Social Network

Analysis in a Supply
Chain Context

Dr Leila Alinaghian

2021–2022 FT MSc

www.cranfield.ac.uk/som
An Overview of the Module

2
Intended Learning Outcomes
On successful completion of this module you should be able to:
Social
Network
1. Describe, discuss and apply the key concepts,
Analysis principles and metrics of social network analysis.

Internal 2. Apply social network analysis to quantify and assess

Supply network metrics and interpret their implications for
Chains
individuals and networks in internal supply chains.

External
3. Apply social network analysis to visualise, analyse
Supply
Chains and evaluate external supply chains.

Use a social network analysis software package to visualise,

analyse and evaluate internal and external supply chain networks.

3
Module Structure
Social Internal External
Network Supply Supply
Analysis chains Chains

Session I Session 3 Session 5

Social Network Social Network
An introduction
Analysis for Analysis for
to Social Network
internal supply external supply
Analysis I
chains chains I

Session 2 Session 4 Session 6

Visualizing and Social Network

An introduction analyzing
to Social Network Analysis for
networks using external supply
Analysis II Software chains II

4
Module Structure

Session 6
Session 5
Session 4
Session 3
Session 2
Session 1

Applications to Supply
Chain Management
Concepts, principles
and metrics

5
Teaching and Learning Activities
Sessions ILOs Teaching and learning activities

§ Lecture
1
ILO I § Mini class interactions, discussions and game
§ Node-level metrics group activity
2
§ Network-level metrics group activity

§ Lecture
3 ILO 2 § Mini class interactions and discussions
§ Industry guest speaker from Facebook

§ Computer lab workshop

4 ILO 4
§ Practice questions

5 § Lecture
§ Mini class interactions and discussions
ILO 3
§ Buzz group discussions
6
§ Case study* group discussion and reflection
*Pre-session reading 6
Session 3 Industry Guest Speaker

Roshanak Sadeghi

7
Session 4 – Network Analysis
Software

§ UCINET 6 for Windows is a downloadable software package for

the analysis of social network data. It was developed by Martin
Everett, Linton Freeman and Steve Borgatti.

§ NetDraw is bundled with UCINET and used to read and display

network visualisations.

§ UCINET / NetDraw URL: http://www.analytictech.com. UCINET

can be downloaded and used for free for 90 days/Netdraw is free.

§ Available on Cranfield PCs.

8
Session 4 – Accessing Network
Analysis Software

VMware Horizon

Accessing UCINET
See instructions on Canvas Session 4

9
Module Reading
Borgatti, S.P., Everett, M.G. and Johnson, J. C. (2013).
Analysing Social Networks. London, UK: SAGE Publications.

Session reading

10
Assessment

Individual assignment
100% of SNCC Credits; 1500 words

A small dataset consisting of supply relationships of a

company will be provided to qualitatively/quantitatively
analyse and discuss
§ The network, its composition and the pattern of
relationships
§ The operational and strategic performance
implications of the network structural properties

11
Session 1 and 2
An Introduction to Social
Network Analysis

Dr Leila Alinaghian

Feb 2022

www.cranfield.ac.uk/som
Session 1 and 2 Outline

What is Social Network Analysis and Why?

Key Network Elements and Principles

Node-level Metrics and Group activity

Network-level Metrics and Group activity

13
What is Social network
Analysis about?

14
Why Networks?

Battle of Lexington and Concord, 1775

15
Express riders knocked on acquaintances doors
to raise a military force from the civil population.
16
Why Networks?

17
Why Networks?

18 Source: Uzzi and Dunlap, 2005

Network Elements
and Principles

19
The Network View

20
Network View History
§ A multi-disciplinary method from the very beginning:
Mathematics Graph theory (1700), Psychology (1940s),
Anthropology (1950s and 1960s), Sociology (1970s),
Computer science (1980s) and Physics (2000s).

§ Network analysis as a field in its own right emerged during the

1960s and 1970s at Harvard Sociology department.

Growth in network studies Number%of%Papers%Published%in%

Business%and%Management%Journals%
160$
140$
120$
100$
Axis%Title%

80$
60$
40$
20$
0$
!20$1995$ 2000$ 2005$ 2010$ 2015$
Axis%Title%

21
Two Key Network Elements

A network is made up of nodes and

ties that connect these nodes.

Node Tie
Actor, Vertex Link, Edge

22
Your Interaction Network at Cranfield

What are the

ties that connect
individuals?

