Yes, i am aware of my elongated absence from this blog. And i have to plead…well i don’t know what my excuse is, so i’ll just say “howdy all” instead.
A recent article from PLoS One by Christakis and Fowler seems to be getting much less publicity than did their series of papers from the Framingham Heart Study. We’ve talked briefly about some contentions with that work here before.* The thing is that, by my reading, this newest paper is much more compelling and interesting than even the sum of their previous networks-based research, imo.
The new paper is an elegant finding – in essence that we would be better equipped for predicting flu epidemics if our estimates were based on surveillance of the nominated friends of a random sample, than we would get from tracking the random sample itself. It is firmly rooted in previous social networks research and a core idea/finding therein – Felds’ 1991 (gated) “Why your friends have more friends than you do.” And perhaps more importantly, is very clearly and simply potentially useful.
_________
*Incidentally, while i have been somewhat critical of their FHS work, i ended up trying out their Connected book for my current Intro Sociology class. i’ll have to get back to you on how effective a book it is for those purposes.
Permutations 
hey Jimi, that is a neat article. Thanks for the pointer. I do wonder about the practicality of the proposed sampling procedure. If the goal is simply to have respondents of high degree and centrality, perhaps we can do better than to draw one random sample and then ask them to nominate friends. If the goal is to sample the people most likely to get sick first, perhaps we should survey health care workers, or teachers, or someone else likely to be exposed early.
[…] Seasonally Adjusted By Adam Permutations points out an elegant paper from Christakis and Fowler (gloriously open access). They exploit a […]