Saturday, November 7, 2015

Dynamic Spread of Happiness in a Large Social Network

By: James H. Fowler and Nicholas A Christakis


Happiness is a fundamental object of human existence to the point that the World Health Organization emphasizes happiness as a component of health.  A large variety of voluntary and involuntary factors determines a person’s happiness.  Emotional states can be “caught” or transferred directly from one individual to another in many different ways.  This study looked at how happiness spreads over long periods through a social network in both direct relationships and indirect ones (i.e. friends of friends) as well as whether there were geographical or temporal constraints on this spread through the network.

Looking at 5,124 individuals – deemed “egos” – each connected to others via friendship, family, marriage, coworkers, or geographically.  Each person who had a relationship with an ego – deemed an “alter” – brought the totally participants up to 12,067 studied individuals.  There were 53,228 observed social ties between 5,124 egos and alters, averaging 10.4 ties to family, friends, and coworkers.  Though there were substantial variations from person to person, with some egos ranging from several people with no friends to one person who was nominated as a friend by eight different participants.  The 3 types of defined friendship are the “ego perceived friend,” the “alter perceived friend,” and the “mutual friend.”  The first being the alter thinks of the ego as a direct friend, the second being the ego thinks an alter as a friend, and the last being both the ego and the alter consider themselves friends.

The study defines happiness as a positive emotion used a conventional measure using 4 items from the CES-D (Center for Epidemiological Studies depression scale) in which people were asked how often they experienced certain feelings during the previous week.  Happiness was considered a perfect score on each of the 4 items, but close to perfect results was also recognized and weighed.  The social network consists of the individuals (nodes) and relationships (social ties) between them.  Once all the nodes and ties are known, a picture can be drawn of the network to discern position within it. 

Examination of the social network indicated that happy people tend to be connected to one another.  The below figure mapped the largest connected network in 1996 and 2000 based on a restricted set of ties among siblings, spouses, and friends.  Each node is colored according to a person’s happiness on a spectrum from blue (unhappy) to green (intermediate) to yellow (happy).

The clusters of happy and unhappy people seen in the network are significantly larger and calculable.  The relationship of an ego and alter’s happiness is calculable by measuring the probability that an ego is happy when an alter is happy and comparing it to the same probability in a simulated network.  This data also suggests a relation between network centrality and happiness as people at the core of their local networks are more likely to be happy.  It should be noted that happiness itself does not increase one’s centrality at subsequent time points.


While there are many determinants of happiness, whether an individual is happy also depends on whether others in the individual’s social network are happy.  Happiness is not just a function of individual experience or choice, but is also a property of groups of people.  Social network analysis makes it possible to visualize the changes and ripple effects through a network to generate large scale structures and identify clusters within a network.  I found these results interesting considering that happiness generally requires close physical proximity to spread and that the effect decays over time.  Nonetheless, the analysis does not allow for the actual cause of happiness (or other emotions) to be identified.  It also recognizes that while the person to person effects tend to be string, they decay throughout the network so a particular mood or behavior ripple is very limited. 

BMJ 2008; 337 doi: 
(Published 05 December 2008)

Cite this as: BMJ 2008;337:a2338


  1. I find this article interesting solely on the basis that the authors attempted to measure such a subjective topic as happiness. The author’s research would have had more depth if an objective measurement was used in conjunction with happiness. For example, a Princeton study indicates happiness rises with income, peaking at $75,000 (,9171,2019628,00.html). Applying an objective control measure like income, may have revealed new information concerning how happy people connect with other happy people, and added critical thought to their disappointing key finding - “Examination of the social network indicated that happy people tend to be connected to one another”. In essence, I am disappointed that the author’s research did not infer any worthwhile information.

    1. I was attracted to this article because I wanted to see how simple and valid the method could be while handling subjectivity. I agree with your insight and I too would have liked to have seen a more objective measurement to found the study on. The key finding is clearly fairly intuitive, though, looking at the network map, I found the shifts in clusters between participants a notable documentation of this inherent quality.

  2. Katie, it is really hard to show tangible findings about an abstract topic like happiness. Thus, this article seems really interesting to me. I am curios to learn if the article shows any correlation between the number of ties and happiness just like "the social tie theory" which argues that an individual who has high numbers of social ties is less likely to commit a crime.

    1. The article does comment on the correlation between the number of ties (centrality) and happiness. The research suggests that people at the core of their local networks seem more likely to be happy, while those on the periphery seem more likely to be unhappy. Additionally, it is not only the number of direct ties, but also the number of indirect ties that influence predicted future happiness. More simply, the better connected are one’s friends and family, the more likely one will attain happiness in the future. Though as I mentioned in the article discussion, network centrality leads to happiness rather than the other way around.

  3. With regarding to the findings, can we say if the person who is the center node of a social network is unhappy, then his/her connections are likely to be unhappy. Do the actions of center node form the social network or social networks' dispositions guide the center node (with the light of this study shed)?

  4. I think this article has some really interesting applications, particularly when attempting to change the mindset of a group of individuals. If happiness can cause a ripple effect within a network, could a government employ a similar analytic-based strategy to realign the people's view? The only aspect I think this paper fell short on was Eigenvector centrality, which would have identified who the most influential nodes are. Theoretically, if you could make the influential nodes happy (or any other emotion), that would be passed on to more people and the ripple would likely not die as quickly.

  5. It's interesting to see that Princeton ties higher income to happiness. Some of the richest countries in the world are not the happiest, whereas some of the poorer less developed countries surpass them on a happiness scale. I, too, would be interested to see how they measured the happiness in this study.