Social networks and their economics influencing consumer choicedr soc

Social
Networks
and their
Economics
Influencing
Consumer Choice
Daniel Birke



Social Networks
and their Economics



Social Networks
and their Economics
Influencing Consumer Choice
Daniel Birke
Visiting Researcher, Aston Business School, Birmingham,
and works in a leading international management consultancy
in Germany.


This edition first published 2013
C 2013 John Wiley & Sons, Ltd
Registered office
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ,
United Kingdom
For details of our global editorial offices, for customer services and for information about how to apply
for permission to reuse the copyright material in this book please see our website at www.wiley.com.

The right of the author to be identified as the author of this work has been asserted in accordance with the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise,
except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of
the publisher.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may
not be available in electronic books.
Designations used by companies to distinguish their products are often claimed as trademarks. All brand
names and product names used in this book are trade names, service marks, trademarks or registered
trademarks of their respective owners. The publisher is not associated with any product or vendor
mentioned in this book.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in
preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not
engaged in rendering professional services and neither the publisher nor the author shall be liable for
damages arising herefrom. If professional advice or other expert assistance is required, the services of a
competent professional should be sought.
Library of Congress Cataloging-in-Publication Data
Birke, Daniel.
Social networks and their economics : influencing consumer choice / Dr Daniel Birke.
pages cm
Includes bibliographical references and index.
ISBN 978-1-118-45765-8 (cloth)
1. Social networks–Economic aspects. 2. Consumer behavior. I. Title.
HM741.B57 2013
658.8 34–dc23
2013017209
A catalogue record for this book is available from the British Library.

ISBN: 978-1-118-45765-8
Set in 10/12pt Times by Aptara Inc., New Delhi, India

1 2013


Contents
List of figures

ix

List of tables

xi

Preface

xv

Acknowledgements
1 Consumer choice in social networks
1.1 Motivation
1.2 Using mobile telecommunications to illustrate the economics of
social networks
1.3 Structure of the book
1.4 Why you should read this book
References
2 Research into social networks in economics, sociology and physics
2.1 Introduction
2.2 The economics of networks: Key findings from economics

and marketing
2.2.1 Introduction
2.2.2 Definition of network effects
2.2.3 Direct network effects
2.2.4 Indirect network effects
2.2.5 Implications for company strategies
2.3 Social network analysis: Key findings from sociology
2.3.1 A short history
2.3.2 Network analysis basics
2.3.3 Design of social network studies
2.4 Key findings from physics research into complex networks
2.5 Empirical research on social networks and network effects
2.5.1 Introduction
2.5.2 Big data: Massive electronic social networks

xvii
1
1
3
4
6
9
11
12
13
13
14
15
17
18

24
24
27
29
30
32
32
32


vi

CONTENTS

2.5.3

Challenges when identifying causal relationships
in social networks
2.5.4 Empirical strategies to identifying causal effects
in social networks
2.5.5 Empirical challenges and advances in the economics of
network literature
2.6 Summary
References

33
35
38
40
40

3 Marketing in social networks: The iPhone
3.1 Executive summary
3.2 Case study 1: Social network and viral marketing
3.3 Case study 2: Social advertising on Facebook
3.4 Introduction to the empirical study
3.5 Product diffusion in social networks
3.6 Modelling diffusion in social networks
3.7 Model estimation
3.7.1 Description of the data used: Very large-scale mobile
network
3.7.2 Description of the statistical method used: Survival analysis
3.8 Model results
3.8.1 Non-parametric tests
3.8.2 Variable definitions
3.8.3 Model results: Impact of the social network on
iPhone adoption
3.8.4 iPhone virality over time
3.9 Discussion
References

47
47
48
52
54
55
57
59

4 Switching and churn in social networks
4.1 Executive summary
4.2 Case study: Customer retention in social networks
4.3 Introduction to the empirical study
4.4 Key findings from the switching cost literature
4.5 Modelling concept
4.6 Description of the data used: Another large-scale mobile
network
4.7 Model results
4.7.1 Non-parametric tests
4.7.2 Variable definitions
4.7.3 Model results: Impact of the social network on
customer churn
4.7.4 Robustness tests
4.8 Discussion
References

