Using Social Network Analysis to Evaluate Communities of Practice: A Methodological Journey

1
Using Social Network Analysis to Evaluate
Communities of Practice A Methodological Journey

• Kathryn Everest
• Director, Strategy Consulting
• Jive Software

2
What is Social Network Analysis (SNA)?

• A set of methods and metrics that shows how
  entities interact
• the current patterns of communication,
  information-sharing, decision-making and
  innovation within a particular organization or
  group
• Focuses on three elements
• A group of entities (in this case researchers and
  institutions) and the roles they play
• The relationship between the entities
• And attributes might be present that are creating
  a bias in the relationship of the entities
• When applied to organizations, it can be referred
  to as ONA Organizational Network Analysis
• The outcome of an SNA
• helps us to see where collaboration is breaking
  down, where talent and expertise could be better
  leveraged, where information is getting bogged
  down or where opportunities for innovation are
  being lost
• SNA data
• gives us the picture we need to create a set of
  remedial actions for individuals, influencers and
  stakeholder to improve productivity, efficiency
  and innovation

3
The entities, the relationship between entities,
and the attributes of the entities are depicted
in SNA maps

• Social Entities
• People (Researchers, Health Professionals,
  Experts)
• Resources (Collaborative spaces, Information
  repositories)
• Relationship
• Information sharing
• Advice
• Collaboration
• Trust
• Awareness
• Access
• Attributes
• Contributing factors such as location, tenure,
  specialty, language, funding, culture

4
SNA also provides insights about roles within the
network, as well as metrics that allow us to
measure where we are and where we want to be

• Roles
• Central roles those who many people go to (or
  could be bottleneck)
• Peripheral people - goes to others but no-one
  goes to them
• Boundary Spanners connects different groups
• Brokers connects many people
• Isolate - not connected

Network Measures Density 6 Cohesion
4 Centrality 6 Central People DPa(34), CR(29),
BB(20), MDo(19), DPr(17)

• Density --- Robustness of network (group measure)
• Number of connections that exist in the group out
  of 100 possible in that network
• General level of linkage. More points connected
  means quicker and more accurate information flow
• Cohesion --- Ease with which a network can
  connect
• Distance is the shortest path between two people
• Aggregate measure at network level reflects
  average distance between people
• Centrality --- Identifies influential people
  (individual measure)
• Number of direct connections that individuals
  have with others in group
• Individuals who have more ties to others may be
  in more advantaged positions they may have
  access to more of the information or knowledge in
  the network

5
Why do an SNA?
6
Develop before and after measurements (e.g.
What impact has your community had on the
network?)
1 Area 1 (Circle) 2 Area 2 (Square) 3 Area 3
(Up Triangle) 4 Area 4 (Plus) 5 Area 5 (Down
Triangle)
1 Geo 1 (Red) 2 Geo 2 (Pink) 3 Geo 3
(Black) 4 Geo 4 (Dark Blue) 5 Geo 5 (Grey)
Awareness Network - Density 4.83 861 Ties
7
The SNA can provide understand the potential
impact losing core researchers to a network
Sample hypothesis Without core XX researchers,
collaboration supporting XXX research would be
severely affected

• Research collaboration with core researchers

• Research collaboration without core researchers

Example
8
Identify issues and opportunities in the network
N Ind/Pri Hosp Rehab Col/Uni Other
Individual/Private 14 24 8 31 16 20
Hospitals 9 6 17 19 15 6
Rehab Facility 10 29 20 27 29 36
College/University 8 11 14 16 20 14
Other 7 22 6 36 30 43
Note Removed Gov/Agency and K-12
9
Identify Brokers

• Consultant
• Name
• Name
• Name
• Name
• Name
• Name
• Representative
• Name
• Name
• Name
• Name
• Name
• Name
• Name
• Name
• Name

• Liaison
• Name
• Name
• Name
• Name
• Name
• Name
• Gatekeeper
• Name
• Name
• Name
• Name
• Name
• Name
• Name
• Name
• Name

