Archive for the ‘Social Network Analysis’ Category

ARIN6901: “Big Society” in the UK: tapping weak ties and the ‘influentials’ for results

October 9, 2010

In a situation remarkably similar to our own Australian federal election, the UK population were also on tenterhooks for weeks in May of this year as they waited for an outcome to their General Election. Eventually, a coalition was formed by David Cameron’s Tory party and Nick Clegg’s Liberal Democrats. One of the central policies of the Tories is “Big Society” – misunderstood during the election but still central and still in the planning stages. It is explained here by Tory team member Ian Birrell:

At its core, the big society is an attempt to connect the civic institutions that lie between the individual and the state – and these range from the family and neighbourhood to churches, charities, libraries, local schools and hospitals. It is born out of recognition that our centralised state has become too big, too bureaucratic and just too distant to support many of those most in need of help, and that it deters people from playing a more active role in public life.

In political terms, this means passing power to the lowest level possible: radical public service reform, so that schools, social services, planning and even prisons are more responsive to the needs of those using them; and social action, to encourage more people to play a role in society. Not just charities, but neighbourhood groups, workers’ co-operatives, social enterprises and, yes, businesses. (Birrell, 2010)

So, the general idea is to allow the execution of power to filter down to the lower, more connected levels of society, rather than just being something that is executed “up there” in the upper echelons.

Naturally, the question is: how can this be implemented?

It comes down to investigating the existing network structure of different communities, pinpointing those considered ‘opinion leaders’ or as ‘influentials’ in sociological terms or network terms, to create effective change, rather than maintaining superstructural agencies that are far removed from the reality and nuances in these varied communities.

Respected thinktank, the RSA (the Royal Society for the encouragement of Arts, Manufactures and Commerce) has published a report entitled Connected Communities based on the findings from a year of research. The report makes detailed recommendations around Big Society, encouraging the use of social network analysis in addition to older approaches to public policy in order to achieve effective community regeneration (RSA, 2010).

The report identifies what are known as ‘linchpins’ in different communities that took part in the research, recommending that these people can play pivotal roles in new policy development and implentation. The linchpins may already be involved in their own community-building projects or may just be well connected enough to allow for the advancement of new policies and projects.  ‘Linchpins’tend to be well-connected, likely to be based on many ‘weak ties’ in Granovetter’s terminology,  and as defined by the RSA, appearing to fit neatly with the ‘opinion leader’ archetype originally proposed by Katz and Lazarsfeld in 1955 and described by Duncan Watts and Peter Dodds as “individuals who are highly informed, respected or simply ‘connected’” (2007: 442). In an article looking at how diffusion can occur much faster when initiated by opinion leaders, Valente and Davis identify many studies which indicate the importance of these opinion leaders in utilising the power of interpersonal contacts in influencing adoption behaviour.

So, there is a rich history of studying diffusion after the fact – it will be very interesting to see how successful a program will be that has been established from the start with a network theory approach to diffusion in a community setting.



Birrell, I. (2010) ‘Big society’? Let me explain
Accessed from:
Last accessed: 9/10/10

Rowson, J., Broome, S. and Jones, A. (2010) Report: Connected Communities: How social networks power and sustain the Big Society
Accessed from:
Last accessed: 9/10/10

Valente, T.W. and Davis, R.L. (1999) ‘Accelerating the Diffusion of Innovations Using Opinion Leaders’ in Annals of the American Academy of Political and Social Science, Vol. 566, pp. 55-67

Watts, D. and Dodds, P. (2007) ‘Influentials, Networks, and Public Opinion Formation’ in Journal of Consumer Research, vol. 34

Williams, R. (2010) ‘Big society’ facilitators are found within communities
Accessed from:
Last accessed: 9/10/10

ARIN6901: Using the “Friendship paradox” in Social networks to predict epidemics

September 17, 2010

The metaphor of ideas, innovations or movements diffusing throughout a network as a type of disease, as seen by use of descriptive words such as epidemics or infectious as Duncan Watts identifies, has been criticised by Duncan Watts in the midst of such a metaphor assuming normative status in the wider population.  Watts asserts that the spread of a disease contagion in a network works on a different mechanism to that of a social contagion and explains the logic of this in his 2003 book Six Degrees: The Science of a connected age.