71
71
72
75
76
78

59
60
62
62
63
64

65
67
68

79
81
81
81
83
85
86
88


CONTENTS

vii

5 How social networks influence consumer choice of mobile phone
carriers in the UK, Europe and Asia
5.1 Executive summary
5.2 Case study: Using homophily for social network marketing
5.2.1
Mobile phone carriers
5.2.2
Online retailers
5.2.3
Online social networks
5.3 Introduction to the empirical study
5.4 Methodology

5.4.1
Design of the social network survey
5.4.2
Description of the statistical approach used: Quadratic
assignment procedure
5.5 Understanding the properties of the social networks
5.5.1
Descriptive social network statistics
5.5.2
Graphical analysis of a social network
5.6 The impact of friendship on operator choice
5.7 Robustness of results
5.7.1
Non-respondents
5.7.2
QAP and multicollinearity
5.7.3
Ethnicity
5.8 Are stronger relationships more influential?
5.9 Friendship networks and choice of handset brand
5.10 Multi-country case study of operator choice in social networks
5.10.1 Malaysia
5.10.2 The Netherlands
5.10.3 Italy
5.10.4 Cross-country comparison
5.11 Discussion
References

100
102

102
106
108
112
112
114
116
117
120
122
123
124
127
132
133
134

6 Coordination of mobile operator choice within households
6.1 Executive summary
6.2 Case study: Social network marketing to communities
6.2.1
International communities
6.2.2
Families
6.3 Introduction to the empirical study
6.4 Data
6.5
Descriptive statistics
6.6 The model
6.7 Multinomial logit model

6.7.1
Model parameters
6.7.2
Base model
6.7.3
Relationship types within households
6.8 Multinomial probit model
6.8.1
Independence of irrelevant alternatives

137
138
138
139
140
142
143
144
146
148
148
149
152
153
153

91
92
92
93

95
95
96
98
98


viii

CONTENTS

6.9

6.8.2
Multinomial probit motivation
6.8.3
Estimation results
Discussion
References

7 How pricing strategy influences consumer behaviour
in social networks
7.1 Executive summary
7.2 Case study: Pricing digital products with network effects
7.2.1
Facebook
7.2.2
LinkedIn
7.3 Introduction to the empirical study
7.4 The mobile telecommunications industry in the UK

7.5 The model: Price discrimination between on- and off-net calls
7.6 Estimation results: Adapting consumption choice
to price signals
7.7 Discussion
References

155
157
158
158

161
161
162
164
164
165
167
169
173
175
176

8 Conclusions
8.1 Main results
8.2 Implications of interdependent consumer choice
8.2.1
For marketing practitioners
8.2.2
For academic researchers

8.2.3
For regulatory policy
8.3 Looking ahead: How social network analysis is changing research
and marketing practice
References

177
177
178
178
179
180

Appendix A Success factors for viral marketing campaigns
A.1 Proposition excellence
A.2 Observability of the product or its use
A.3 Design the campaign around a good understanding of the specific
role of word-of-mouth in propagating your product
A.4 Word-of-mouth for economic benefit
A.5 Exploit storytelling and tap into the zeitgeist
A.6 Exploit influential expert users
A.7 Appeal to communities of interest
A.8 Conclusion
References

183
185
186

Appendix B

193

Index

Student questionnaire

180
181

187
187
188
189
189
190
191

197


List of figures
2.1

Overlapping social networks.

17

2.2

Technology diffusion in markets with network effects.

21

2.3

Increase in the use of ‘social networks’ in the title of academic
publications.

24

2.4

A sociogram with undirected (a) and directed relationships (b).

25

2.5

Examples of transitive (a) and intransitive ties (b).

26

2.6

Sociogram of an undirected graph (a) and a direct graph (b).

28

2.7

Transition from a regular (a) to a random (c) graph via a small-world
graph (b).

31

2.8

Penetration of social networking on smartphones.

33

2.9

(a) Outside-in and (b) inside-out approach for measuring network
influence.

36

3.1

Social network pressure and influence.

48

3.2

Applications of social network analysis across the customer life cycle.

50

3.3

Spreading of new products in networks.

58

3.4

Two-stage adoption process.

