Wk4 Case Study: Innovation
Wk4 Case Study: Innovation Review the following case study –
Dean Diehl Disertation
Dean Diehl Disertation – Alternative Formats
.Write a 1,000-1,500 word paper including the following headings and content:
Include at least two PCRs (
, Citation, and Reference) – one from this dissertation and one from one of your textbooks.
WAS
ON-DEMAND MUSIC STREAMING A
DISRUPTIVE INNOVATION?
by
Dean Diehl
Dissertation
Submitted to the Faculty of
Trevecca Nazarene University
School of Graduate and Continuing Studies
in Partial Fulfillment of the Requirements for
the Degree of
Doctor of Education
In
Leadership and Professional Practice
May 201
9
WAS ON-DEMAND MUSIC STREAMING A
DISRUPTIVE INNOVATION?
by
Dean Diehl
Dissertation
__________________________________ _______________
Dissertation Adviser Date
___________________________________ __________________
Dissertation Reader Date
___________________________________ __________________
Dissertation Coordinator Date
___________________________________ __________________
EdD Program Director Date
___________________________________ __________________
Dean, School of Graduate & Continuing Studies Date
02/26/2019
02/26/2019
02/26/2019
02/26/2019
02/26/2019
i
© 20
19
Dean Diehl
All Rights Reserved
ii
ACKNOWLEDGEMENTS
I would like to thank my dissertation advisor, Dr. Jea Agee, for his invaluable
assistance in completing this project as well as Dr. Randy Carden, Dr. Glenn Schmidt,
and Dr. Tim Brown for their input and guidance. I would also like to thank Dr. Jim Hiatt,
Dean of the Skinner School of Business and Technology as well as Greg Runyan,
Chairman of the Skinner School of Business and Technology for their encouragement
and accommodation as I completed this work.
iii
ABSTRACT
by
Dean Diehl, Ed.D.
Trevecca Nazarene University
August 2019
Major Area: Leadership and Professional Practice Number of Words 10
6
Disruptive innovation theory, introduced and developed by Dr. Clayton Christensen in
the late 1990s, has come to be confused with any innovation that encroaches upon
existing options. In order to clarify the theory of disruptive innovations, scholars have
repeatedly called for the application of the core concepts of the theory to the data
surrounding the introduction of innovations from various fields. This study applied the
concepts of disruptive innovation theory to the data surrounding the introduction and rise
of on-demand music streaming between the years of 2001 and 2017 in order to test
whether on-demand music streaming constituted a disruptive innovation as defined by the
theory.
i
v
TABLE OF CONTENTS
I. INTRODUCTION …………………………………………………………………………………………..
1
Statement of the Problem …………………………………………………………………………………
3
Rationale ……………………………………………………………………………………………………….
4
Research Questions ……………………………………………………………………………………….
13
Contribution of the Study ……………………………………………………………………………… 1
5
Process to Accomplish …………………………………………………………………………………..
16
II. REVIEW OF THE LITERATURE ………………………………………………………………….
21
Historical Perspective ……………………………………………………………………………………
23
Digital Downloads: A Sustaining Innovation …………………………………………………… 4
2
Conclusion …………………………………………………………………………………………………..
46
III. METHODOLOGY ……………………………………………………………………………………….. 4
7
Research Design …………………………………………………………………………………………..
49
Participants ………………………………………………………………………………………………….
52
Data Collection …………………………………………………………………………………………….
55
Analytical Methods ………………………………………………………………………………………. 5
8
IV. FINDINGS AND CONCLUSIONS …………………………………………………………………
63
Findings ……………………………………………………………………………………………………… 64
v
Summary of Findings ……………………………………………………………………………………
83
Limitations …………………………………………………………………………………………………..
87
Implications and Recommendations ………………………………………………………………..
90
vi
LIST OF TABLES AND FIGURES
Figure 1.1 Innovation in Video Software (Shapriro, 2014). ……………………………………….. 6
Figure 2.1 Diffusion of innovations over time and by frequency (Rogers, 2006). ………..
25
Table 3.1 US Music Consumers (MusicWatch, 2018). …………………………………………….
