Social Networks

Social network analysis is the branch of sociology that treats relationships, rather than individuals or whole societies, as the basic unit of study.

13 frameworks, from 1934 to today · Sociology · Social Sciences

still a working view ended · each leads to its entry below

Overview

Social network analysis is the branch of sociology that treats relationships, rather than individuals or whole societies, as the basic unit of study. Its practitioners ask who is connected to whom, how those connections are patterned, and what those patterns do: how information travels, who gets a job, why a rumour spreads, which communities hold together and which fracture. What makes the field distinctive is a wager—that the shape of the ties among people has consequences of its own, not reducible to the attributes of the people themselves. A doctor's advice, a rumour, a new contraceptive, a job opening: each moves through a structure, and the structure explains things that individual traits cannot.

Sociometry and the first pictures of groups

The field's origin story begins in the 1930s with Jacob Moreno, a psychiatrist who developed sociometry: a method for asking people whom they liked, trusted, or would work with, and then drawing the answers as a diagram. Moreno's Who Shall Survive? (1934) presented these "sociograms" as tools for understanding group life, and he founded the journal Sociometry in 1937. The innovation was conceptual as much as technical: a group could be seen as a structure of choices, with stars, isolates, and cliques visible at a glance.

Around the same time, Kurt Lewin's group dynamics research treated the group as a system of interdependent parts, and his students carried the idea forward. Lewin's field-theoretic language—forces, valences, regions—was not network analysis in the later sense, but it established that group-level properties were real and measurable. After Lewin's death in 1947, researchers at the University of Michigan's Research Center for Group Dynamics, and later at MIT, turned to the systematic study of who talks to whom in small groups. The sociometric tradition gave the field its first data and its first pictures; what it lacked was a way to say precisely what the pictures meant.

Balance theory and the logic of signed ties

The first real theory of network structure came from Fritz Heider, whose balance theory (developed in the 1940s and published in The Psychology of Interpersonal Relations, 1958) concerned triads—sets of three people—and the sentiments among them. Heider's insight was that people prefer balanced configurations: if A likes B and B likes C, then A should like C; if A dislikes B and B likes C, A should dislike C. Imbalance—"my friend's friend is my enemy"—creates psychological pressure to change.

In 1956, Dorwin Cartwright and Frank Harary translated this into graph theory, showing that a signed network (one where ties are marked positive or negative) is balanced when its positive and negative ties partition the group into two camps with positive ties inside and negative ties between. The theory made a testable prediction: real groups should tend toward such partitions. It also introduced a lasting idea—that the sign of a tie matters, and that triadic configurations have consequences. Balance theory's limits became clear as well: many real networks are not balanced, and the theory says little about how large structures form.

Structuralist network analysis and the Harvard years

The decisive turn came in the 1970s, when a group of Harvard researchers made network analysis a general structural method. Harrison White, a mathematical sociologist, and his students—among them Mark Granovetter, Barry Wellman, and Ronald Breiger—developed the formal apparatus: blockmodels, which group structurally equivalent actors (people who occupy the same position in the network even if they are not directly connected), and algebraic representations of roles. The key claim was structuralist: social positions, not individual attributes, explain behaviour, and positions can be read off the pattern of ties.

The empirical payoff came in studies of community. In the 1950s and 1960s, researchers such as Elizabeth Bott had shown that conjugal roles varied with the density of couples' separate social networks; in the 1970s, Barry Wellman and colleagues in Toronto studied "personal communities," showing that urbanites' ties were not the dense, neighbourhood-bound bundles of the old Chicago School picture but sparse, spatially dispersed, and specialized—different ties for different purposes. The finding reframed the classic question of community: the question was not whether community had been lost but what shape it now took.

The strength of weak ties

In 1973, Mark Granovetter published "The Strength of Weak Ties," one of the most cited papers in sociology12. Granovetter distinguished strong ties (frequent, emotionally close, reciprocal) from weak ones (acquaintances), and argued that weak ties are disproportionately important for information flow1. The reason is structural: strong ties tend to cluster—your close friends know each other—so information circulating among them is redundant1. Weak ties bridge clusters, carrying novel information between them1.

Granovetter's evidence came from a study of how people found jobs: a large share of professional, technical, and managerial workers had heard of their jobs through acquaintances rather than close friends. The finding was counterintuitive—surely close friends help most?—and it made the case that network structure, not tie strength alone, governs access to opportunity. The paper also introduced a lasting vocabulary: bridges, the weak ties that connect otherwise separate clusters, became a central object of study.

Structural holes and social capital

In 1992, Ronald Burt extended the argument with the concept of structural holes: gaps between clusters that a well-placed actor can bridge. Burt argued that people whose networks span holes—who have contacts in otherwise disconnected groups—enjoy advantages in information and control, because they are the only route between the groups. His empirical work on managers in a large firm showed that those with networks rich in structural holes were promoted faster and evaluated more favourably.

