Sometimes it seems to me as if a lot of people love quantitative methods as the easy way out, “the figures show…” and then you can’t argue any more. However, if you are really strict in using quantitative network analysis approaches, there are a lot of methodological pre-conditions:
- All actors need to be of the same kind, that is either individuals or organisations or departments, etc. You can’t do a proper quantitative analysis of a map that mixes individuals and organisations. One colleague of mine adds: All actors need to have a phone number – and while you can also include people who don’t have a phone, this excludes functional groups such as “THE farmers”
- You need to know everyone and agree on who is a member of the network. While you might be the one who draws the boundaries (or they might be drawn for you, e.g. by organizational boundaries) it has to be clear, who is in and who is out, because
- You have to have a complete picture, either by asking all network members about their links to all other members. Or by choosing links that you can observe, such as email communication.
If you don’t fulfill these preconditions, you’ll have great distortions in your quantification: “bigger” actors (e.g. organizations) will have more links than smaller actors (individuals), those you involve in your study by interviewing them will have higher centrality levels than those you don’t and what on earth do you do with the fact that people disagree about network boundaries?
But, a lot of networks have unclear boundaries, are messy, have individual and organizational actors and you’ll rarely be able to include each and every member in your interviews. So I see that the longer I to this, the more I tend to focus on the power of visualization, as I want to stay as close to the (messy) reality as possible, instead of adding so many assumptions and arbitrary boundaries that my participants won’t recognize their own networks afterward. I still use quantitative network measures, but I don’t fool myself into believing that they can prove anything (in the kind of networks that I draw) – they only point me towards interesting areas of the network that need closer examination.