53
Table 3.2 US Raw Sales Data for the First Two Weeks of 2017 (Nielsen, 2018). ………..
56
Table 3.3 US Converted Data for First Two Weeks of 2017 (Nielsen, 2018). ……………..
57
Table 4.1 Playback Media Performance …………………………………………………………………
65
Table 4.2 2008 Total Music Consumption by Format (Nielsen, 2018). ………………………
69
Figure 4.1 2008-2010 Weekly US Consumption by Format (Nielsen, 2018). ……………..
70
Table 4.3 2008-2010 Correlation between CDs, DL Albums, DL Songs, and Streams
(Nielsen, 2018). ……………………………………………………………………………………………
71
Table 4.4 2008-2010 Average Consumption by Format (Nielsen, 2018). ……………………
73
Table 4.5 2008-2010 Paid-to-Non-Paid Ratio (Nielsen, 2018). …………………………………
74
Table 4.6 2008-2010 Average % of Consumption in Streaming (Nielsen, 2018). ………..
75
Figure 4.2 2011-2017 Weekly US Consumption by Format (Nielsen, 2018). ……………..
81
Table 4.7 2011-2017 Correlation between CDs, DL Albums, DL Songs, and Streams
(Nielsen, 2018). ……………………………………………………………………………………………
82
Figure 4.3 2008-2017 Weekly US Consumption by Format (Nielsen, 2018). ……………..
86
Table 4.8 Growth in Streaming % of Total by Genre from 2011-2017 (Nielsen, 2018). . 91
1
CHAPTER ONE
INTRODUCTION
Creative Destruction is the essential fact about capitalism.—Joseph A. Schumpeter
All innovation is disruptive. Not every innovation, however, is a disruptive
innovation properly understood (Schmidt & Druehl, 2008). Confusion over what
constitutes a true disruptive innovation has led many leaders to make tactical and
strategic business errors, often with tragic results (Christensen, Raynor, & McDonald,
2015). Christensen et al. (2015) stated, “The problem with conflating a disruptive
innovation with any breakthrough that changes an industry’s competitive patterns is that
different types of innovation require different strategic approaches” (p. 4). Leaders must
learn to distinguish true disruptive innovation from other forms of innovation.
Simply stated, a disruptive innovation is one in which the innovation’s initial
performance is considered to be inferior to existing options in those attributes most
valued by the mainstream market, called core competitive dimensions, leading
mainstream consumers to dismiss the innovation. A disruptive innovation, however,
survives because it finds a place among low-end consumers of the existing market or
creates a new market due to its unique business model or its superiority to existing
options in one or more attributes, called secondary competitive dimensions. Over time,
the innovation improves its performance in the core competitive dimensions while
maintaining its unique advantages until it becomes acceptable to the mainstream,
2
allowing the innovation to encroach upon or disrupt existing options thus shifting the
competitive landscape (Christensen, 1997; Schmidt & Druehl, 2008).
Largely originating with the Clayton Christensen book, The Innovator’s Dilemma
(Christensen, 1997), and refined over the last two decades, disruptive innovation theory
generated much praise and more than a little criticism. Danneels (2004) stated, “One can
see from a search for disruptive technology on the web how loosely the term has come to
be used and how it has become separated from its theoretical base” (p. 257). Even
Christensen et al. (2015) agreed, stating, “Despite broad dissemination, the theory’s core
concepts have been widely misunderstood and its basic tenets frequently misapplied” (p.
4).
Properly applying a theory strengthens the theory. Scholars writing about
disruptive innovations have been consistent in pointing out the need for additional
involvement from both academics and practitioners in the process of identifying and
clarifying the key characteristics of disruptive innovations (Christensen et al., 2015;
Danneels, 2004; Schmidt & Druehl, 2008). It is particularly important to study industries
not previously examined in order to establish those characteristics of disruptive
innovations with broad applicability versus industry-specific characteristics (Danneels,
2004).
Clarifying and demonstrating the essential characteristics of disruptive
innovations is necessary to arriving at a predictive model of disruption. As Danneels
(2004) put it, “The real challenge to any theory…is how it performs predictively” (p.