58

3.5

Time structure of the model.

59

3.6

Duration data.

62

3.7

Correlation of iPhone purchases over time.

66

3.8

Level of iPhone virality over time.

66

3.9

iPhone diffusion speed over time.

67

4.1

Illustrative benefits of combining traditional churn models with social
network model.

73

Illustrative results from prediction of churn influencers.

74

4.2


x

LIST OF FIGURES

5.1

Illustrative levels of handset homophily.

94

5.2

Predicting user characteristics in a social network.

94

5.3

UK 2005: Student class social network.

107

5.4

UK 2005: Full student class social network.

114

5.5

UK 2005: Nationality and ethnicity of students.

117

5.6

UK 2006: Student class social network.

118

5.7

Malaysia: Student class social network.

124

5.8

The Netherlands: Student class social network.

126

5.9

Italy: Student class social network.

128

6.1

Typical family communications patterns.

140

7.1

Number of mobile phone subscribers in the UK (in thousands).

167

7.2

Development of subscriber market shares.

168

7.3

Development of on- and off-net call volumes.

172

7.4

Price and volume ratios between off- and on-net calls.

172

7.5

Fitted and observed values for off-/on-net calls.

175

A.1

Personalised Nokia handsets.

185


List of tables
3.1

Example of survival analysis applications.

61

3.2

iPhone uptake rate by number of ‘infected’ neighbours.

63

3.3

Definition of variables.

64

3.4

Regression results from log-normal base model.

65

4.1

Churn and iPhone uptake rate by number of ‘infected’ neighbours.

81

4.2

Definition of variables.

82

4.3

Regression results from log-normal base model.

84

4.4

Regression results from survival analysis model.

87

5.1

Customer-related knowledge by industry.

93

5.2

Permutation of rows and columns (QAP).

102

5.3

Nationality and gender of respondents.

102

5.4

Out-degree and nationality.

103

5.5

Frequencies for choice criteria.

104

5.6

Do you know which operator your friends/family/partner uses?

104

5.7

Duration of mobile phone usage per week.

105

5.8

Number of SMS sent per week.

105

5.9

Mixing patterns between students from different nationalities.

106

5.10 Determinants of choosing the same operator (UK 2005).

109

5.11 Calculation of operator coordination measure.

110

5.12 Degree of coordination (UK 2005) by operator.

110

5.13 Degree of coordination (UK 2005) by nationality.

111

5.14 Friendship determinants.

112


xii

LIST OF TABLES

5.15 Predicted probabilities of calling each other.

112

5.16 Non-respondents and nationality.

113

5.17 Non-respondents and gender.

113

5.18 Non-respondents and operator choice.

113

5.19 Regression results from robustness checks.

115

5.20 Determinants of choosing the same operator (UK 2006).

119

5.21 Determinants of mobile handset choice.

120

5.22 Handset choice by operator (expected values in brackets).

121

5.23 Sample size and response rates.

122

5.24 Do you know which operator your friends/family/partner use?

125

5.25 Determinants of choosing the same operator (The Netherlands).

126

5.26 Determinants of choosing the same operator (Italy).

129

5.27 Operators chosen when respondents have multiple operators.

130

5.28 Degree of coordination by operator.

130

5.29 Operator coordination between respondents and fathers (expected
figures in brackets).

130

5.30 Operator coordination between respondents and mothers (expected
figures in brackets).

131

5.31 Operator coordination between respondents and siblings (expected
figures in brackets).

131

5.32 Operator coordination between respondents and partners (expected
figures in brackets).

132

5.33 Degree of coordination in different countries.

132

5.34 Observed versus expected percentage of same operator dyads
amongst friends.

133

6.1

Number of respondents.

143

6.2

Survey participation of wave 3 respondents.

144

6.3

Observed number of operators per household (wave 3).

145

6.4

Expected number of operators per household (wave 3).

145

6.5

Determinants of operator choice (MNL model).

150

6.6

Predicted probabilities of operator choice (MNL model).

152


LIST OF TABLES

xiii

6.7

Coordination of operator choice by type of relationship.

153

6.8

Hausman-test results for IIA assumption.

154

6.9

Determinants of operator choice (MNP model).

