Social Network Analysis – measuring the immeasurable.

Social Network Analysis – measuring the immeasurable.

Euan Semple comments on the topic of social network analysis  (or measuring the immeasurable!).

Euan identifies two points that make him nevous about SNA:

"The first is because the activity is invariably couched in terms of
one group – managers, the business – mapping the relationships of
everyone else – the people prepared to open up and use the social tools
in the first place.

The second is because they seek to make explicit something that is
much better left implicit. We can all work out what the network is and
where the good guys are from the using the tools and inhabiting the
environments they create without having to have it drawn out for us.

If I felt that someone else was mapping my conversions and the
relationships they represented – and wasn’t prepared to have the same
done to them, I would soon stop talking.
"

As I have commented on the blog, I detect a form of management paranoia; they don’t really
understand what social networking is all about, they don’t want to dip
their toes into what they consider to be muddy water, yet at the same
they want to understand it in the only way that makes sense to them –
numbers and statistics.

What worries me are the conclusions they may draw from this imprecise and flawed method of evaluation!

Feel free to share...Share on FacebookShare on Google+Tweet about this on TwitterShare on LinkedInBuffer this pagePin on Pinterest

4 thoughts on “Social Network Analysis – measuring the immeasurable.

  1. Hullo Steven.

    I posted the following on Euan’s site:

    I don’t agree, primarily because you over-generalise.

    A network analysis need not be on one group. It could for example map artefacts, projects, intra and inter project communications, trust networks, and intra and inter organisational interactions. My own research – http://www.durantlaw.info/Doctorate – seeks to do exactly this. In other words it is recognising that at one time we can belong to many networks.

    Your second reason needs expansion. I have blogged on deconstructing complexity using network analysis techniques – see http://www.durantlaw.info/Deconstructing+Complexity . In this case it was about understanding a portfolio of projects. Network analysis techniques were very useful in making the implicit, and in this case the invisible, projects explicit.

    Now you might offer the criticism that I am not doing social network analysis. But remember projects are run by people who reside in organisations. Making the invisible or obscured explicit can, and in my example does, aid management and the knowledge workers. The techniques I am using come from social network analysis and in some circles might be called organisational network analysis ala Rob Cross.

    If this doesn’t convince you have a look at this website http://netmap.wordpress.com/case-studies/ which highlights the work of Eva Schiffer in Ghana.

    Social network analysis need not be evil, and is not the same as social computing.

    Regards

    Graham

  2. Graham,

    thanks for the comments, and the website links. Some interesting material on your web site which I’ve only had a chance to quickly scan. I’ll spend some more time reading the various papers when I have a chance.

    The one point we can agree on is that I over-generalise. The rest I guess we’ll have to agree to disagree. I should explain that my post was predicated on some practical experience with a particular client (who will remain nameless), which resonated with the points made by Euan. I’ll defer to your greater knowledge of SNA tools and techniques, but would caution on (a) any assumptions you’ve made in developing the relationship maps, and (b) the interpretation that other people may make when looking at the data. The deconstructing complexity models look very interesting, but the conclusions you are drawing assume that the raw data is 100% accurate and that these relationships do have the significance you have attached to them.

    Coming back to the real world (and the client I referred to above), data is being collected on the number of conversations between individuals, the number and type of contributions (e.g. wikis, blogs, documents etc.), and the total number of members in the network community. From this data, the client believes he can determine the success or otherwise of each community of practice in the network.

    The data and statistics look wonderful when presented in a report, with nicely coloured graphs showing trends. However, my argument is that this data (and the interpretation thereof) does not provide a measure of the success (or failure) of a project or community, since no-one is analysing the actual conversations taking place and how these conversations are affecting (a) the individuals and (b) the outcomes.

    Just because a blog attracts hundreds of comments does not mean the final outcome has been beneficial to the individuals involved, or has contributed in a positive way to the purpose of the community. But on the other hand, it might have. So, I think all I’m saying is that, by all means use data to get a better understanding of what is happening in a community, but beware of drawing conclusions as to success or failure of the community (or network or project) using this data. The fact that the conversations are occurring at all can be inferred as having some benefit, but some things are better left as implicit than explicit.

  3. Dear Steve

    Thankyou for an interesting and considered reply. I must say I was disappointed with Euan’s response, which I don’t even understand, but I digress.

    The case-study you provide appears to have some design problems, but that is difficult to confirm without seeing data, and understanding the communities. A problem with any network analysis is determining the unit of analysis. This unit then determines the system boundary, which in turn determines how missing data is or is not dealt with, which in turn determines how it is analysed and visualised.

    The examples I provided come from a closed system and data were collected via personal semi-structured interviews. Essentially it was a census methodology, which captured 98% of the people in the organisation. Even this approach has its problems, but less so if the assumptions are made explicit and caution is applied in presentation and interpretation – see this blog http://www.durantlaw.info/SOILs for a mistake I made.

    Coming back to your example, I agree your conclusions, particularly if the communities are open. If they are open then it is impossible to accurately know how many people actually use the community, and to know where to draw the boundary for analysis – at best all you get is an indicator of use. In this circumstance social network techniques might be valid if they are being used as a diagnostic tool to elicit further questions.

    You might be interested in this presentation http://www.durantlaw.info/sites/durantlaw.info/files/actKM%20Jan-Jun%202006%20Network%20v2.pdf on the actKM community of interest. Note in particular the caution slide (slide 15) and the comment about equilibrium. Network analysis treats the community as if it is in limbo and has reached a state of equilibrium – clearly this is never the case! Slides 37 to 43 illustrate this and show one way of dealing with it.

    Again thanks for your considered response. I do hope you enjoy my website and blog and learn something. I would be even happier if I convert you!

    Regards, Graham

Leave a Reply

Your email address will not be published. Required fields are marked *