Books & TeachingNetwork Analysis

Books about social networks for beginners, grouped by what you want from them

A reading guide to books about social networks for beginners: popular science first, then the textbooks, then the anthropology where network analysis began.

Portrait of Stefan NovakStefan Novak · · 8 min read
Books about social networks for beginners, grouped by what you want from them

For a first book about social networks, pick a popular account: Duncan Watts's Six Degrees if you want to know why the world feels small, Albert-László Barabási's Linked for hubs and how networks grow, or Nicholas Christakis and James Fowler's Connected for the claim that habits and moods travel along ties. Once you have data of your own, move to a textbook, either Borgatti, Everett and Johnson's Analyzing Social Networks for social scientists or Mark Newman's Networks for the full mathematics. Behind both lies an older line of work in anthropology, where some of the first network studies came out of fieldwork.

The books were chosen for clear writing and a steady reputation among teachers and readers, and each note below is written from the publisher's description, the catalogue record and the book's standing in the field.

Which book for which purpose

What you want Start with Go on to
The big ideas, no equations Watts, Six Degrees Barabási, Linked
Why behavior seems to spread Christakis and Fowler, Connected Shalizi and Thomas (2011) on the statistical catch
To analyze network data you have collected Borgatti, Everett and Johnson Wasserman and Faust
The mathematics, properly Newman, Networks Wasserman and Faust for the social science side
How anthropologists use networks Bott, Family and Social Network Hage and Harary; McCarty and colleagues
Friendship and the size of personal circles Dunbar, Friends Our explainer on the evidence for Dunbar's number

All three of these were written for general readers by researchers who helped build the field. Between them they cover the three ideas a beginner meets first: short paths, hubs and influence.

Six Degrees, Duncan J. Watts (2003)

Watts is the first name in the Watts and Strogatz model. In a 1998 paper in Nature, he and the mathematician Steven Strogatz showed that a network made of tight local clusters needs only a handful of random long-range ties for the paths between any two members to become short. The book tells that story from the inside, then follows network thinking out into the Internet, economies and organizations. Read it first if "six degrees of separation" is what brought you here. Our explainer on small worlds and six degrees works the same model by hand on a ring of 20 houses.

Linked, Albert-László Barabási (2002; paperback 2014)

Barabási, a physicist at Northeastern University, writes about the pattern he and Réka Albert reported in Science in 1999. In many real networks a few nodes, the hubs, hold a large share of all the links. The two authors called such networks scale-free and traced them to two ingredients: networks grow, and newcomers prefer to link to nodes that are already well connected. The publisher's description runs from cocktail parties to terrorist cells to bacteria, which tells you the book's ambition: one set of principles for very different systems. It suits readers who enjoy that kind of sweep, and it is a useful contrast to anthropology, where a village network is small enough that every hub has a name.

Connected, Nicholas A. Christakis and James H. Fowler (2009; paperback 2011)

Christakis, a physician and social scientist now at Yale, and Fowler argue that emotions and health behaviors pass from person to person along ties, so that a friend of a friend you have never met can make a difference to you. The research behind the book used records from the Framingham Heart Study. In the New England Journal of Medicine in 2007, they followed 12,067 people assessed repeatedly from 1971 to 2003 and reported that clusters of obese people extended to three degrees of separation; a 2008 paper in the BMJ reported the same reach for happiness.

Read it alongside its critics. Cosma Rohilla Shalizi and Andrew Thomas (2011) showed that in observational network data, influence along ties, the tendency of similar people to form ties, and the effect of each person's own traits are generically confounded, so telling them apart takes strong assumptions. That point is worth more to a beginner than any single finding, and Connected makes it easy to see why it matters.

† Links to shops are affiliate links: if you buy through one, we may earn a commission.

