A beginner’s guide to social network analysis for social media and strat comm professors.
From a social network analysis fan with much to learn!
http://Netlytic.org
Overview of how to use the network visualization tool https://netlytic.org/home/?page_id=2
Tutorial for using Netlytic: https://youtu.be/F6scVtMGKFE
Additional Resources
♣ Basics of social network analysis slides
♣ Blog post “A Quick, Interactive Activity for Introducing the Concept of Digital Influencers”: http://mattkushin.com/2018/03/19/digital-influencers-easy-classroom-activity/
♣ Blog post detailing the below assignment: http://mattkushin.com/2017/04/24/teaching-basic-social-network-analysis-of-instagram-and-twitter-data-using-netlytic-org-post-4-of-4/
10. Directional
• An edge can be directional.
John
Sally
Examples:
1) Sally mentioned John in
an Instagram comment
2) Sally emailed John.
3) Sally Tweeted John.
4) Sally has a crush on
John (but he doesn’t
have one back).
11. Uni-Directional
• Or, an edge can be uni-directional.
John
Sally
Examples:
1) Sally and John know
each other.
2) Sally and John emailed
back and forth.
3) Sally said she has a
crush on John and John
said he has a crush on
Sally.
12. • Centrality is a measure of:
– “What characterizes an important node (e.g., person) in a
network?”
• Note: Many other stats used to analyze network. We are only scratching the
surface.
13. Degree Centrality
• Several types of centrality exist.
– We’ll look at: Degree centrality
• The # of ties a node has. This is the likelihood of
info flowing through this person.
– Note: Only considers walks of length of 1 (that is, from 1
node to the other)
#6 has 1 tie.
#2 has 3 ties.
14. 2 Types
• In-Degree: # of ties directed to the node.
• I.e., # of people mentioning that person in a Tweet.
– Indicates Popularity / prominence
15. Has 32 in-degree edges.
Why?
32 other Twitter accounts mentioned @ShepherdU in a Tweet.
Indicating it is popular – talked about often.
Network of hashtag: #ShepherdU
16. 2 Types
• Out-degree - # of ties the node directs to
others. – i.e., gregariousness.
All the people
I Tweet to, RT, or
mention.
17. Out-degree
• Significance?
– Indicates involved/care about topic
• Indicates they are active about a topic.
– Could be an influencer:
• May indicate influence (may not, as no one may
respond or pay attention.)
18. @ShepherdU has 0 out degree edges.
Why?
It hasn’t mentioned anyone in any Tweets.
Indicating, it doesn’t often a) Tweet at people or 2) RT others.
Primarily used to send out communication.
Network of hashtag: #ShepherdU
19. These are 3 OUTDEGREE connections because:
@khattakfugan sent the Tweet.
She mentions @UGRADPakistan (by Retweeting them) and @IREXintil and @ShepherdU
Network of hashtag: #ShepherdU
20. Other Stats
• Reciprocity
– Proportion of ties showing two-way
communication
• Calculated: Reciprocal ties / total number of ties
– High reciprocity (approaching 1) = lots of 2-
way convos
– Low (approaching 0) = one-way convos.
21. Other Key Stats
• Centralization – average degree
centrality of all nodes.
– High centralization (closer to 1) = a
few central people are dominating
information flow.
• Example: Graduation Ceremony
– Low (closer to 0) = info flows freely.
• Example: Party after graduation
ceremony.
22. Other Key Stats
• Diameter – longest distance between 2
network participants.
– Indicates: network’s size.
• Ex:
• For info to pass from John to Trevor, it has to
go through:
– John - > Sally - > Billy -> Robin -> Trevor
• Diameter = 5.
23. Other Key Stats
• Density – How closely connected is this network?
– Shows speed information can flow.
• Calculated: # of existing ties / # of possible ties.
– Dense networks: More close-knit
• Many people talking to one-another.
– Not dense:
• Few people talking to one another.
• A density score close to 1 is very dense.
• A density score close to 0 lacks density.
24. Example: Density
• This network has high density (but not
perfect density) because most people are
talking to each other.
25. Diameter – 3 people is farthest distance between communicators.
Centralization – Somewhat around ShepherdU – dominate as the topic of discussion.
Reciprocity – few people talking to one another.
Density low – Only a few people talk to each other in this network.
27. Analyzing Social Media
• Helps us understand interactions online.
Who Tweets, mentions, etc., whom.
28. Measuring Online Networks
• For hashtags or search terms:
– Outdegree: Who posts a lot?
• Represents gregariousness, and may indicate:
– Passion/interest
– Influence / Thought leadership
– In Degree: Who is popular in this network?
• Ex:
– High mentions
– People whose content is being RT’d.
– What clusters of people are talking to one
another, and what about?
29. • For mentions of your client
– What communities (who) is talking about your
client’s @username?
• i.e., what can you find out about who these groups
of interacting people are?
– Who all is mentioning (talking about) your
client the most?
• What are these folks saying?
30.
31. #CincoDeMayo Tweet by Bleacher Report mentioning @SHAQ
Why is Shaq
being talked
about during
Cinco De
Mayo!?
32.
33. Considering…
• We can see a lot of people are sharing
humorous or other #cindodemayo
related content.
• Not having conversations
• Density – SUPER Low
– Almost no one is connected to one-
another
• Reciprocity – very low
– Not a lot of people talking to one-another.
Mostly sharing, RTs each other.
• Centralization – very low
– Info flowing freely (not dominated by a
few).
Hinweis der Redaktion
-- so companies are trying to identify thought leaders, build relationships with them, and spread info
Bloggers, mommy bloggers, twitter chats, etc.
If this person talks about our product, they have higher:
Credibility
More likely to influence.
So we may want to target them!
It’s good to have a basic understanding of how social networks work.
I’m not an expert, more of a fan. And I thought it’d be valuable to discuss.