The Blackwell Companion to Sociology (75 page)

with?), on the researcher's taste (does this version look better structured than that one?), on the ease of use of available computer packages, and so forth. The choice one makes is often the only one the rest of the world ever gets to see, so it is a weighty matter. It might seem that we could escape responsibility by sticking with the original data: use sociograms, in which all the actors are represented by symbols and all the relationships by various kinds of lines between them. This is getting to be a lot of fun, as people are developing nifty ways to show networks in three dimensions, from a mobile viewpoint, with gorgeous colors. (See Lin

Freeman's delightful web page http://eclectic.ss.uci.edu/~lin/galery.html). Yet we still have to look at an arrangement at any one time, and the way the actors and lines are arranged affects how people interpret what they see; different arrangements of the same network lead people to different conclusions (for example,

McGrath et al., 1997).

Centrality is another important structural topic that can be addressed only by

studying whole networks. A more central actor is in some way better placed in

the network as a whole: better able to get in touch with other actors, more visible to them, more able to control the flow of information among them, and so on.

320

Bonnie Erickson

This broad concept has alternative interpretations. One popular measure of

centrality is degree, or the number of other actors to which an actor is linked.

There is just one measure of degree if relationships are non-directional, or the same from actor A to actor B as from actor B to actor A, as in A and B are

partners. But often ties are not the same both ways: A is B's superior, so A gives B

orders but B cannot boss A around, or A loves B with unrequited passion. Then

indegree is the number of ties an actor receives, and outdegree is the number the actor sends. Though these tend to be related, they can differ in interesting ways.

Elite actors tend to attract more attention (indegree) than they return, like

famous sociologists who are aware of only a fraction of the people who pay

attention to them and their work. Sometimes one kind of degree is clearly the

more theoretically appropriate. For example, in a network of competitive bridge

players (Erickson and Nosanchuk, 1984), being known to a larger number of

fellow players (having high indegree) means playing both one's good and bad

performances before a larger audience. Thus more successful players get more

esteem from fellow players, and get especially high esteem if they have high

indegree. At the same time, more aggressive players receive more aversion,

especially if they have high indegree.

Another popular version of centrality is betweenness, or the extent to which

an actor is part of paths linking others and hence controls connections between

them. The Medici had the highest betweenness in the elite of Florence, and it is this that Padgett and Ansell (1993) emphasize as the structural root of their

power.

Studying whole networks gives us important information about the position of

social actors within a social structure. Once we have this, we can easily combine it with attribute information to do more familiar kinds of work. For example, we can classify scientists into more or less central groups, and then consider how

their later careers are affected by structural position, productivity, gender, and so forth.

Although whole network studies are wonderful they are not always feasible. A

popular and powerful alternative is the study of actor-centered networks con-

sisting of a focal actor and those tied to that actor. This approach combines

neatly with sociology's most popular research tool, the survey: sample and

interview people as usual, but include questions about each person's contacts.

Not only is this strategy often the only practical one, but it is also one that lets us ask some profound questions. People's networks are their immediate social

environment, the part of society they live in. What are such networks like,

how do they get to be that way, and what difference do they make to people's

lives?

Concerning what networks are like, we know a good deal about close ties and

not nearly enough about weaker ones. It is much easier to ask people about those near or dear to them: there are not too many people to ask about, and people

know the answers. Researchers have successfully asked people about the people

they feel close to, the people they discuss important matters with, the people they can call on for help of various kinds, the closest people in particular role

relationships like co-worker or friend, the people they talk to or e-mail most

Social Networks

321

often, and so forth. We know that the close are few, ranging from a couple of

people to a couple of dozen on average, depending on the kind of closeness.

Close relationships call for a good deal of investment of time, emotion, commit-

ment, risk, and so forth, so people cannot maintain a huge number of close ties ±

and the people who can maintain a larger number are, unfairly, the people with

other useful resources such as better jobs and education (for example, Fischer,

1982). We know that, few though close ties are, they are specialized. For

example, people get different types and amounts of help from ties of different

degrees of closeness, from people in different role relationships, and from

people with different attributes like gender (for example, Wellman, 1999).

Specialization occurs partly because of convenience (it is neighbors who have

lawnmowers next to your lawn), partly because of our cultural definitions of

appropriateness (you would readily ask your neighbor to lend the mower, but

ask your parents to lend the down payment on your house), and partly because

of socialization (women learn to listen to people's troubles more patiently

than men).

We know that the people we are close to tend to be very much like ourselves,

and the closer they are the more alike they are. This similarity, which turns out to have far-reaching consequences (discussed below), arises from both choice and

constraint. People feel more in common with, and more attracted to, other

people who are like them in salient respects (homophily). In addition, people

more often meet similar others because similar people spend time in the same

places (Feld, 1982). Sometimes they obviously have no choice, like schoolchil-

dren who spend most of their days with other pupils of the same age, but even

apparently free choice of settings leads to constrained choice of friends because similar people make similar choices. People join voluntary associations that

attract people like them in age, gender, occupation, and education, then, given

a choice already narrowed to people rather like them, they make friends of

people still more like them (McPherson and Smith-Lovin, 1987).