157

6.10 Predicted probabilities of operator choice (MNP model).

157

7.1

Observed shares (by volume of calls).

170

7.2

Expected shares (by volume of calls).

170

7.3

Expected shares (by volume of calls) second quarter 1999.

171

7.4

Regression results for off-/on-net call volumes.

173

A.1

Is your product/campaign suitable for word-of-mouth marketing?

190



Preface
To understand what influences consumers in their purchasing decisions has been at
the heart of marketing for decades. Intuitively, everybody understands that purchasing
decisions are based on our own individual preferences and that we are at the same time
influenced by our friends and peers in what we do, how we behave and what products
we consume. However, until recently, it was difficult to measure this interdependence,
mainly because data on social networks were difficult to collect and not readily
available. Nowadays, more and more companies, like mobile phone companies or
social networking sites like Facebook, collect such data electronically. There is,
therefore, a strong academic and practitioner interest in measuring how consumers
are influenced by their social network in the products they consume.
This book uses the author’s unique experience in carrying out academic research
on consumer choice in social networks, starting up a company that successfully
commercialised these insights and working in a top-management consultancy advising companies on Marketing and Sales. It is relevant for both an academic and a

practitioner audience:

r From an academic perspective, the book is most relevant to final year undergraduate, postgraduate and university researchers in industrial economics and
consumer marketing. Each chapter uses different empirical studies demonstrating how consumption interdependences can be measured. A number of
different research techniques (primary and secondary surveys, electronic data
collection) and different statistical techniques (survival analysis, multinomial
logit, time-series statistics, permutation tests) are used. The case studies and
related questions can be used in the class room.

r The book is also directly relevant for marketers interested in how to turn
social network data into actionable insights and campaigns. Based on the
author’s experience working together with a large number of marketing and
sales departments, each chapter starts with an executive summary of relevant
aspects from a practitioner point of view. Furthermore, each chapter is preceded
by a case study discussing practical implications of the research in areas such
as social network marketing, retention, pricing strategy and so on. For example,
Chapter 4 on how switching of mobile phone providers is influenced by one’s
peers, is preceded by a case study on how several mobile phone providers
are using these insights to reduce customer churn among their subscribers.


xvi

PREFACE

Appendix A includes a discussion of the success factors for viral marketing
campaigns.
This book mainly covers the following topics:

r Network effects and the analysis of social networks: Overview of the state-ofthe art research.

r Consumption interdependences between friends and peers: Who is influencing
whom through which channels and to what degree?

r Statistical methods and research techniques that can be used in the analysis of
social networks.

r Social network analysis and its practical application for marketing purposes.
This book contains an accompanying website. Please visit www.wiley.com/
go/social_networks


Acknowledgements
I am very grateful to my wife Yundan and my children for coping with their husband/
daddy locking himself in the office to write this book. To them I am dedicating this
book.
This book has benefited from a number of people and I am very grateful for this
help and support. First and foremost I would like to thank my PhD advisor Peter Swann
who supported me from the first meetings at Manchester Business School, through
meetings at Bridgewater Hall to the award of my PhD at Nottingham University
Business School, and since then as a very good friend. I in particular enjoyed the
stimulating discussions which helped me not only to write my PhD thesis, but to
understand what is needed to become a good academic.
Towards the end of my PhD in 2006 I started with Idiro Technologies, a software
company specialising in analysing very large social networks in order to derive
marketing recommendations. I had four fantastic years with Idiro and thoroughly
enjoyed being able to translate my PhD research into practical use and being able
to work with our customers on combining the model predictions with the other
elements of successful marketing campaigns. I am in particular grateful to Aidan
Connolly, Brendan Casey and my team members. A special thanks goes to Simon

