I recently mapped the Twitter traffic at the 2016 annual conference of the Canadian Evaluation Society. This was presented live several times during the conference, as the network evolved. The idea was to encourage participants to add to the “buzz” of the conference, and to inform each other of important ideas and upcoming events. In addition, the CES will now have a database including the text of all conference tweets, replies and mentions, which should help with future conference planning by identifying what generated interest and comment. Here is a summary.
The network – distinct structure and key players
We observed a network comprising both a large number of casual participants and a smaller set of active communicators who helped tie the conversations together across the conference. In this image, the most frequent communicators are represented by the largest points.
This social network covers all Twitter messages referencing @CES_SCE_2016 and #EvalC2016 from May 31 through June 8, thus including the run up to the event and final comments. 213 people sent or were referenced in Twitter messages, for 2,093 unique connections between them. The network is of course centred on the main conference address.
A large inner ring of participants were relatively closely connected to the hub, with multiple messages and interconnections, while a separate subject were more peripherally connected. A smaller subset of more active Twitter users are grouped together with multiple interconnections. While highly interconnected among themselves, these highly active participants helped to span the network, forming clusters of interconnection with many more Infrequent communicators.
Out degree and in degree 1 – a small group sent the most messages
“Out Degree is the social network analysis term for the number of messages an individual has sent out. “Messages” include tweets, replies and mentions of others on Twitter. Here, we can see that the main conference address has, of course, the highest out degree, but also that there is a relatively small group who have sent a large portion of the messages.
Out degree and in degree 2 – activity and influence are mostly correlated, with some exceptions
The graphic on the left shows the degree of correlation between sending and receiving messages. Those participants receiving significantly more than sending are highlighted blue, with the opposite marked in red.
The more people get replies to their messages or are mentioned in other messages, the higher is their in degree. The table lens at the right shows that those sending the most messages (the first column) are often but not necessarily those receiving he most replies or mentions (the second column). This can point to a difference between activity and influence where the difference is pronounced. In this case, it is mostly the keynote speakers and several institutions that have received significantly more messages than they sent. Understanding these patterns as the conference unfolds can help organizers to encourage dialogue.
Connectors and clusters 1 – some people are “bridges”
In many social networks, there are players who connect many others, who would otherwise be unconnected, or at least much less so. In technical terms, this is called a high “betweenness centrality”. There is clearly a set of such individuals here, shown by the larger points. For the mostpart, these correspond to high sending-activity participants, although not entirely. Several players here are clearly positioned as bridges between outlying individuals and the more central group. Identifying your bridges helps you to make sure more participants can be engaged.
Connectors and clusters 2 – most people are part of a cluster
“Clusters” in social networks are basically groups of people who are more densely connected to each other than they are with the rest of the network. This, in the case of a something like a conference, probably denotes underlying commonalities of interest, as well as a propensity to reply to and mention other relatively active network members. Clues to these commonalities can help to ensure high levels of engagement and discussion as the event progresses.
How different people spanned the network
Among the interesting things you see in an analysis of this kind is:
1) how certain individuals connect otherwise more peripheral players to the “action”;
2) how some parts of the network are better connected than others; and
3) how a set of individuals, together, could connect most or all of a network in ongoing communications.
Watching for this and identifying key connectors can help organizers ensure that social media talk will draw people in to key events and discuss ideas, creating more “buzz” and participation.