Social Network Analysis - Revealing network patterns

[Lucia Nass]

Last year I was introduced to Organisational Network Analysis (ONA), or SNA - Social Network Analysis as it is more widely known. If you are new to SNA, I can recommend this 5 minute intro to Social Network Analysis by the Stockholm Resilience Center. A whole new world opened for me; I never imagined that you could actually quantify relationships among organisations, and do it with a very short questionnaire, simple to administer. It seemed a bit absurd at first. Quantifying social relationships, so you could monitor change and growth in networks, really? However, as soon as I saw a few simple examples, and understood the basic metrics in ONA, I began to see the multiple uses I could put it to.

I am currently putting the final touches to my first SNA report. You may read about the background of that initiative in my Learning among organisations blogpost. I have learned a lot going through this SNA process, first time without the seasoned SNA experts doing it for me. Fortunately I had asked Ed Canela, to be on my side (yes, the same Ed from some of the blogs here on Beads), a bit less green than myself in SNA matters. And Joitske Hulsebosch shared some experiences, as well as her SNA delicious list, from which I can highly recomend the practical Networkmapping manual by Louise Clark. Even then, it's not easy cutting through the SNA jargon. Density, centrality, eigenvector, degrees-in and degrees-out. Oooof... Don't worry, it's really not that complicated. Well, actually it is complicated, but I can maybe help to make it more accessible by using simple language.

 

A SNA visual: What am I looking at?

A SNA visual represents the network of your survey respondents, or part thereof. This visual shows just part of a network of almost 300 organisations.To be able to read this visual, you must know the following:

• Each round dot (jargon: node) represents a person, namely one of the survey respondents
• Each square dot represents an organization identified by the survey respondents (i.e. in the past 6 months I worked with…)
• All dots have a label with the name of the organization. One organization may have several dots in the visual. For instance in figure 1, the round dots CARE1 and CARE2 are people (respondents to the survey), and the square dot CARE represents the organization identified by at least one respondent.
• Arrows between the dots represent a relationship between the given organizations. The arrow shows who identified the relationship, i.e. “I worked with…” – the arrow points from the respondent (“I”) to the organization that was identified. For instance the illustration shows that IUCN was identified by 7 respondents to the survey.
• The placement of the dots has an approximate significance – unless otherwise indicated, those on the outer rim of the network are less connected than those in the center.
• The color of the dots is explained in the legend to the visual. For instance, in the illustration the light green color identifies the Ecosystems and Livelihoods Adaptation Network | ELAN and its partners, namely CARE, IIED, IUCN, and WWF. The dark green color identifies all the other organizations that have been identified through the survey as having a relationship with ELAN and its partners (jargon: ego-networks combined).

The interpretation of visuals is best done by the organizations themselves. However some broad conclusions can generally be drawn. For instance in the illustration, we immediately see something interesting. CARE respondents (light green circles) report many linkages to others, while WWF and IUCN respondents (light green circles) report only a few links to others. Yet when we look at the links others report to the ELAN partners, the situation is different. IUCN, WWF and IIED (light green squares) get mentioned more often than CARE. This is a good illustration of the different roles organisations may play in a network. The role of CARE is most likely that of network promoter. IUCN and WWF are more likely to be resource hubs, two very different functions.

Through SNA we can identify a variety of roles that organisations may have in a network. I focus on three network roles that are most relevant for overall network functioning, they are the Broker, the Resource Hub and the Promoter (I won't bother you with the SNA metrics jargon):

• Brokers are organisations which can directly reach many others in the network. They play a vital role in knowledge communication processes. Their position in the network is such that if organisations want the shortest routes to others in the network, they will cross the path of the Brokers. Brokers have the potential to move knowledge around in the network. Brokers that are unaware of their position in a network can unintentionally block others from accessing knowledge. Question for Brokers: In what ways can your organisation better harness the potential to move knowledge around in the network?
• Resource Hubs are organisations to which many others report linkages. As they are known by many others, Resource hubs have the potential to make resources widely available in the network. If you know the 'resource hubs' you can for instance discuss: Can we collaborate with these organisations to make our knowledge available?
• Promoters are organisations that report links to many other organisations. They are particularly influential in the network. Promoters have the potential to help others in the network to increase their network.

What I discovered in working with SNA is that researchers should not draw too many conclusions from the metrics data and visuals. The trick is to find and reveal patterns, and then to define questions that help SNA participants make sense of what they see. Unfortunately, the Bangkok based conference where I had planned this 'sense making' was cancelled due to the floods. I can however tell you from other expriences that the strength of SNA visuals is maybe not so much that it reveals new insights to the key players in the network, -- they usually know already "who's who". The main benefit I have seen is in helping those who want to improve their position in the network to understand what is currently going on, and strategise, build new relationships with that understanding. That's good news for marginalised groups and smaller, less powerfull organisations.

A few notes on key learned points regarding the design, and administration  of the SNA survey, and running the actual analysis:

• We used Survey Monkey, but needed an excel data typist to set up the network file. There are specialist programs that actually run surveys as well as do the SNA analysis, but I have yet to find an affordable, user friendly one.
• Although the survey was short, i.e. could be filled in 5 minutes including open questions, we still collected too much information. For instance, we asked for sex of the respondents (for gender based analysis), and country headquarters of the organisation (for south-north dimensions in the analysis), but that proved not very relevant to the SNA. It did help us to say something about the respondents though.
• The conference organisers offered to support distribition of the survey, because they had an e-mail list of all 800 registered participants. It was a great offer, but the first survey-link they send was in a mail with a lots of other infomation, too far ahead of the conference. The intial response was low. The second time, based on my suggestions, the link appeared prominently in mail with "5 things to do before you come to Bangkok". This generated 100 responses in the next 3 days, up to the point that the conference had to be cancelled.
• Respondents needed to identify organisations "you worked with in the past 6 months" (maximum 10). About 20% of the respondents provided names that could not be used in the analysis, because they were too general (e.g. rural community, or line ministry). I must find out how other SNA researchers overcome such problems. If you know many of the organisations in a network already, it is better to list them, and ask respondents to indetify those they have worked with from the list, and to add any other organisations that are misisng from the list.
• We used UCINET software for the actual Social Network Analysis (generates metrics and visuals), which is free for the first 60 days. In this case, where we only had one round of surveys, with respondents each indentifying max 10 other organisations, it was very interesting to look at the 'ego-networks': all the relationships identified, viewed from one orgaisation. I also liked the 'sub-groups' analysis, which shows groups of loosely connected, more closely connected and tightle connected organisations. What struck me was the dominance of UN organisatons in the tightly networked sub-group, and the dominance of bilaterals (e.g. GIZ, DFID, JICA) in medium connected group. They really have their own 'circles'! Amazing...