23
Tie Types

Mental relations

• A knows B
A • A likes/hates B
B
Formal relations

• A is a friend of B
• A teaches/supervises B

Interactions and flows

• A helps/ have lunch with B

• A transfers information to B

24 Source: Adapted from Borgatti and Halgin, 2011

Tie Direction

A B A B

Directional Non-directional
A tie from A to B does not A tie from A to B implies a
necessarily imply a tie of the tie of the same kind from B
same kind from B to A to A

25
A Network View to Supply Chains

Supply Chains

26
Nodes and Ties in Supply Chains

Nodes Ties

27
Nodes and Ties in Supply Chains

Mental relations

• A knows B
• A Competes with B

Formal relations

• A is supplier of B
• A is a partner of B

Interactions and flows

• A transfers materials/ information to B

• A and B are connected via financial flow
A B
28
Network Representations

Graph Matrix

29
Graph

30
Adjacency Matrix

James Ivy Harriet Fiona George Carol David Donna Andrea Bob
James 0 1 0 0 0 0 0 0 0 0
Ivy 1 0 1 0 0 0 0 0 0 0
Harriet 0 1 0 1 1 0 0 0 0 0
Fiona 0 0 1 0 1 1 1 0 1 0
George 0 0 1 1 0 0 1 1 0 1
Carol 0 0 0 1 0 0 1 0 1 0
David 0 0 0 1 1 1 0 1 1 1
Donna 0 0 0 0 1 0 1 0 0 1
Andrea 0 0 0 1 0 1 1 0 0 1
Bob 0 0 0 0 1 0 1 1 1 0

31
The Key Network Principles

1 Connections

2 Patterns of connections

32 Source: Granovetter, 1973; 1985

1. Connections

Kinship Co-membership/Friendship

33 Source: Uzzi and Dunlap, 2005

2. Patterns of Connections

34
2. Patterns of Connections

35
Network Analysis

36 36
Network Metrics

37
Two Levels of Analysis

Node-level
Three metrics

Network-level
Three metrics

38
Node-level
Metrics

39
Centrality Metrics

The relative importance of a node due to

its structural position in the network.

40
Centrality Metrics – Example

Who do you think is important in

this network?

Two nodes are connected if they

41
regularly talk to each other
Centrality Metrics

1. Degree Centrality

2. Closeness Centrality

3. Betweenness Centrality

42
Degree Centrality

Degree centrality is measured by the number of

direct ties to a node.

43
Degree Centrality Calculation

Row sums di= Σ xij

j

James Ivy Harriet Fiona George Carol David Donna Andrea Bob Degree
James 0 1 0 0 0 0 0 0 0 0 1
Ivy 1 0 1 0 0 0 0 0 0 0 2
Harriet 0 1 0 1 1 0 0 0 0 0 3
Fiona 0 0 1 0 1 1 1 0 1 0 5
George 0 0 1 1 0 0 1 1 0 1 5
Carol 0 0 0 1 0 0 1 0 1 0 3
David 0 0 0 1 1 1 0 1 1 1 6
Donna 0 0 0 0 1 0 1 0 0 1 3
Andrea 0 0 0 1 0 1 1 0 0 1 4
Bob 0 0 0 0 1 0 1 1 1 0 4
44
Degree Centrality Calculation
Divide the degree centrality by the maximum possible
connections in a graph with n node (i.e. n – 1)

Degree centrality Normalised degree centrality

James 1 1/(10-1) = 0.11
Ivy 2 2/9 = 0.22
Harriet 3 3/9 = 0.33
Fiona 5 5/9 = 0.55
George 5 5/9 = 0.55
Carol 3 3/9 = 0.33
David 6 6/9 = 0.66
Donna 3 3/9 = 0.33
Andrea 4 4/9 = 0.44
Bob 4 4/9 = 0.44
45
Degree Centrality Implications

§ Index of exposure to what is

flowing through the network (e.g.
A
virus, information)
§ Resource access advantages
(validity and diversity) and risks
(operational burden)
§ Influence and power on
decisions or behaviour of others
in the network

46
Degree Centrality Implications –
Example

47 Source: Krackhardt, 1992

Degree Centrality Implications –
Activity
§ A government grant of £100k will be awarded to two
companies who mange to seal a partnership deal to
work on a project.
§ The companies below have the opportunity to form a
partnership (only) with those to which they are
connected with a line.
§ They need to negotiate the split of partnership grant
(£100k) to seal a deal. No agreement means no grant!

48 48
Degree Centrality Implications –
Activity

Who do you think would have more

power in negotiating a deal?

