Following Up: Using Twitter to understand event connections

In a previous post we used Twitter data to see if people attending the LeWeb'12 London tech conference were more likely to have connected or exchanged information due to the event.

This post discusses further insights from our Twitter research on event analytics, looking in particular at the strength of connections made at the event.

The Twitter followship

One idea underlying our analysis is that people's actions in Twitter somehow reflect their 'real-world' networking (in the sense that they indicate whether they have become aware of each other or exchanged information). If John and Juan attend LeWeb12 and a Twitter following relationship between them appears, perhaps this means they met there. Chalk one up for LeWeb12!

Alas, things aren't that simple. Before even getting into what would have happened if John and Juan hadn't attended LeWeb'12 London (the issue of additionality - see the end of this blog), there is the issue that by contrast to other social media platforms such as Facebook and LinkedIn, Twitter allows "unreciprocated" follow connections: Generally, Juan can follow John's tweets without John's permission. 

In our analysis of LeWeb'12 London we therefore look at three possible types of new Twitter connection between those attending the event, illustrated in the diagram below.

1. One-way (unreciprocated) following connections where, say, Juan starts following Kevin Rose of Google Ventures who spoke at LeWeb'12 London and has 1,4 million Twitter followers...but Kevin Rose does not follow him back.

2. New reciprocal following connection where Juan and John, who were unconnected in Twitter before LeWeb'12 London start following each other after the event.

3. Consolidated reciprocal following connections where a previous one-way following relationship becomes mutual e.g. Juan was following John on Twitter before LeWeb'12 London, and John starts following Juan back after the event.

We are particularly interested in reciprocal connections because:

1. They tell us about the 'intensity' of the tie created between people. It seems fair to assume that a one-way link between two individuals is weaker than a reciprocal one (in the latter case, the two individuals involved are at the very least aware of each other).

2. With completely 'new' reciprocal connections, it may be reasonable to assume that the event was more important in making the two individuals meet.

Analysing our LeWeb'12 London data shows that 1,520 new Twitter follow connections were created between participants. Of these:

  • 70% (1,072) involved unreciprocated follows (around half of these were audience members following speakers on Twitter)
  • 21% (314) of the follows went towards forming 157 completely new reciprocated connections
  • 9%  (134) involved the 'consolidation' of 134 existing one-way following connections to form 134 new reciprocal connections

The visualisation below shows the network of 291 new reciprocal connections created at LeWeb'12 London). The purple lines are between speakers following one another on Twitter, the gold lines are following relationships between audience members, the intermediate coloured lines are reciprocal following relationships between speakers and audience members.

 

If we wanted to find out where LeWeb'12 London is helping to build new relationships potentially leading to future exchanges of ideas and collaboration, the links between the people below are probably a good place to start. 

 

Talking publicly...

Connections aside, we can use the public interactions between individuals through tweets to look at whether they are communicating with each other and what about. 

Let's have a look at @mentions (tweets from participants where they mention other participants' Twitter ids) in the run-up to LeWeb'12 London, during the conference, and afterwards.

This is shown in the figure below for individuals that had already been connected to other attendees before LeWeb'12 London (Labelled "Pre-LeWeb"), and those who created a new connection at the event (Labelled "New").

Unsurprisingly, the month in which LeWeb'12 London took place (June) displays a spike in mentioning between participants, with a total of 4,103 interactions. When we look at the mentions between those who only connected at the event (red columns), we see 585 interactions.  Communication between newly connected people continues in the longer term, with a further 480 interactions between people who connected at LeWeb'12 London in the three months after the event.

...And saying something

We can also use the content of the tweets between participants who connected at LeWeb'12 London to study what they were discussing. By way of overview, the word cloud below shows those words that were most often mentioned. 

We see that 'meet' and 'meeting' appear often in this word cloud- together, they are mentioned 49 times in tweets between newly connected participants. These words are perhaps suggestive of something more than just an online connection having been made between conference participants.

 

Using social data to measure the effects of events on innovation networks

We'll shortly be publishing a research report with new results from our analysis of LeWeb'12 London, including how 'close' new connections at the event were on Twitter before the event took place. This provides some further indicative evidence of how likely the connections were to have formed even in the absence of the conference.

We're also getting the next stage of our festival analytics research programme underway, where we want to explore the question of 'network additionality' more deeply. Specifically, we have issued an Invitation to Tender for research proposals that formally incorporate control groups into their analysis of the effects of events on networking - in order to establish the effects on expanded networks of events over and above connections that would have formed anyway. We are also interested in proposals that can link the changes in networks resulting  from events to innovation and economic outcomes. Can we get a measurable grasp on 'real-world' outcomes that are generated as a consequence of the expanded networks?.

If you have any questions about this work, drop us a line at [email protected] and [email protected]  or alternatively @johnardavies and @JMateosGarcia

Author

John Davies

John Davies

John Davies

Principal Data Scientist, Data Analytics Practice

John was a data scientist focusing on the digital and creative economy. He was interested in the interface of economics, digital technology and data.

View profile
Juan Mateos-Garcia

Juan Mateos-Garcia

Juan Mateos-Garcia

Director of Data Analytics Practice

Juan Mateos-Garcia was the Director of Data Analytics at Nesta.

View profile