The concept connected network analysis to the broader idea of social capital—the resources available through relationships34. James Coleman had argued in the 1980s that dense, closed networks generate trust and enforce norms, which helps children succeed in school and communities solve collective problems. Burt's structural holes and Coleman's closure were in tension: one celebrated brokerage, the other density. The debate sharpened the field's central question—which network shapes produce which goods—and made social capital a bridge between network analysis and the rest of sociology.

Small worlds

In 1967, Stanley Milgram's "small-world" experiments asked people to forward a letter to a stranger through acquaintances; a surprising share of letters arrived in a handful of steps, and the popular phrase "six degrees of separation" followed. The finding was suggestive but methodologically fragile. The decisive formal step came in 1998, when Duncan Watts and Steven Strogatz published a model showing that a network with mostly local ties plus a small number of random long-range ties has both high clustering (like a lattice) and short average path lengths (like a random graph). The "small-world" property—short paths in a clustered network—was now a precise structural idea, and it explained how information could travel widely without everyone knowing everyone.

Scale-free networks and preferential attachment

In 1999, Albert-László Barabási and Réka Albert proposed that many real networks are scale-free: their degree distribution follows a power law, meaning a few nodes have enormous numbers of ties while most have few5. Their mechanism was preferential attachment—new nodes tend to connect to already well-connected nodes, so the rich get richer56. The model generated networks with hubs, and it explained why some networks are robust to random failure but fragile to targeted attack on hubs. The claim was influential and contested: later work showed that many social networks are not strongly scale-free, and that measurement choices matter. But the model put degree distributions and growth processes on the field's agenda.

Diffusion and contagion

How do things spread through networks? The diffusion tradition runs from Ryan and Gross's 1943 study of hybrid seed corn adoption among Iowa farmers—where adoption followed social channels—through Everett Rogers's Diffusion of Innovations (1962), which codified the stages and the role of opinion leaders. Network models of contagion formalized the idea: whether an innovation or disease spreads depends on the network's structure and on the threshold of adoption. Mark Granovetter's 1978 threshold model showed that small differences in individuals' thresholds could produce large differences in collective outcomes, and that the same distribution of thresholds could yield either a cascade or a fizzle depending on the network. The tradition later split into simple contagion (one exposure can suffice, as with a virus) and complex contagion (multiple exposures needed, as with many behaviours), a distinction that shaped how researchers modelled everything from smoking to protest.

Computational social science and causal inference

From the 2000s, the field changed scale. Digitized data—email logs, phone records, online platforms—made it possible to observe whole networks rather than samples. In 2009, David Lazer and colleagues named computational social science as an emerging field, arguing that the volume and granularity of digital traces demanded new methods and new institutions. Studies of massive email and mobile-phone networks confirmed old findings (weak ties bridge, networks are clustered) and raised new questions about how ties form and dissolve.

At the same time, the field turned to causal inference: the problem that network ties are not randomly assigned, so observed associations between network position and outcomes may reflect selection rather than influence. Researchers developed methods—instrumental variables, propensity scores, and later randomized experiments and natural experiments—to separate the two. The classic question, "Do networks cause outcomes or merely correlate with them?", became a methodological programme, and the field increasingly ran experiments, from get-out-the-vote studies to randomized introductions in online platforms, to answer it.

The present

Today the field is plural. Formal network statistics and exponential random graph models describe how ties form; longitudinal and causal designs test what ties do; computational methods handle data at a scale Moreno could not have imagined. The old debates persist in new form: whether structure or attributes explain more, whether closure or brokerage produces social capital, whether contagion is simple or complex. The questions still open are the ones the field began with—what shapes networks take, why, and what those shapes make possible—now asked with better tools and against a much larger evidentiary base.

References

  1. Interpersonal ties (History, introduction), Wikipedia
  2. Mark Granovetter, Wikipedia
  3. Structural holes, Wikipedia
  4. Social capital, Wikipedia
  5. Scale-free network (History), Wikipedia
  6. Preferential attachment (History), Wikipedia

The frameworks of Social Networks

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1934-1960
  2. 1940-1970
  3. 1943-Present
  4. 1946-1980
  5. 1955-1990
  6. 1967-Present
  7. 1970-2000
  8. 1973-Present
  9. 1980-Present
  10. 1988-Present
  11. 1999-Present
  12. 2000-Present
  13. 2009-Present
A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Social Networks is a planet in Nousfarer

Its city has a tower for each of these 13 frameworks. Land, walk in, and climb one concept at a time.

Land on Social Networks

Related fields

More in Sociology

See where Social Networks sits on the Map of Human Knowledge