[Dora Costa and Matthew Kahn, guest-blogging, January 16, 2009 at 12:03am] Trackbacks
Empirical Social Networks Research

In this last blog entry, we would like to respond to some commentators and discuss our new research on the consequences of social networks. Dasarge and Fischer both discuss leadership. We can find some evidence of the importance of leadership. Black soldiers with an abolitionist officer were less likely to desert. But this was less important to their desertion decision than being in a company with guys like them. So to in response to Richard Aubrey, if you wanted to keep your men from deserting during the Civil War (and keeping men from deserting is a good idea if you want to win a war) having a homogeneous unit was the most important factor we could identify. We also find that men who had more of their (non-commissioned) officers in Andersonville were more likely to survive. Officers divided up the rations. We'd like to do more to identify the effects of leadership but our problem is identifying a good leader versus a bad one.

Our new research builds on the insights presented in our recent Heroes and Cowards book. We are convinced that empirical work on social networks offers social scientists with an interdisciplinary inclination to work together. Recently, sociologists, economists, political scientists and legal scholars have all made important contributions to this field.

Research on terrorist organizations indicates that friends and relatives join together and work together. By one estimate, roughly three quarters of mujahedin joined the global Salafi Jihad (of which Al Qaeda is a part) either as a group with friends or relatives or as men with close social ties to members (Sageman 2004: 114). Seventy percent of captured Italian Red Brigade terrorists had joined a friend who was in the terrorist organization -- a next-door neighbor, a school friend with whom he or she had spent vacations, or a cousin with who belonged to the same voluntary association (della Porta 1988). Social bonds, not a shared terrorist ideology, drove the decision to join. The ideology came later (Sageman 2004: 133).

The fundamental challenge in conducting social networks research is to identify who is a member of one's network and what might be the causal mechanism such that participating in such a network causes a behavioral change. Critics will always raise the valid point that people are rarely randomly assigned to a network. Do greens choose to live in Berkeley, California (selection) or does living in Berkeley cause one to embrace environmentalism (treatment)? The gold standard for separating "selection effects" from "treatment effects" is to identify situations when group members have been randomly assigned to a group. This randomization mitigates concerns about self-selection. The 2006 book, Are Judges Political? by Cass Sunstein et. al. highlights how legal scholars have approached these issues.

An ongoing health research agenda has linked social networks to health. People who report themselves to be socially isolated, both in the number and quality of their personal relationships face a higher mortality risk from all causes and from several infectious, neoplastic, and cardiovascular diseases. A large body of literature links stress, whether in the form of war, natural disasters, divorce, lack of control on the job, or even disrupted sleep patterns, to cardiovascular disease. Social networks could either mitigate or accentuate the effects of stress. They could mitigate the effects of stress through beneficial effects on psychological and physical well-being. But, they could accentuate the effects of stress if the stressor leads to the loss of friends or family (e.g. the well-established effect of death of a spouse on the mortality of a survivor).

In a forthcoming Demography paper, we investigate the interaction between stress and social networks in one of the few human populations to provide us with measures of stress, of long-run outcomes, and of exogenous social networks. We find that being in a more cohesive company reduced the negative, long-term consequences of wartime stress on older age mortality and morbidity, particularly from cardiovascular causes. Focusing on the role of stress factors in health and mortality may be a fruitful line of research, particularly as we exhaust the gains from public health advances. Matthew Kahn has returned to conducting research on environmental economics topics. While the U.S Civil War might appear to be light-years removed from urban pollution and Global Warming, social capital actually ties these sets of issues together. After all, there would be no "tragedy of the commons" if people felt altruism for each other and internalized the social costs of their actions ranging from littering to driving a Hummer around. In societies where there is greater social capital, there is less need for formal enforcement of anti-pollution laws. Social networks also can play an important role in the diffusion of new green products. If one's probability of buying a hybrid vehicle or solar panels is an increasing function of whether one's neighbors are purchasing such green products, then "bandwagon" effects can help to mitigate the challenge of climate change. Friends may also more quickly learn from their friends' experience with new unproven products. Such word of mouth learning will accelerate the diffusion of good ideas.

Richard Aubrey (mail):
"in response to Richard Aubrey" implies I said something other than that homogeneous units were best at cohesion.
I believe I mentioned, in the context of affirmative action, that it was nice to see short term costs acknowledged wrt diversity. IOW, diversity has costs and so homogeneity is a good thing.
The British practice of local recruiting (homogeneity) has--most likely--been responsible for the incredible steadiness of their small units for centuries.
1.16.2009 6:54am
Curt Fischer:
Thanks to both guest-bloggers for a fascinating week of posts. It was a very enjoyable experience.

