, network size, and tie properties. | Expanding network size mathematically dilutes homophily and clustering. |