49
Degree Centrality Implications –
Activity

Power is a function of having multiple

potential partners (Degree centrality
as the underlying power principle)

✅ ✅ ✅

50 50
Degree Centrality Implications –
Activity

Power is also a function of connection to

weak actors.

✅ ✅

A node’s bargaining position is a

consequence not only of its alternatives but
also the (lack of) alternatives of its
alternative!
51 Source: Cook, 1983; Scott and Carrington, 2011
Degree Centrality – Directed Network

In-Degree centrality Out-Degree centrality

the number of incoming the number of outgoing
ties a node has. ties a node has.

52
Degree Centrality – Directed Network
Example
A is connected to B if A shares a secret with B
Carol

Andrea Fiona

David Harriet Ivy James

Bob George

Donna

In-degree centrality: How many people share their secrets with the node

Out-degree centrality: How many people the node shares their secrets with
53
Degree Centrality Calculation –
Directed Network
Receiver
James Ivy Harriet Fiona George Carol David Donna Andrea Bob Out-Degree
Sender

James 0 1 0 0 0 0 0 0 0 0 1
Ivy 0 0 0 0 0 0 0 0 0 0 0
Harriet 0 1 0 0 0 0 0 0 0 0 1
Fiona 0 0 1 0 1 1 0 0 1 0 4
George 0 0 1 0 0 0 1 1 0 1 4
Carol 0 0 0 0 0 0 0 0 1 0 1
David 0 0 0 1 0 1 0 0 1 0 3
Donna 0 0 0 0 0 0 1 0 0 0 1
Andrea 0 0 0 0 0 0 0 0 0 1 1
Bob 0 0 0 0 0 0 1 1 0 0 2
In-degree 0 2 2 1 1 2 3 2 3 2

54
Closeness Centrality

Closeness centrality is the sum of the length of the

shortest path connecting a node to all others.

55
Closeness Centrality

In an interview in 1994,
Kevin Bacon
commented that he had
worked with
everybody in
Hollywood or
someone who’s
worked with them!
http://oracleofbacon.org

56
Closeness Centrality Calculation

Length of the shortest path Sum of distances to

connecting James and Fiona other nodes

James Ivy Harriet Fiona George Carol David Donna Andrea Bob Closeness
James 0 1 2 3 3 4 4 4 4 4 29
Ivy 1 0 1 2 2 3 3 3 3 3 21
Harriet 2 1 0 1 1 2 2 2 2 2 15
Fiona 3 2 1 0 1 1 1 2 1 2 14
George 3 2 1 1 0 2 1 1 2 1 14
Carol 4 3 2 1 2 0 1 2 1 2 18
David 4 3 2 1 1 1 0 1 1 1 15
Donna 4 3 2 2 1 2 1 0 2 1 18
Andrea 4 3 2 1 2 1 1 2 0 1 17
Bob 4 3 2 2 1 2 1 1 1 0 17
57
Closeness Centrality Calculation
Divide the closeness centrality by longest
path in a graph with n nodes (i.e. n – 1)

Closeness Normalised closeness Inverse Normalised

centrality centrality closeness centrality
James 29 29 / 9 = 3.22 1/ 3.22 = 0.31
Ivy 21 21 / 9 = 2.33 1 / 2.33 = 0.43
Harriet 15 15 / 9 = 1.66 1/ 1.66 = 0.60
Fiona 14 14 / 9 = 1.55 1/ 1.55 = 0.64
George 14 14 / 9 = 1.55 1/ 1.55 = 0.64
Carol 18 18 /9 = 2 1/ 2 = 0.50
David 15 15 / 9 = 1.66 1/ 1.66 = 0.60
Donna 18 18 / 9 = 2 1/ 2 = 0.50
Andrea 17 17 / 9 = 1.88 1/ 1.88 = 0.53
Bob 17 17 / 9 = 1.88 1/ 1.88 = 0.53
58
Closeness Centrality Implications

§ Index of expected time until arrival for

given node of whatever is flowing through
the network
§ Resource access advantages (Higher
fidelity and timeliness of transfer)
§ Autonomy (More freedom from others’
influence and their controlling actions)
hence higher capacity for independent
actions)