Textbooks: when you have data

The first measures you need are simple. Density is the share of possible ties that are actually present. For n people and L undirected ties:

density = 2L / (n(n - 1))

A hypothetical hamlet of 6 households with 7 ties among them has 2 × 7 / (6 × 5) = 14/30, about 0.47. You can check numbers like these in our network density calculator, and network density and centrality, worked by hand adds degree, closeness and betweenness on one small example. The three books below take you from there to the rest of the field. They differ mostly in how much mathematics they expect.

Analyzing Social Networks, Stephen P. Borgatti, Martin G. Everett and Jeffrey C. Johnson

The second edition (SAGE, 2018) is the gentlest of the three, aimed at social scientists who have a question about a network and want to answer it. Its companion website offers the datasets used in the book and worked tutorials showing how to do each analysis it discusses, with links to the UCINET software. The same authors, joined by Filip Agneessens, also wrote Analyzing Social Networks Using R (SAGE, 2022) for readers who would rather work in R, and Agneessens is listed as a fourth author on the 2024 edition of the main book.

Social Network Analysis: Methods and Applications, Stanley Wasserman and Katherine Faust

Cambridge University Press published this in 1994 as volume 8 of its Structural Analysis in the Social Sciences series, and at 857 pages it is the standard reference. The publisher presents it as both a reference for scholars and a textbook for researchers with network data in hand. Its parts run from the mathematical representation of networks through structural and locational properties, roles and positions, and dyadic and triadic methods, to statistical models for dyads. It rewards dipping in for a precise definition more than reading straight through.

Networks, Mark Newman (2018)

Newman's book first appeared in 2010 as Networks: An Introduction; the second edition dropped the subtitle. The publisher described the first edition as bringing together work on networks from mathematics, physics, computer science, biology and the social sciences: measurement, graph theory, algorithms, spectral methods, random graph and generative models, and processes that spread across networks. Expect linear algebra and probability throughout. It is the second textbook for anyone who wants to know why a measure works, not only how to compute it.

The anthropology of networks

Network analysis owes a debt to a group of anthropologists around Max Gluckman and the Manchester school. John Barnes, a student of Gluckman, was among the first to use the idea of a social network in a scholarly analysis, in a 1954 article in Human Relations on Bremnes, a parish on the Norwegian island of Bømlo. Three books carry the line forward, and one more brings it into present-day fieldwork.

Elizabeth Bott's Family and Social Network (Tavistock, 1957) is the place to start, because its central idea is easy to state and to test. From a study of London families she proposed what became known as the Bott hypothesis: the denser the husband's and wife's separate networks, the more sharply the couple divided their household roles. "Denser" here is the density formula above, which makes the book a natural first exercise. J. Clyde Mitchell edited Social Networks in Urban Situations: Analysis of Personal Relationships in Central African Towns (Manchester University Press, 1969), which gathered network studies of town life in Central Africa. Per Hage and Frank Harary, an anthropologist and a graph theorist, wrote Structural Models in Anthropology (Cambridge University Press, 1983) to give anthropologists the graph theory itself.

For fieldwork now, Conducting Personal Network Research: A Practical Guide by Christopher McCarty, Miranda J. Lubbers, Raffaele Vacca and José Luis Molina (Guilford, 2019) is written at an introductory level. It covers study design, name generators, measures of a personal network's composition and structure, statistical models from regression upward, and the ethics of collecting network data.

Robin Dunbar's Friends belongs here as the evolutionary view of the same personal circles. Dunbar, the psychologist who proposed that our capacity for friendship tops out at around 150 people, looks at how different kinds of friendship and family ties overlap, and at the psychological and behavioral mechanisms that make friendships possible to form and keep. Read it next to what the evidence says about Dunbar's number, which sets the 150 against later reanalyses of the data.

A reading order for one term

  1. Six Degrees, for the shape of the field.
  2. Family and Social Network, for one hypothesis you can test with a density calculation.
  3. Analyzing Social Networks, chapter by chapter with the companion datasets.
  4. Connected together with Shalizi and Thomas, as a lesson in what observational network data can and cannot show.
  5. Newman or Wasserman and Faust, kept on the desk as references for whatever measures your own project needs.

Sources and further reading

More in Books & Teaching