Rees, Sales & Marketing Director of Idiro Technologies for his deep insights into
the mobile telecommunications industry (and many great nights in Istanbul!). Simon
also contributed the discussion of the success factors for viral marketing campaigns
in Appendix A which is a great reference resource for organizations who want to
run a viral marketing campaign. Thank you as well to Robert Walker from Ernst &
Young’s Customer practice who enabled me to take a three months sabbatical to write
this book.
I would also in particular like to thank John Belchamber, Ricardo Correia, Paul
David, Chris Easingwood, Nicolas Economides, Koen Frenken, Sourafel Girma, Gautam Gowrisankaran, Francesco Lissoni, David Paton, Roy Radner, Paul Stoneman,
Arun Sundararajan, Steve Thompson, Reinhilde Veugelers and many others who gave
me helpful comments and suggestions.
This book also would not have been possible without the extensive access to
data that I was able to gain from a number of sources. I would like to thank Idiro
Technologies and two mobile phone companies that shall remain anonymous for
providing me access to the data for Chapters 3 and 4. By enabling and supporting
me to run surveys with their students at the University of Utrecht, University of


xviii

ACKNOWLEDGEMENTS

Nottingham in Malaysia and at the University of Brescia, Koen Frenken, Yoong
Hon Lee and Francesco Lissoni made the data collection for Chapter 5 possible.
Ben Anderson from Chimera, the Institute for Social and Economic Research at the
University of Essex, the ESRC data archive and Nicoletta Corrocher helped me with
data for Chapter 6. Last but not least, I would like to thank Hilary Anderson from
OFCOM who granted me access to the data on which Chapter 7 is based.
I would also like to gratefully acknowledge financial support from Nottingham
University Business School and the ESRC, which allowed me to focus on my research

during my PhD years.
Last but not least I would like to thank my publisher Wiley & Sons and their team
for shepherding and guiding me through the publication process, in particular Richard
Davies, Heather Kay, Debbie Jupe, Ilaria Meliconi, Paulina Shirley and Jo Taylor.
Daniel Birke


1

Consumer choice in
social networks
1.1 Motivation
1.2 Using mobile telecommunications to illustrate the economics of social
networks
1.3 Structure of the book
1.4 Why you should read this book
References

1.1

1
3
4
6
9

Motivation

The basic conjecture of this book is that consumers do not make decisions in isolation, but are influenced by and influence other consumers with whom they interact.
Everyday experience suggests that we are frequently influenced by others: we ask

our peers for restaurant tips, hear about new products from them, make joint consumption decisions for family cars within families, consume similar products to our
peers in order to ‘keep up with the Joneses’ and use similar products as people we
regard highly and aspire to. These processes happen within social networks, which
in this book means all social relationships between people. In recent years, social
networks such as Facebook have become very popular. Thinking about ones social
relationships as a social network has consequently become very intuitive for many
people – whether these relationships are maintained via Facebook, mobile phones or
via traditional offline channels.
However, for a long time much of economics and marketing did not take these
interrelationships into account. There are a number of good reasons for this focus
on treating individuals as atomistic decision units: First, it was difficult to collect
Social Networks and their Economics: Influencing Consumer Choice, First Edition. Daniel Birke.
© 2013 John Wiley & Sons, Ltd. Published 2013 by John Wiley & Sons, Ltd.


2

SOCIAL NETWORKS AND THEIR ECONOMICS

appropriate network data, especially in the past when such data had to be gathered
with the help of network surveys asking respondents to identify relationships they
have with their peers (see also Chapter 5). Secondly, most traditional statistical
methods assume that observations are independent, an assumption that is clearly
violated in social networks and requires different statistical approaches. Thirdly, data
volumes when analysing electronic social network datasets can be huge. Large-scale
social networks can have tens or hundreds of millions of users and for each user, there
can easily be 100–1000 communications per observation period, meaning that data
volumes can be 100–1000 times larger than for comparable individual-level data.
Fourthly, it is more difficult to incorporate interdependences in social networks into
theoretical economic and marketing models.

