Social Network Analysis of Survey Participants
I did a little analysis of our survey participants. The following picture shows the social network between the people who answered our questions. Red nodes are users who retweeted my announcement. The big node in the middle is me (I did not take part in the survey but added me for the network analysis). There is a little bit wrong influence in this analysis because I added some of the participants to my network after the survey because their updates were protected and we needed access or just because they were interesting. You can clearly see that all but two nodes are connected to each other. The survey was promoted 99% via Twitter so this is a logical consequence. The two lonely nodes may come from this posting on the Microblogging Conference blog.
I used a great piece of software called Network Workbench (nwb) for the visualisation. The connection data was retrieved via the Twitter API and imported into nwb as CSV file (one row of the file looks like “usera, userb, true” that means that “usera” is following “userb”).
Here are some more:
Tags: microblogging, sna, twitsurvey, twitter
September 14th, 2009 at 00:35
This is great. I would be extremely grateful if you could go through the following, step by step
‘The connection data was retrieved via the Twitter API and imported into nwb as CSV file’
September 21st, 2009 at 09:46
Hi Josh, thanks for your interest. In short, the steps to get Twitter connections into nwb were:
a) I had a list of my survey participants (Twitter names)
b) my PHP script calls the Twitter API for each of the participants and gets their followed users and followers (note that Twitter knows these two kinds of relationships but nwb only knows one; so you have to decide how you want to transfer Twitter relationsships to a nwb relationship, for example you could say “a nwb relationship only exists if users follow each other in both directions” -> this is something you can achieve with a little bit of PHP
c) for each participant (lets say “Tom”) we now have a list (array) of friends and our PHP script writes for each friend in a text file (for example network.csv):
Tom, Friend1, true
Tom, Friend 2, true
and so on. This can be read as “Tom and Friend1 are friends”. In pseudo code:
for each participant as p {
get friends into array
for each item in friend array as i {
write p.”, “.i["name"].”, true” in CSV file
}}
d) Finally, network.csv could be imported into nwb (I cannot provide you with a step-by-step guide for that since I do not have the software on my pc any longer, sorry, but it was quite easy)
Hope that helps.