Complexity and Social Networks Blog

Hi David

I am one of the many consultants using SNA in their work. But possibly in a different way than most. In the field of evaluation, one challenge is how to represent the intended activities and expected outcomes of a project (development aid projects, in my area of work). Traditionally this has been through the use of linear logic models, which take the shape of “If we do A, and assumption B holds, then C will happen, and if assumption D holds, then E will happen...� The Logical Framework has been one device used to capture these summary representations of projects, and these in turn have been the focus of evaluation activity: did the plan work, and if not why not…

Linear models have the virtue of being like simple story lines, easy to communicate and to understand. But they have their corresponding disadvantages, of ignoring all the complexity in the world, especially that arising from the existence of multiple actors working in parallel, with only local and partial connectedness with each other.

I have started to introduce actor networks as an alternate means of representing project intentions and results, both through some introductory papers at conferences ( http://www.mande.co.uk/docs/nape.pdf ) and through selective introduction of network representations of projects, in my consultancy work. Also via some workshops on innovations in development project evaluation methods ( http://www.mande.co.uk/docs/MMAPP.ppt ) I have also set up a networks section of my website on monitoring and evaluation methods, at http://www.mande.co.uk/networks.htm (None of this information is proprietary).

While I think that the field of evaluation has much to learn from SNA, I think the opposite may also be the case. In evaluation work a lot of attention has been paid to participatory methods of evaluation, where the subjects of evaluation are active participants in the evaluation. There has also been a substantial influence from the field of anthropology/sociology, such that evaluators are increasing concerned with people’s awareness and interpretations of “the facts�, and not just the facts, as though they spoke for themselves. So, I would be interested to see what work has been done with SNA, and complexity studies, on participatory analyses of networks, where network members contribute their own interpretations of what the network in question is all about. In my experience, even relatively small and simple networks become very interesting when the actors intentions about how those networks should work are brought into the picture. I am also interested in the use of two-mode networks (actors x policies, actors x events, etc), because typically participating individuals have much less complete knowledge of these networks than actor x actor networks, and so even the simple act of surveying and feeding back results can be a significant development intervention in its own right.

Perhaps I should stop at this point, but not before quoting Thomas Schelling, as my reference point: "In my own thinking they have never been separate. Motivation for the purer theory came almost exclusively from preoccupation (and fascination with) 'applied' problems; and the clarification of theoretical ideas was absolutely dependent on an identification of live examples" (Schelling, T.O., 1980:vi)

Hi David

I am one of the many consultants using SNA in their work. But possibly in a different way than most. In the field of evaluation, one challenge is how to represent the intended activities and expected outcomes of a project (development aid projects, in my area of work). Traditionally this has been through the use of linear logic models, which take the shape of “If we do A, and assumption B holds, then C will happen, and if assumption D holds, then E will happen...‿ The Logical Framework has been one device used to capture these summary representations of projects, and these in turn have been the focus of evaluation activity: did the plan work, and if not why not…

Linear models have the virtue of being like simple story lines, easy to communicate and to understand. But they have their corresponding disadvantages, of ignoring all the complexity in the world, especially that arising from the existence of multiple actors working in parallel, with only local and partial connectedness with each other.

I have started to introduce actor networks as an alternate means of representing project intentions and results, both through some introductory papers at conferences ( http://www.mande.co.uk/docs/nape.pdf ) and through selective introduction of network representations of projects, in my consultancy work. Also via some workshops on innovations in development project evaluation methods ( http://www.mande.co.uk/docs/MMAPP.ppt ) I have also set up a networks section of my website on monitoring and evaluation methods, at http://www.mande.co.uk/networks.htm (None of this information is proprietary).

While I think that the field of evaluation has much to learn from SNA, I think the opposite may also be the case. In evaluation work a lot of attention has been paid to participatory methods of evaluation, where the subjects of evaluation are active participants in the evaluation. There has also been a substantial influence from the field of anthropology/sociology, such that evaluators are increasing concerned with people’s awareness and interpretations of “the facts‿, and not just the facts, as though they spoke for themselves. So, I would be interested to see what work has been done with SNA, and complexity studies, on participatory analyses of networks, where network members contribute their own interpretations of what the network in question is all about. In my experience, even relatively small and simple networks become very interesting when the actors intentions about how those networks should work are brought into the picture. I am also interested in the use of two-mode networks (actors x policies, actors x events, etc), because typically participating individuals have much less complete knowledge of these networks than actor x actor networks, and so even the simple act of surveying and feeding back results can be a significant development intervention in its own right.

Perhaps I should stop at this point, but not before quoting Thomas Schelling, as my reference point: "In my own thinking they have never been separate. Motivation for the purer theory came almost exclusively from preoccupation (and fascination with) 'applied' problems; and the clarification of theoretical ideas was absolutely dependent on an identification of live examples" (Schelling, T.O., 1980:vi)

Schelling, I would note, wrote this while he was at the Kennedy School.

I want to highlight a particular methodological issue raised here, regarding the debriefing process in SNA. There are actually multiple issues here: (1) how one handles the human subjects issues around revealing the positions of particular people, and (2) the potentially very useful feedback one gets from the subjects regarding why the network looks the way it does. A small illustration: the last time I taught SNA I did a social network survey of the class, which, it turns out, was broken into several small cliques. In discussion, it emerged that this structure simply reflected the "recruitment" process into the class, where one person noticed the course, and persuaded a few friends to check it out, and they jointly decided to take the class. It would have been difficult/impossible to figure this out without the discussion. Further, in a context where one was trying to produce rigorous research based on these data, it would have been very easy to make incorrect inferences about them.

Your example from your SNA class is a good one, and it connects to one of my worries. Listing in to some of the mailing lists, it does appear that there is a lot of SNA analysis which could be described as "de-contexualised number crunching". I wonder whether this is partly due to the increased availability of network analysis software that enables a wide array of quantitative analyses, which might be easier and quicker to do than going out and talking to people. Somewhat ironical, given that SNA is about contacts between people. On a more constructive note, I would be interested to hear more about other people's experiences, like yours above, with engaging people as analysts of network data (in which they are the subjects). I think this is different to work on "cognitive social networks", but corect me if I am wrong.