Here is a link to the abstract of an interesting article by Fowler & Christakis, published in the British Medical Journal in December 2008:
I think it is a delightful statistical analysis of social networks, based on a cohort of about 5000 people from the Framingham Heart Study, followed over 20 years. This article should really be read in its entirety, in order to appreciate the sophistication of the techniques.
They showed that happiness "spreads" in a manner analogous to contagion. Having happy same-sex friends or neighbours who live nearby, increases one's likelihood of being, or becoming, happy. Interestingly, spouses and coworkers did not have a pronounced effect.
Also, the findings show that having "unhappy" friends does not cause a similar increase in likelihood of being or becoming "unhappy" -- it is happiness, not unhappiness, in the social network, which appears to "spread."
So the message here is not that people should avoid unhappy friends: in fact the message can be that befriending an unhappy person can be helpful not only to that unhappy individual, but to that unhappy person's social network.
There has been some criticism of the authors' techniques, but overall I find the analysis to be very thorough, imaginative, and fascinating.
Here are some practical applications suggested by these findings:
1) sharing positive emotions can have a substantial positive, lasting emotional impact on people near you, including friends and neighbours.
2) nurturing friendships with happier people who live close to you may help to improve subjective happiness
3) this does not mean that friendships with unhappy people have a negative emotional impact, unless all of your friendships are with unhappy people.
4) in the treatment of depression, consideration of the health of social networks can be very important. Here, the "quantity" of the extended social network is not relevant (so the number of "facebook friends" doesn't matter). Rather, the relevant effects are due to the characteristics of the close social network, of 2-6 people or so, particularly those who have close geographic proximity. As I look at the data, I see that having two "happy friends" has a significantly larger positive effect than having only one, but there was not much further effect from having more than two.
5) I have to wonder whether the value of group therapy for depression is diminished if all members of the group are severely depressed. I could see group therapy being much more effective if some of the members were in a recovered, or recovering, state. This reminds me of some of the research about social learning theory (see my previous post: http://garthkroeker.blogspot.com/2008/12/social-learning-therapy.html)
6) on a public health level, the expense involved in treating individual cases of depression should be considered not only on the basis of considering that individual's improved health, function, and well-being, but also on the basis of considering that individual's positive health impact on his or her social network.
7) There is individual variability in social extroversion, or social need. Some individuals prefer a very active social life, others prefer relative social isolation. Others desire social activity, but are isolated or socially anxious. Those who live in relative social isolation might still have a positive reciprocal experience of this social network effect, provided that relationships with people living nearby (such as next-door neighbours or family) are positive.
I should conclude that, despite the strength of the authors' analysis, involving a very large epidemiological cohort, my inferences and proposed applications mentioned above could only really be proven definitively through randomized prospective studies. Yet, such studies would be virtually impossible to do! I think some of the social psychology literature attempts to address this, but I think manages to do so only in a more limited and cross-sectional manner.