59
Closeness Centrality – Directed
Network Example
A is connected to B if A shares a secret with B
Carol

Andrea Fiona

David Harriet Ivy James

Bob George

Donna

In-Closeness: How fast a node hears others’ secrets

Out-Closeness: How fast a node’s

60
secrets spread (reach others)
Closeness Centrality Calculation –
Directed Network
Undefined
Length of the shortest path Length of the shortest path distances are
connecting Ivy to James connecting James to Ivy
replaced with n
Out-
James Ivy Harriet Fiona George Carol David Donna Andrea Bob
Closeness
James 0 1 10 10 10 10 10 10 10 10 81
Ivy 10 0 10 10 10 10 10 10 10 10 90
Harriet 10 1 0 10 10 10 10 10 10 10 81
Fiona 10 2 1 0 1 1 2 2 1 2 22
George 10 2 1 2 0 2 1 1 2 1 22
Carol 10 6 5 4 5 0 3 3 1 2 39
David 10 3 2 1 2 1 0 3 1 2 25
Donna 10 4 3 2 3 2 1 0 2 3 30
Andrea 10 5 4 3 4 3 2 2 0 1 34
Bob 10 4 3 2 3 2 1 1 2 0 28
In-Closeness 90 28 39 44 48 41 40 42 39 41

61
Betweenness Centrality

Betweenness centrality measures how often a

node connects nodes that would otherwise be
disconnected.

62
Betweenness Centrality Calculation

The number of times a node acts as a bridge

along the shortest path between two other nodes.

B (i) = Σ g(i)jk /gjk

The number of The number of shortest

shortest paths that path connecting any two
actor i is on nodes

63
Betweenness Centrality Implications

§ Index of power as a result of

§ Controlling the flow
§ Liaising otherwise separate parts of the network
§ Gatekeeping/Brokering roles
§ Vulnerability to disruption
64
Betweenness Centrality Implications
– Example
Russian trade routes in the 12th 13th century

Moscow high in betweenness centrality

Betweenness centrality matters when ties are hard to create

65 Source: Pitts, 1979

Node-level Metrics
Activity

66
Global Operations and Supply Chain
function Interaction Network
The network mapping below reveals an informal structure – who interacts
with whom to get the work done (e.g. information, expertise, advice) – within
the Global Operations and Supply Chain function of a multinational company

Shapiro Stock
Cole
Cohen Jones

Kelly
Hughes Andrews

Williams
Smith
Miller
Hussain Taylor
Cross

Moore Ramirez 67 Bell Sen

Group Activity

1. Construct an adjacency matrix representing the interaction

network (an Excel template can be found on canvas – Session 1
and 2).
2. Compute degree centrality scores of each member of
global operations and supply chain function. Identify the
most/least central individual based on degree centrality
scores.
3. Compute closeness centrality scores of Cole, Andrews
and Moore. Among the three, who is the closeness to
everyone in the network?
4. Discuss the implications of these network positions.

68
Network-level
Metrics

69
Two Levels of Analysis

Node-level
Three Metrics

Network-level
Three Metrics

70
Network-level Metrics

1. Density

2. Sub-groups

3. Distribution of Degree Centralities

71
Density

The number of ties that are present divided by

the total number of possible ties

A B

72
Density Calculation Count 1s

Number of ties that are present = 36 = 0.4

Total number of possible ties 90
assuming no ties to self n x (n –1)

James Ivy Harriet Fiona George Carol David Donna Andrea Bob
James 0 1 0 0 0 0 0 0 0 0
Ivy 1 0 1 0 0 0 0 0 0 0
Harriet 0 1 0 1 1 0 0 0 0 0
Fiona 0 0 1 0 1 1 1 0 1 0
George 0 0 1 1 0 0 1 1 0 1
Carol 0 0 0 1 0 0 1 0 1 0
David 0 0 0 1 1 1 0 1 1 1
Donna 0 0 0 0 1 0 1 0 0 1
Andrea 0 0 0 1 0 1 1 0 0 1
Bob 0 0 0 0 1 0 1 1 1 0
73
Density Implications

§ Cohesion

§ Knitting together/ Trust

§ Redundancy of information

74
Sub-Groups (Communities/Clusters)

A 11
B 11

12 10

12 10
2 6 2 6
9 9

1 4 5 8 1 4 5 8

3 7 3 7
Density = 0.31 Density = 0.31

75
Sub-Groups Implications

Groups of actors who interact with each other to such an

extent that they could be considered to be separate entity.

§ A tendency to have
homogeneous
opinions/common traits.
§ They tend to share norms
and exert peer pressure on
their members to conform to
these norms.
§ Each member of group would
know more or less what the
other members know.
76
Sub-Groups and Bridges

A tie between two nodes, removal of which would

break up a network into disconnected parts

§ More likely that bridge

Jack% actors possess non-
redundant
Jenny% information and thus
John% pass on new
James%
information and
insight.