In the last couple of years, massive electronic social network datasets have become
available and companies now have the computational capabilities to analyse them
properly, although this can still be a challenge when working with datasets covering
tens or hundreds of million users. At the same time, researchers such as Snijders, van
de Bunt and Steglich (2010) have developed new statistical techniques and running
experiments using electronic social networks have become increasingly popular (see
e.g. Bakshy et al., 2012). There has also been some progress on theoretical models
(e.g. Sundararajan, 2007), but in general there has been a shift in emphasis towards
more empirical work, something which is also driven by the increased emphasis
of marketing practice on quantitative marketing and, therefore, greater demand for
quantitative analystics skills.
A particular challenge when analysing influence processes in social networks is to
identify the causality of events. There can be a large number of reasons why decisions
of consumers in a social network are not independent (Manski, 1993) and one of the
key challenges of research into social networks is to tease such different effects
apart. For example, individuals who interact with each other typically behave in a
similar way because they share the same environment, because they receive similar
information and because of psychological factors like group pressure. If we want to
use the social network structure to achieve a certain outcome, say because we want to
promote a certain product, then identifying the main cause(s) is key to pursuing the
right marketing strategy. If, for example, a lot of consumers who are likely to buy a
product share the same context, then the company might want to target a certain social
or geographic segment. For a wine retailer it is, for example, critically important to
find the right location if a high number of consumers is geographically concentrated
in a particular location as consumers will then buy their wine independent of the
social networks that exist in this location. However, if a high number of customers of
an online wine retailer recommend this retailer to their friends, then it is important
to understand how customers interact with each other and which customers are most
likely to successfully recommend the service.
This book uses a variety of electronic and non-electronic datasets to study how

consumers influence each other, and establishes causality by using the time structure
of events occurring over the network and via cross-country case study research. I also
use social network analysis of very large electronic datasets, an approach that has the
potential to revolutionise marketing and can help extend our understanding of human
behaviour.


CONSUMER CHOICE IN SOCIAL NETWORKS

1.2

3

Using mobile telecommunications to illustrate the
economics of social networks

A large part of the book is based on empirical analysis of the mobile telecommunications industry. Telecommunications networks are a prime example of markets
where consumers influence each other and where this influence can be measured by
data. Similar phenomena exist in other industries, such as social networking sites
and finance. Furthermore, data on social network interactions are available to varying
degrees in a number of other industries. Even if there is no electronic data readily
available this book shows that social network surveys can be used in such cases (see
Chapter 5). In mobile telecommunications, consumer demand is interdependent for
a variety of reasons:

r First, every subscriber to a mobile phone network benefits from other subscribers also using mobile phones as it allows communication with a greater
number of users. Network effects therefore influence the overall diffusion
pattern of mobile phones. The same is true for social networking sites like
Facebook, LinkedIn and so on. Likewise, network expansion drives the usage
volume of people already using mobile phones. The usage volume of existing subscribers therefore increases with the total number of mobile telephone

subscribers.

r Secondly, in mobile telecoms and other industries it is becoming increasingly
important to create product eco-systems. While Apple, Google or Facebook
directly provide the basic functionality of their respective products, they also
create platforms and interfaces that allow other companies to offer their products and services via their platforms. This means that higher user numbers
make platforms more attractive for these third parties to develop their offering
for a particular platform, and this in turn makes the platform more attractive
to end users as well. Often, complementary services developed for a platform
also create direct network effects, such as Apple’s FaceTime application which
allows iPhone and iPad users to communicate with each other for free on their
mobile devices. Such complementary services are, therefore, a good way of
making products stickier.

r Thirdly, in mobile telecommunications, calls to the same network are typically
cheaper than calls to other networks, and it is therefore beneficial for consumers
to subscribe to the same network as the people they are calling. Mobile phone
networks from different companies are highly compatible with each other from
a technological point of view, but network effects are often induced by network
operators through higher prices for off-net than for on-net calls, something
which Laffont et al. (1998) termed tariff-mediated network effects. Tariffmediated network effects can take the form of a general price discrimination
between on- and off-net calls or can be created through discounts for certain
types of on-net calls. Probably the most famous example of such a scheme
is MCI’s Friends and Family plan, which was introduced at the beginning of


4

SOCIAL NETWORKS AND THEIR ECONOMICS

the 1990s and allowed MCI customers to call up to 20 other MCI customers
at a cheaper rate. In most European countries such price differentiation is
common place, but there are also exceptions, like the Netherlands, where
operators charge the same prices for calls to the same network and calls to
other networks.

r Fourthly, the use of mobile phones is conspicuous and sends out social signals about the users. Using an attractive handset, like for example an iPhone,
enhances the social standing of its owners and peers might be influenced by
their peer group in their choice of mobile phone, just as drivers of luxury
cars influence others in their neighbourhood and peer group to buy similar
prestigious car brands.

r Fifthly, users of relatively complex products such as mobile phones and the
services running on them benefit from information exchange with their peers.
Such information exchange can be about new services, about the advantages
and disadvantages of existing services or simply about how to use certain
services or functionalities.
Variations of these factors will also be important in many other industries and can be
observed in electronic data, for example, in many online businesses such as online
social networks. While most of the empirical data for this book come from the mobile
telecommunications industry, the insights and methods are, therefore, more generally
applicable.

