Fundamentals of Social Network Analysis

Based on Wasserman and Faust (1994) Chapters 1 & 2

Social Network: “…consists of a finite set or sets of actors and the relation of
relations defined on them.” (20)

Actor: “discrete individual, corporate, or collective social units.” (17). Also called
“node”

EX: Cornelia School pupils; U of M professors; U.S. voluntary associations;

European auto corporations; Australian tribes; South American cities;
African nations

Relation (tie): “The collection of ties of a specific kind among members of a

group…” (20) Also called “edge” or “line”

EX: best friends; coauthors; monetary donations; interlocking directors;

marital ties; intercity airflights; military alliances

The fundamental network unit is the dyadic relationship, a tie between two
actors. All larger, more complex units are constructed from dyads: triads, cliques,
subgroups, groups, positions, blocks, etc.

Social relations vary along 3 dimensions – direction, strength, content:

Direction: which actor sends (initiates) and/or receives the relation? Ties
may be “MAN” – mutual, asymmetric, null. Asymmetric ties are called
directed ties, in contrast to undirected or nonties.

Strength: the emotional intensity or frequency of interaction. Simplest

strength measure is binary (present/absent); other strength indicators may
take discrete or continuous values
EX: class attendance is binary; talking duration is continuous

Relational Content: a specific substantive connection among actors.

Relations are multiplex when actors are connected by more than one type

SOC8412 Social Network Analysis Fall 2009

 of tie (e.g., friendship and business). Analysts have proposed many
typologies of network contents; one example:

 Resource exchanges: Transactions in which one actor yields

control over a physical good or service to another actor in
return for some other kind of commodity (including money).
 Information transmissions: Communication exchanges in
which technical data, work advice, political opinions, or office
rumors flow from one actor to another.
 Power relations: Asymmetrical interactions in which one
party exerts control over another’s behaviors either by
applying force (coercion) or, more typically, by a superior
exercising the taken-for-granted expectation that commands
will be obeyed by subordinates (“legitimate power” or
authority in Weber’s sense).
 Boundary penetrations: Coordinated actions by two or
more actors to reach some mutual objective.
 Sentimental attachments: Emotional affiliations among
individuals that create solidarity and generate obligations of
mutual assistance and support.

P. 65 in David Knoke. 2001. Changing Organizations: Business Networks in

Social network data “consist of at least one structural variable measured on a

set of actors” (28). Methods of data collection depend on researcher’s decisions
about which variables, actors, and populations to examine.

Structural variables measure dimensions of dyadic relations

Composition variables are actor attributes/characteristics

(“substantialism”) EX: gender, race, class, age, sales, profit

Network mode refers to “a distinct set of entities on which structural variables are
measured’ (29) [In matrix format, senders appear in rows, receivers in columns.]

One-mode data identical sending & receiving actors; most common type

Two-mode data different send-receive actors. EX: producers & consumers

Affiliation network links actors to “events”. Note duality of actors and

events: persons connected via participating in same events, events
connected via overlapping participants. EX: SMOs & protests

SOC8412 Social Network Analysis Fall 2009 2

A researcher must decide how to bound a population (specify a set of actors) to
be sampled or completely enumerated (31-32).

Nominalist: analyst’s theoretical interests determine which actors possess

the relevant attribute(s). EX: grade in school; residential neighborhood;
professional association membership; global information sector industries

Realist: members themselves recognize and accept a shared identity that

defines the group’s boundary. EX: St. Paul politicians; environmental
activists; criminal gangs; AIG executives

Network sampling (random selection from a population) is problematic, both

conceptually and empirically (33-35). Wasserman & Faust also say little about
ego-centric networks, one focal actor (ego) and its direct contacts (alters),
except where ego-nets are embedded within larger networks.

Their focus is the complete network, an entire population of actors. Note the
requirement for a very high response rate from all population members (~100%).

Network data collection methods, cross-sectional & longitudinal; some studies

combine multiple designs (details pp. 45-54); see also Knoke & Yang (2008).

 Self-administered mail- or hand-back questionnaires

Formats: roster-free recall; fixed-free choices; ratings-rankings
 Personal and telephone interviews (at schools, homes, offices)
 Observations made at physical sites (playgrounds, factory floors)
 Archival records such as news reports, corporate databases,
citation indexes, biographies
 Other methods: cognitive social structures; small-group
experiments; ego-centered survey (GSS 1985 & 1987); diaries;
Small World (6 Degrees of Separation and the Kevin Bacon game)

Accuracy, validity & reliability, and measurement error issues abound in network
analysis, as in other social research methodologies.

SOC8412 Social Network Analysis Fall 2009 3