77
Distribution of Degrees

Distribution of degrees is the

(frequency) distribution of nodes’ degree
centrality

7
6
5
4
3
2
1
0
David Fiona George Andrea Bob Harriet Carol Donna Ivy James

78
Distribution of Degrees Example 1
A Centralised
B
One or a few nodes with an
E exceptionally high degree
centrality and very many nodes
D
with low degree centrality.
C

Degree Centrality 5
A 1 4
3
B 1
2
C 1
1
D 1
0
E 4 E D C B A

79
Centralisated Networks

Centralised: Non centralised:

Trade monopoly Distributed trade

80 80
Distribution of Degrees Example 2
A B
Block-diagonal

Most nodes have similar

E C degrees.

Degree Centrality 3
A 2
2
B 2
C 2 1
D 2 0
E 2 A B C D E

81
Distribution of Degrees Example 3
B D G F Scale-free

H A few nodes have disproportionately

A
C I many connections, while most of
E the other nodes have only a few
connections. The distribution follows
Degree Centrality a power-law or Pareto distribution.
A 8
B 3 9

C 2 8
7
D 2 6
E 1 5

F 1 4
3
G 1 2
H 1 1
0
I 1
A B C D E F G H I

82
Degree Distribution Implications –
Communication Network Game
Groups of 5 + 1 observer (Groups of 6)

§ Each person is given 5 letters chosen from a set of

six (e.g., A, B, C, D, E, F) in a coloured envelope.
§ One of the six letters is in everyone’s possession.
The objective is to discover which letter they all
have in common.
§ You need to communicate with each other to solve
the puzzle as fast as possible.
§ The task is regarded as completed when every
member of the five-person team knows the answer.
83 Source: Adapted from The Bavelas-Leavitt Experiment
Communication Rules

§ A colour is assigned to each member

§ The dots at the back show the members
of the team to whom you are connected

§ Use post-it notes to communicate

§ One note for each communication
§ NO TALKING!
§ Don’t reveal the dots/puzzle

84 84
Team Performance

The team performance is measured based on

§ Whether you solved the puzzle

§ How fast

§ How many notes you used

85 85
Note to the Observers

§ Start the timer when the game begins

§ Stop the timer when every member of the five-

person team announces the answer

§ Count the number of written notes that have been

used (one note for each communication)
§ Check the answer and report the time and
number of notes to the instructor

86 86
Distribution of Degree Implications

5 4
4 3
3
3 2
2 2
1 1 1
0 0 0

87
Metrics – Valued
Networks

88
Tie Value
5

1
A B

How often
Daily = 5
Weekly = 4
Monthly = 3
Quarterly = 2
Yearly = 1

89
Degree Centrality – Valued Network
Example
Dichotomize network ties
using a cut off point to
4 1 convert the network into a
5 non-valued network.
1
1
5 5 5 5
1
3
1 1
3 4
2
The cut off point will be
determined based on the
5 4 context and the nature of
connections.

90
Degree Centrality – Valued Network
Example
Carol For example:
3 and above = Connected
Below 3 = Not connected
Andrea Fiona

David Harriet Ivy James

Bob George

Donna

91
Network-level Metrics
Activity

92
Global Operations and Supply Chain
function Interaction Network
Further study has revealed the frequency of interactions within the
informal network of Global Operations and Supply Chain function.

4
How often Shapiro Stock
5 2
Daily = 5 Cole
4
Cohen 1 Jones
Weekly = 4
4 5 4 4
Monthly = 3 3 3
3 Kelly 4 3
Quarterly = 2
4 3 Hughes Andrews Williams
Yearly = 1
3
4
Smith 1 Miller

2 4 Hussain 4 3 Taylor 5
Cross

Moore Ramirez 93 Bell Sen

Group Activity

1. Construct an adjacency matrix representing the

interaction network (an Excel template can be found on
canvas – Session 1 and 2).
2. Dichotomize network ties using a cut off point of 3 (i.e.
Consider below 3 as not connected)
3. Calculate network density
4. Develop network degree distribution and determine the
type of network
5. Discuss the implications of these network metrics.

94
Post-Session

§ Indicative reading

Next Session

§ Session 3 – 11th Feb

§ Session 4 – 14th Feb
VMware Horizon

95
Thank you
Dr Leila Alinaghian
MSc, MPhil (Cantab), PhD(Cantab), FHEA
Cranfield School of Management
College Road, Cranfield, MK43 0AL

[email protected]

https://www.linkedin.com/in/leila-alinaghian-phd-fhea-
79602a56/
https://www.cranfield.ac.uk/som/people/dr-leila-
alinaghian-784115

www.cranfield.ac.uk/som