Online communities have different dynamics to face-to-face communities of practice. New research suggests that for online communities, behaviour spreads more quickly when the community is grouped into networks with overlapping connections than when left to develop more random and distant links.
If you are managing online communities then the research is worth a closer look as it may provide hints on how to design and support it so behaviour and influence can spread amongst the participants.
For communities of practice not online this clustering effect is not new. Research has been around since the 1970's which show the importance of bringing groups together geographically, thematically and/or by profession or person type. As humans we like to be with people similar to ourselves and are most influenced by those the same as us - if Joe can make the change then I will probably be able to as well.
There is a lot of buzz about social movements, mobilising and organising for change in healthcare. One of the success criteria for this work will be the depth to which those doing the organising as well as those participating understand network dynamics, and design accordingly. Random networks are less influential.