Do greens choose to live in Berkeley, California (selection) or does living in Berkeley cause one to embrace environmentalism (treatment)?

I will be moving to Berkeley later this year. I suppose by the standards of the average VC commenter I am unapologetic environmentalist, but I suspect that by the standards of Berkeley I am a right-wing environment-destroying corporatist. If I had to pick one I would say that "treatment" is more important than "selection", although I do hope that in my case the "treatment" proves to be not very effective.
1.16.2009 8:30am
Tracy Johnson (www):
Should have track-backed to their other posts in the week.
1.16.2009 11:22am
Stuart Buck (mail) (www):
Critics will always raise the valid point that people are rarely randomly assigned to a network. Do greens choose to live in Berkeley, California (selection) or does living in Berkeley cause one to embrace environmentalism (treatment)? The gold standard for separating "selection effects" from "treatment effects" is to identify situations when group members have been randomly assigned to a group. This randomization mitigates concerns about self-selection.

I'm not sure why this is such an issue. Random assignment is both possible and useful when we're talking about medical treatments, prescription drugs, etc., but not for social networks. Social networks have the most power precisely because of someone's personal choice to be involved in that network.

The example I like to give is this: If you were randomly assigned (never mind how) to join a motorcycle gang (if you like motorcycle gangs, imagine some group you find completely repugnant), would you suddenly grow your hair long, buy leather pants and a Harley, and ride on the highway with no helmet, just to fit in? Probably not. Peer effects arise mostly when and where you chose to be part of the group, and the power of peer effects comes precisely from the fact that you admired the group and wanted to fit in.

So no one should expect to find that same peer effect outside of a chosen social group, nor is it useful (in a real-world sense) to try to figure out how much of a "peer effect" would arise in a random assignment situation.
1.16.2009 11:35am
Hannibal Lector:
A fascinating post. I plan to buy the book.
1.16.2009 11:43am
Mark Swanson:
It is reported (sorry, no reference but many popular detailed histories repeated it) that the German army in WW2 followed a policy of grouping new recruits into 4s in "boot camp" and afterwards always assigned them by that same group for all training and combat assignments. This was supposed to partially explain the German army's remarkable resilience.
1.16.2009 2:10pm
Richard Aubrey (mail):
The US Army once had OSUT (One Station Unit Training) which meant the same unit went through Basic and AIT. This was a change because Basic is traditionally for everybody. Some weapons, marching, military paperwork, minimal tactical work, field and compass training and so forth so that, if they happened to be a cook someplace which was in danger of being overrun, they wouldn't be trying to get the can opener on the end of a rifle.
To do OSUT, everybody in the Basic company would be going to the same AIT, learning the same specialty.
Organizationally complicated. But the impulse was for unit cohesion.
I recall reading many years ago about various German WW II units with a standard reference (recruiting district....someplace.\). Which, when you consider the size of the Wehrmacht, and the losses to be replaced, would have been difficult to sustain.
Read about Zhukov's secret defeat, a huge offensive that failed, before he got the hang of it. The fighting was immensely savage. After going back and forth over an objective, the Russians would be driven off and the Germans holding it with ten percent of the original unit plus a half dozen signalmen they found someplace. Some kind of resilience, alright.
1.16.2009 2:59pm

Post as: [Register] [Log In]

Remember info?

If you have a comment about spelling, typos, or format errors, please e-mail the poster directly rather than posting a comment.

Comment Policy: We reserve the right to edit or delete comments, and in extreme cases to ban commenters, at our discretion. Comments must be relevant and civil (and, especially, free of name-calling). We think of comment threads like dinner parties at our homes. If you make the party unpleasant for us or for others, we'd rather you went elsewhere. We're happy to see a wide range of viewpoints, but we want all of them to be expressed as politely as possible.

We realize that such a comment policy can never be evenly enforced, because we can't possibly monitor every comment equally well. Hundreds of comments are posted every day here, and we don't read them all. Those we read, we read with different degrees of attention, and in different moods. We try to be fair, but we make no promises.

And remember, it's a big Internet. If you think we were mistaken in removing your post (or, in extreme cases, in removing you) -- or if you prefer a more free-for-all approach -- there are surely plenty of ways you can still get your views out.