In a recent development in this debate, specifically looking at the relationship between disease outbreak and social networks, Harvard medical professor and sociologist Nicholas Christakis and James Fowler, professor of medical genetics and political science at the University of California, have put forward a proposition that the so-called “Friendship paradox” holds the key to predicting the movement of diseases throughout a network, with potentially enough forewarning to enable prevention of an epidemic. Interestingly, and this is where he differs in his views from Duncan Watts, he believes this same rule can be applied more widely, to areas such as product adoption, and to the spread of social norms and information. It is the Friendship paradox application to diffusion that is the new discovery.

Since mapping an entire network may be extremely expensive or unethical, if at all possible, Christakis and Fowler decided to apply the Friendship paradox in order to predict the spread of the flu amongst Harvard students. The Friendship paradox holds that if you randomly choose a person from a group, and then look at that person’s friends, those friends will have more friends themselves than that original person does. In short, according to Christakis, “your friends have more friends than you do” (TED Talks, 2010).

The methodology involved the selection of an original random group of 319 undergrads who each nominated up the three friends, which produced a second group of 425 friends. The second group were found to be more central in the connections amongst Harvard students owing to the Friendship paradox. By monitoring flu infections in a set period, they found that on average, the second group developed the flu approximately two weeks prior to the random group using one method of detection, and a full 46 days prior to the epidemic peak by using another method (University of California, 2010).

Phase Diagram Harvard Study

Phase Diagram Harvard Study showing the point of diversion between the two groups

So, what this means is that the spread of a disease can be tracked by locating the better connected members of a network (the second group) and then identifying who they are connected to as these people will be the ones who can benefit from this advanced warning. Christakis in his TED Talks video does qualify the actual number of days of advance warning depends on a number of factors including the nature of the pathogen or even the specific structure of the human network.

Support for the study results has come from John Glasser, a mathematical epidemiologist at the Centres for Disease Control in the U.S., who said

“This study may be unique in demonstrating that social position affects one’s risk of acquiring disease. Consequently, epidemiologists and social scientists are modeling networks to evaluate novel disease surveillance and infection control strategies.” (University of California, 2010)

Christakis and Fowler conclude that this type of network analysis is a more effective means for disease control than more traditional methods which are usually one to two weeks behind in tracking the outbreak rather than in advance as the Friendship paradox method appears to be capable of. Applying the Friendship paradox method more widely could be done by marrying the method with Google Trends search data (that is, with flu-related search terms) to map the emergence of a flu epidemic (Christakis and Fowler, 2010).

My critique of Christakis’ TED Talks presentation is that he glosses over the many ethical and logistical complications of applying the Friendship paradox in an effort to garner interest in it. However, it is important that this information is promoted and TED Talks is an effective way of spreading it to a diverse audience.

In attempting to make sense of the findings, I have concluded that in the case of disease epidemics, the key is that disease spread is due to the quantity of links or ties that someone has and a smaller number of degrees on average, whereas with social innovation or information that Granovetter discusses, that is due more so to the diversity of ties than the quantity of ties (however, that being said, diversity of ties can also sometimes equate to a high number of ties).

Clearly there is much opportunity for research to build on these results, however the challenge will be to repeat the results on a larger scale and in other areas apart from a flu outbreak in a relatively closed community such as Harvard. Application in the real world with tangible positive results is the aim.


Anonymous (2010) ‘Infectious personalities’
The Economist. London: May 15, 2010. Vol. 395, Iss. 8682; p. 89

Christakis, N.A. and  Fowler, J.H. (2010) ‘Social Network Sensors for Early Detection of Contagious Outbreaks’ in PLoS ONE, 5(9): e12948 DOI:
Accessed from:
Last accessed: 17 September 2010

Granovetter, M. (1973)’The strength of weak ties’ in American Journal of Sociology, Vol. 78. No. 6, pp. 1360-1380.

TED Talks (2010) Nicholas Christakis: How social networks predict epidemics
Accessed from:
Last accessed: 17 September 2010

University of California – San Diego (2010, September 16). ‘Friendship paradox’ may help predict spread of infectious disease in ScienceDaily.
Accessed from:
Last accessed September 17, 2010

Watts, D. (2003) Six Degrees: The science of a connected age,
New York and London, Norton.