Graphical Social Network Analysis of NHS Change Day 2014
1. #nhschangeday 3rd to 5th July 2014
through the lens of graphical Social Network Analysis using
NodeXL
A very basic introduction with example graphs (and lots of caveats)
Mark Outhwaite
Outhentics Consulting
Helping find direction through complex times
Phone: +44 (0)2032399438
Mobile: +44 (0)7768131770
Skype: markout
Website: www.outhentics.com
More about me: http://www.linkedin.com/in/markatouthentics
Follow me on Twitter: @mark_outhwaite
Blog: http://outhentics.blogspot.co.uk/
To schedule meetings with me go to: http://doodle.com/outhentics
2. Introduction
• This presentation provides a brief and very simplified introduction to the techniques used to analyse the Twitter
activity taking place during the NHSCHANGEDAY online event which took place on 4th July 2014 – Social Network
Analysis (SNA)
• Graphical SNA allows you to explore and discover the connections and patterns between people whether they
are in organisations, talking across organisational boundaries or in social settings. Who are the ‘hubs’ and
‘bridges’, the experts and the lone voices? How do the actual patterns of conversation actually contrast with the
expected patterns? How do networks evolve over time and what is their response to change.
• An oversimplified introduction to the concepts involved is included but I strongly recommend that you review
these two sources for a much better introduction and an excellent illustration as I am just a beginner.
– http://www.slideshare.net/Marc_A_Smith/2013-nodexl-social-media-network-analysis
– http://www.pewresearch.org/fact-tank/2014/02/20/the-six-types-of-twitter-conversations/
• The NodeXL software used is Open Source and freely available for download for use with Excel 2007 onwards
– http://nodexl.codeplex.com/
• NodeXL can take data directly from a wide variety of sources including Twitter, Facebook, YouTube email,
Exchange servers and a variety of other sources as well as data captured through other routes such as part of
organisational SNA surveys (eg – name the 4 people whose advice you most value)
• The data source in this case is a Twitter search using the NodeXL import function
– Search was for all Tweets containing #nhschangeday ‘since:20014-07-03’ and was conducted on 06 July
2014
– You should be aware that due to Twitter API limitations on external searches that it is possible that not all
Tweets meeting those conditions were retrieved
– Twitter searches are normally only possible for a maximum of the previous 7 days activity. So if you want to
monitor and map a conversation over an extended period you will have to schedule regular searches and
downloads
• If you want a copy of the data for your own use in NodeXL or you use another package (eg GEPHI) please let me
know.
• If anybody feels the need for an opportunity for a further discussion about this presentation or the other uses of
graphical SNA then please let me know – I am happy to host an online live webinar (GoToMeeting). If you are an
expert in the area and would be prepared to talk on the webinar even better.
• Continue the conversation on #nhschangeday
3. Connections and patterns in Twitter
• ‘I Tweet therefore I am’ – unless you tweet, retweet,
reply or mention we do not know you exist.
• An individual, for example ‘@mark_outhwaite’, is
described as a node
• If an individual is mentioned in a Tweet by another
person or vice versa then a relationship or ‘edge’ is
created between the pair.
• The relationship is ‘directional’ depending on who
mentions who in their message.
• The ‘edge’ or connection gets stronger/closer as there
is more activity between the nodes.
• If I tweet and there are no forwards, mentions, replies,
quotes or favourites then there is no relationship
• A cohesive group or sub-group is created when there is
a well connected group – lots of ‘edges’ connecting the
members of the group and many of those being strong
connections.
• Centrality – the number of direct (normally incoming)
connections an individual has to other members of the
group. So if @mark_outhwaite receives lots of
mentions from the same people then he will appear
closer to the centre of the group.
• Betweenness – a measure of how much of a bridge you
act as between other people. For example you may be a
bridge from your group to one or more people in
another group, or between smaller clusters within your
group
4. SNAPSHOTS FROM THE ANALYSIS
These are just some static snapshots from the data.
NodeXL has dynamic filters which allow you to explore the data in more detail live on screen.
5. The full picture of #nhschangeday
This is the full picture
of all activity during
the period.
Groups are created
based on the intensity
and type of
relationships between
participants.
The words most
mentioned by the
groups are shown in
each group box.
The edges are
‘bundled’ tightly to
keep the map tidier.
6. Who are the main bridges between groups?
We use a measure of
‘betweenness
centrality’ to identify
individuals who are
most likely to be
acting as bridges or
information brokers
between the different
groups.
7. Who has more than 2000 followers?
Here we pull out
those people with the
most followers. In this
case 2000 or more
followers.
This means that it is
these people whose
tweets have the most
reach.
In the case of
nhschangeday there
will be many people
who read the tweets
but did not retweet,
reply or mention
8. Who mentions a lot of other people in their tweets
>=10 other people ‘namechecked’ in their tweets
There will be some
people who mention
a lot of other people
during the period.
They may be doing a
lot of retweeting of
other people’s tweets
or holding a lot of
conversations
(mentions) with
others in their groups.
9. Who gets mentioned a lot in other people’s tweets
>= 10 mentions by other people
Some people get
mentioned a lot in
other people’s tweets
– normally because
they get retweeted a
lot.