Close ties matter: people with better social support live longer, people with

serious problems in their inner circles (like the death of a loved one) get

depressed, people with strong ties to criminals go bad, while former criminals

with good jobs and marriages go straight. Those near and dear are especially

important for our access to the kind of support that depends primarily on

another person's willingness and availability: getting companionship from

friends, relying on long-term nursing care from a close relative. Just how this

works varies from one context to another because cultures define appropriate

help differently, people have different priorities of help to seek (with companionship high for some and survival necessities for others), the people they are close to have different resources to offer, and so on. There is a great deal of interesting work to be done on comparisons of different subgroups and countries. Wellman

(1999) includes reports from some very different countries, while Degenne and

Forse (1999) offer rich comparisons between France and other places. At the

same time, some contextual differences widely thought to make a difference do

not: people in the big, supposedly impersonal city have as many and as helpful

supporters as do people living in the fabled coziness of small towns (Fischer,

322

Bonnie Erickson

1982). With all this rich brocade of detail, elegant universal generalizations are hard to come by, and much of the fun lies elsewhere: relishing and making sense

of endless variations.

Meanwhile, weak ties matter too ± more than people often realize. For

example, people often get jobs through personal contacts, and often weaker

ties produce more jobs or better jobs than do strong ties. Such surprising findings inspired one of the best known paradoxes of network analysis, ``the strength of

weak ties'' (Granovetter, 1995). How can weak ties be strong? In part through

sheer numbers. Though close to at most a few dozen people, the average North

American knows about 1,400 (Killworth et al., 1990). Network size is import-

ant, as fundamentally important for as us as is the speed of light for others, yet our knowledge of it is quite crude. The fundamental problem is that we cannot

ask people to report their network size: they do not know. Thanks to the

remarkably ingenious and sustained research program of Bernard, Killworth,

and collaborators, we have some still spotty and rough ideas. Notably, network

size grows with social status, so the benefits of larger networks are unequally

distributed and help to reproduce inequality. For example, the more people you

know, the greater the chance that you will just happen to talk to one, who will

just happen to learn you could be open to a job offer or who will just happen to mention an opening; many job changes occur in this apparently haphazard way

that systematically favors the well connected (Granovetter, 1995).

Granovetter argues that weak ties are strong not because of their weakness as

such, but because weak ties include structural bridges linking clusters of otherwise separated people who have different pools of resources. Not all weak ties

are bridges (some weak ties link people in the same cluster) but almost all bridges are weak ties. Strong ties rarely bridge because the people close to us are too

much like us: they know much the same things and much the same people as we

do. Thus, despite their eagerness to help us, often they can do little to help us reach scarcely distributed resources like better jobs, while our mere acquaintances offer varied and wide-ranging contacts with many resources outside our

and our intimates' reach. Granovetter's original research suggested that male

managers, professionals, and technical workers got better jobs through weak

ties, perhaps because weak ties combine the structural advantages of wide

searches (by both the job seeker and potential employers) with the decision

advantages of having a trusted contact vouch for some of the many assets that

are vital to higher level work but very hard to measure. After 25 years of further research inspired by his, Granovetter (1995) thoughtfully assessed how and why

the results of using weak or strong ties vary greatly, in a fascinating account too rich to review here.

Lin takes a somewhat different approach: weak ties matter because they

connect us to a wider variety of people, including people varied in occupational prestige, and hence weak ties are more likely to include some people with higher occupational prestige than our own who can help us into better jobs than we

already have. One interesting implication is that the game is different for elites: their strong ties to powerful people like themselves are worth more than their

weaker ties to humbler folk. Lin has shown that people's acquaintances include

Social Networks

323

more occupational diversity than their friends or kin, that people with more

occupational diversity in their networks are better able to use a high prestige

contact to get a job, and that the high prestige of the contact (not the weakness of the tie as such) predicts the higher prestige of the jobs obtained through

weaker ties. See Lin (1999) for a useful review of his own and other work on

networks and occupational status attainment.

Burt (1992) takes yet another view: what really matters is structural holes, or

the gaps between groups of people to which a person is connected. It is these

gaps which make the others dependent on that person (as discussed above) and

hence help that person to profit. The ties the person has to these very separate groups are not likely to be strong ones, because people who have strong friends

in common tend to know each other or soon come to do so. Again, like

Granovetter and Lin, Burt carefully points out that weakness as such is not the

strength of weak ties; instead the ties that give structural advantage are usually weak ones, for good network reasons. Burt goes on to argue that structural holes help people to use their resources to better advantage, getting higher rates of

return from the same assets because of the bargaining power of brokerage

positions; thus good networks are a multiplier for other forms of capital.

The value of diversity in weak ties is a common thread in these accounts, and

also in some other intriguing work. Coser (1975) argues that people who interact with different kinds of people, in different role relationships, in different settings, develop a range of useful personal resources, including greater skill in the use of abstract language and a greater sense of personal autonomy. Other work suggests that these are considerable benefits, since (for example) more autonomous