Kieran Healy’s Weblog Sociology and other distractions

Weak Ties and all That

Liz Lawley and Brad DeLong run across Mark Granovetter’s classic 1973 article, “The Strength of Weak Ties”. It’s well worth reading. If you’re interested, you can check out the syllabus for the Economic Sociology seminar I taught last Fall. (Update: Whoops, the link goes to the right syllabus now.)

Liz seems mainly interested in using the article to get into the literature on social networks rather than economic sociology. Social networks have become quite the thing recently, with which the publication of a number of popular books on the subect. Some of them are written by physicists and other natural scientists who have suddenly discovered the notion of scale-free networks and its applicability to social interaction. In sociology, the most prominent young researcher in the field is Duncan Watts, who recently published Six Degrees, a pop account of recent work in the field.

Bear in mind, though, that this is a research program that’s been developing in sociology since the early 1970s, and that some of the popular accounts are shockingly unaware of this (particularly the ones written by the physics people). The least-informed ones begin with Stanley Milgram’s notion of the small-world problem in the late ‘60s, skip lightly over the intervening period (perhaps mentioning Granovetter’s “Weak Ties” piece and maybe the pathbreaking work of his advisor, Harrison White) and then scoot straight to about 20 minutes ago.

I’m not an expert on networks, but several of my colleagues most certainly are. If you’re serious about getting into the literature on social networks, then take a look first at Forse and Degenne’s Introducing Social Networks, and then Wasserman and Faust’s Social Network Analysis: Methods and Applications. Or just look at Ron Breiger’s Soc 526 Syllabus or (even better) his suggested reading list to get a sense of how rich this field is.

While I agree that many of the “physics people” suffer from the delusion that this field began circa 1999, I’m a bit boggled, as a former colleague of his, that Duncan doesn’t qualify as one of those physics people. We are, after all, talking about a man whose Ph.D. was awarded in “theoretical and applied mechanics”. And the idea of a scale-free graph does come from physics, specifically the work of Albert and Barabasi…

Yeah, I know Duncan. He’s actually a good illustration of his own theory, as he’s a very effective bridge between two largely unconnected networks. (Very decent guy, too.) Barabasi’s book doesn’t do much to correct the misapprehension that the SFN concept is equivalent to the whole field.

Thanks, Kieran. Useful pointers.

You’re right that many people seem to be approaching social network analysis in computing/network contexts as somehow completely without prior theoretical basis, there are a number of people who do want to find ways to look at the intersections of the existing theoretical work, the impact of current technologies on the assumptions in those works, and the development of both theories and tools to support the new environments.

Massively networked environments are changing interactions very quickly, and the recursive relationship between the sociological theory and the tools themselves has always fascinated me.

I’m about to add another book to the reading list…The Social Construction of Technological Systems, which I found useful as a graduate student in thinking about the interconnections.

For truly cutting edge empirical work, look at some of the work Tom Snijders at Groningen is doing: http://stat.gamma.rug.nl/socnet.htm His work and some other recent work by Wasserman and colleagues (the p-star models) are allowing researchers to actually construct quantitative tests of network-models. Most of the earlier quantitative social-network analysis is almost exclusively descriptive.

Gee, Kieran, this sure sounds like a lot of work. Couldn’t we just talk out our ass about things we don’t really understand particularly well? I thought that was the whole point of blogging!

W&F is quite a bear and only recommended if you are really serious about SNA. If someone is interested in playing and displaying networks stuff get Steve Borgatti’s UCINET program. It comes with the Padgett and Ansell Medici family data.

Also a recent book by Monge and Contractor regarding network theories and communications is quite good.

The mystery of the small world problem is that so many people want to believe it. Evidence isn’t there folks. This is the short version of my review of the literature. This one appeared in Psychology Today. The massive one with all the footnotes you can find in Society or, I think, on my website.

Six Degrees of Separation: An Urban Myth?

By Judith Kleinfeld

In the first issue of Psychology Today, back in 1967, Stanley Milgram described the familiar “small world experience”:

Fred Jones of Peoria, sitting in a sidewalk cafe in Tunis, and needing a light for his cigarette, asks the man at the next table for a match. They fall into conversation; the stranger is an Englishman who, it turns out, spent several months in Detroit. . . . “I know it’s a foolish question,” says Jones, “but did you ever by any chance run into a fellow named Ben Arkadian? He’s an old friend of mine, manages a chain of supermarkets in Detroit. . . .”
“Arkadian, Arkadian,” the Englishman mutters. “Why, upon my soul, I believe I do! Small chap, very energetic, raised merry hell with the factory over a shipment of defective bottle-caps.”
“No kidding!” Jones exclaims in amazement.
“Good lord, it’s a small world, isn’t it?”

Milgram’s small world experiment took this idea a step further: his subjects could reach anyone in the country, maybe anyone on the planet, through a chain averaging just a few people.

In the intervening decades, Milgram’s findings have slipped away from their scientific moorings and sailed into the world of imagination. The “six degrees of separation” between any two people has been integrated into the intellectual world of educated people, and it has turned up in the media, movies and Web sites. A variant involving the actor Kevin Bacon has become a popular parlor game.

But Milgram’s startling conclusion turns out to rest on scanty evidence. The idea of “six degrees of separation” may, in fact, be plain wrong-the academic equivalent of an urban myth.

The question of how people are hooked up had long been a game among mathematicians: If you choose any two people in the world at random, how many acquaintances would be needed to create a chain between them? Ithiel de Sola Pool at the Massachusetts Institute of Technology and Manfred Kochen of IBM collaborated on mathematical models, but never felt that they had broken the back of the problem.

But Milgram believed he had made substantial progress, if not solved the problem outright. Rather than theorize, Milgram experimented. He asked “starters” from places such as Nebraska to send a folder through the mail to a target person in cities like Boston. The starters had to send the folder to someone they knew on a first-name basis. That person was to send the folder to someone closer, and so forth. Incredibly, Milgram reported that it took only five people in six jumps to reach a random stranger.

I had always regarded Milgram’s work as one of the great, counterintuitive studies in the social sciences and wanted to replicate it in the electronic age. In order to do so, I tracked down the details of Milgram’s papers in the Yale archives.

What I found was disconcerting: very few of his folders reached their targets. In his first, unpublished study, only 3 of 60 letters-5 percent-made it. Even in Milgram’s published studies, less than 30 percent of the folders got through. Few replications spanning cities had been done, and these showed few folders made it through, especially across class and race boundaries.

Perhaps people didn’t bother sending the letters on. That was Milgram’s explanation. But that seems unlikely. The folder was not a simple chain letter, but an official-looking document with heavy blue binding and gold logo. If the subjects knew how to reach the targets, they would have passed the folder along.

There is some evidence that Milgram might be right in spite of his own research. Duncan Watts at Columbia University and his colleagues have created mathematical models that show how a small world could work. Their research has created interest in fields such as disease transmission and corporate communication.

It is just as likely, though, that Milgram was wrong. But if we don’t live in a small world after all, why do people find this idea so easy to believe? My research suggests that first, the belief that we live in a small world gives people a sense of security, a feeling that we are all somehow holding hands. And small world experiences that we encounter naturally buttress people’s religious faith as evidence of “design.”

But also, there is a difference between what ordinary people mean by a “small world experience” and what mathematicians mean. When we say, “It’s a small world,” we are not talking about the chances of connection between two people taken at random. We are talking about the chances of meeting a person who knows someone from our past. Over a lifetime, these chances are high, especially for educated people who travel in similar networks.

And when an especially unlikely connection occurs, the world does feel small, whether or not the scientific evidence agrees.