1.3

Structure of the book

The book consists of six main chapters: one chapter reviewing the relevant prior
research and five empirical chapters looking at various aspects of how consumers
influence each other in a social network. Each empirical chapter starts with an executive summary and one or two case studies on how social network analysis can and

is used for marketing purposes.
Chapter 2 starts with a short history of the relevant literature and research from
economics, marketing, sociology and physics. One of the exciting aspects of studying how social networks influence consumer behaviour is that a number of very
different subject areas can contribute to our understanding. The chapter reviews the
key relevant research strands in each area and shows how they add to our overall
understanding of the underlying processes. In general, the book draws from two main
bodies of literature that help us understand how consumers influence each other: economics/marketing and sociology/social network analysis. The economics literature
in particular sheds light on aggregate phenomena, like overall competitive outcomes
in markets with network effects; whereas the social network analysis literature offers
a wide variety of lessons on how to influence behaviour at an individual level within a
social network. Furthermore, Chapter 2 focuses on practical implications for companies and discusses how causal relationships can be identified when studying dynamic


CONSUMER CHOICE IN SOCIAL NETWORKS

5

processes in social networks – something that is of particular importance for marketing interventions.
Chapters 3 and 4 study how the diffusion of a new product like the iPhone
(Chapter 3) and switching decisions between rival networks (Chapter 4) are influenced
by social networks. The analyses use call detail records from all subscribers of two
large European mobile phone operators to construct a social network and track
product uptake and switching decisions over a period of four months. Based on
survival analysis models, the results show that the more network connections that
have taken up the iPhone, the more likely it is that the focal consumer also takes
up the iPhone. Interestingly, this contagion effect decreases only slowly over time
after an initial peak at product launch. Likewise, one friend switching operators has
a strong impact on the switching decision of the focal consumer. These two chapters
are particularly relevant for companies and researchers with access to large-scale
social network data who would like to understand how to leverage the opportunities

provided by such data for both research and marketing. Chapter 3 is accompanied by
two case studies. The first discusses how mobile phone companies can approach social
network marketing for customer acquisition, product upsell/ cross-sell and customer
retention. The second discusses different ways in which social advertising is and can
be used on Facebook. A third case study preceding Chapter 4 focuses specifically on
customer retention, which is arguably the earliest and still most common application
of social network marketing in the mobile telecommunications industry.
Chapter 5 demonstrates a different way of collecting social network data through
the use of a social network questionnaire. This approach is particularly useful if
electronic social network data is not available at all or if researchers/marketers are
interested in particular individual-level data which are not available from electronically collected datasets. The chapter is based on primary survey data from a number
of university classes in Europe and Asia and uses a statistical permutation method
called Quadratic Assignment Procedure (QAP) to account for a correlation in error
terms for non-independent observations in a social network. The results demonstrate
that friends tend to choose the same mobile phone carrier and that this coordination
is stronger the closer the relationship. Interestingly, using variations in the pricing
strategy between operators and countries, this chapter shows that this coordination
is caused by price differences, rather than by alternative potential causes such as
peer pressure or unobserved socio-demographic similarities among friends. Besides
deciding to use the same operator as their peers, consumers also react to the consumption decisions of their peers by choosing to be part of several networks at the
same time if their friends are on different networks. The accompanying case study
discusses how social network marketing can use homophily, the commonly observed
tendency that similar people interact with each other more frequently, to identify
consumers who are potentially interested in a particular product or to close gaps in
the knowledge of certain individual-level variables.
Chapter 6 analyses how households – the core of most people’s social network –
coordinate their consumer choice. The chapter is based on a large traditional threewave survey of British consumers and employs multinomial logit and probit models
to estimate the extent to which households coordinate their choice of mobile phone