Liaison
Consultant
Representative
Gatekeeper
10
Measure and identify issues in individual networks
Out In
Mean 6.425 6.425
Std Dev. 4.53 3.33
Min 0 1
Max 21 17
11
Understand the makeup of your network
Authority
High Involvement
13
11
In-degree
Apprentice
Low Involvement
10
16
Out-degree
Notes Removed all nodes with a 0
out-degree Higher in-degree and out-degree than
the mean
Framework Source Steve Borgatti
12
Provide a perspective of a person and their
network
Description Awareness Awareness describes a
person is aware of the knowledge and skills of
others in the network. In-degree reflects the
number of people who indicated that they believe
they are aware of your knowledge and skills.
Out-degree reflects the number of people you
selected as being aware of their knowledge and
skills. The mean represents the mean or median
number of the network. SME SME describes
whether a person will call upon another person
when looking for expertise related to TOC topics.
In-degree reflects the number of people who
indicated that they would call upon you for
expertise. Out-degree reflects the number of
people you selected you would call upon. The
mean represents the mean or median number of the
network. Betweenness Betweenness describes how
often you fall between two people in the network.
The Mean of the network is 51.4 with a maximum
value of 533. Comm Comm (Communications)
relates to how many people communicate with each
other on a daily, weekly (once or more per
week), weekly (once a week or less), or monthly
(once a month or less). In-degree reflects the
number of people who indicated they communicate
with you, and out-degree reflects the people you
selected. The mean represents the mean or median
number of the network. Network Composition by
Location This is analysis looks at location of
the people in your network. XX represents the
percentage of people in your network who are
located in XX, XXX represents the percentage who
are located in XXX, and Other represents
locations other than XX and XXX. H is the value
which represents the heterogeneity of your
network. H is a number between 0 and 1 with 1
representing a maximum value. The average result
was .46.
Centrality
In-degree Out-degree Mean
Aware 11
SME 10
Betweenness 51.4
Comm-Daily 1.2
Comm-Weekly 1.8
Comm-Weekly 3.3
Comm-Monthly 6.5
Network Composition by Location
XX XXX Other H

13
Example
Centrality
In-degree Out-degree Mean
Aware 12 11 11
SME 8 11 10
Betweenness 25 54.4
Comm-Daily 2 2 1.2
Comm-Weekly 0 1 1.8
Comm-Weekly 2 4 3.3
Comm-Monthly 8 1 6.5
Network Composition by Location
XX XXX Other H
50 31 19 .617
14
How do you do an SNA?

• Gather attribute/demographic and relationship
  data
• Getting the data from public sources / mining
  data
• Survey (bounded and unbounded)
• Analyze the data
• Pick a tool (I use UCINet, but there are others
  including NodeXL, Pajek, InFlow just to name a
  few)
• And there is a lot to both

15
I use Optimice http//www.onasurveys.com
16
To use it, you need to register for an Optimice
account

• There is no charge to create and account, develop
  and administer a survey
• A free account however only allows you to
  download a subset of the data
• An account is 75USD per month or 599 per year
• You only need to have a valid account when you
  download data
• You can conduct as many surveys as you want, with
  as many people as you want
• You can allow your account to lapse and your data
  is maintained
• They will expunge data on request

17
1. Log in (after you create your account)
18
2. Create a new survey
19
3. About the survey
20
3. Contd
21
Create an introductory email to attach the survey
(should reference another communication which
puts the project in context)

• Dear Name
• What is this project?
• Why are you asking these questions?
• What will happen to the results?
• Will people see how I answered?
• When will I hear more?
• Signed by,
• To answer the survey please click on this link
  URL

22
4. About Respondents (traditional survey and
attribute questions)
23
Respondent List
24
Creating the list of participants
Notes First two columns MUST BE Name and
Email Create as a CSV file to upload to the
survey tool No spaces in the headings
25
5. About Relationship (Network Questions)
Note the more names you have, the fewer
questions you should consider. Make your
questions count!!
Checkbox
26
Choice Across and Down
27
Matrix Question
28
Testing the survey
29
Monitor / Test
30
Tracking Progress Two ways
?
?
31
When the data is collected, download the data
32
Analyze Data
33
What can you do? Way too much to cover, so be
focused about what you need
Demonstration