250). Christensen et al. (2015) concurred stating, “As an ever-growing community of
researchers and practitioners continues to build on disruption theory and integrate it with
3
other perspectives, we will come to an even better understanding of what helps firms
innovate successfully” (p. 11).
With that context in mind, one innovation that bears examining is on-demand
music streaming. On the surface, the history of on-demand music streaming followed the
pattern of a disruptive innovation. However, while the popular press has covered
streaming in the music industry extensively, few scholarly articles exist, and, most of
what does exist relied on incomplete summary data available to the public. An in-depth
analysis of on-demand music streaming supported by comprehensive data from inside the
industry is the kind of study called for by disruption scholars in the hope of further
refining the theory of disruptive innovations.
Statement of the Problem
The purpose of this study was to apply disruptive innovation theory to data
surrounding the introduction and rise of on-demand music streaming in the United States.
Through the collection and analysis of quantitative and qualitative data in the form of
archival sales records and documents, this study considered whether on-demand music
streaming possessed the essential characteristics of a disruptive innovation as defined by
the theory. Conducted in response to a call from disruption theorists for the application of
disruptive innovation theory to industries and innovations not previously studied, this
study attempted to identify patterns and uncover anomalies that would strengthen the
theory.
According to the theory, for on-demand streaming of music in the United States to
have been a true disruptive innovation, it would have initially been inferior to existing
options in a core competitive dimension. As a result, mainstream consumers of music
4
would have rejected on-demand music streaming. In spite of this rejection, on-demand
music streaming would have appealed to the low-end of the market or established a brand
new market through its unique business model or through its superior performance in
some secondary competitive dimension. Finally, over time on-demand music streaming
would have improved performance in the core competitive dimension until it became
acceptable to mainstream consumers leading to the disruption of existing options and a
shift in the overall competitive landscape of music (Christensen, 2015; Schmidt &
Druehl, 2008). It was the goal of this study to test the facts of on-demand music
streaming against these essential elements of a disruptive innovation.
Rationale
Disruptive innovation theory has been disrupted. Twenty years after first
introducing the concept of disruptive innovations, initially called disruptive technologies,
Clayton Christensen summed up the current state of the theory in a 2015 Harvard
Business Review article titled, “What is Disruptive Innovation” (Christensen, Raynor &
McDonald, 2015) where he commented, “Disruption theory is in danger of becoming a
victim of its own success.” Christensen went on to say, “In our experience, too many
people who speak of ‘disruption’ have not read a serious book or article on the subject”
(Christensen et al., 2015, p. 4).
Disruptive innovation theory has been criticized as too narrow (Downes & Nunes,
2013), too broad (Danneels, 2004), and even outdated (Wessell, 2016). There have been
calls for clearer definitions and categorizations (Schmidt & Druehl, 2008) as well as calls
for broadening the definitions (Wessell, 2016). It is safe to say disruptive innovation is a
5
theory in need of further testing of its core concepts against real-world innovations to
define just what a disruptive innovation is, and what it is not.
Disruptive innovations theory is an offshoot of diffusion of innovations theory, a
theory that goes back to the early 1960s with roots in sociology, psychology, and
marketing. Often associated with the work of Everett Rogers (2003), diffusion of
innovations theory deals with the way new ideas and products move, or diffuse, through a
community. Rogers (2003), in summarizing the concept, wrote, “Diffusion is the process
by which 1) an innovation 2) is communicated through certain channels 3) over time 4)
among members of a social system” (p. 11., emphasis in original).
In diffusion, a key concept to understand is compatibility, or “the degree to which
an innovation is perceived as consistent with the existing values, past experiences, and
needs of potential adopters” (Rogers, 2003, p. 240). Opinion leaders, the key influencers
within an industry’s market, look for innovations that are, as Valente (2006) put it,
“compatible with the culture of the community” (p. 68). Innovations perceived as
incompatible are often delayed or rejected by opinion leaders (Valente, 2006). Therefore,
dependence on adoption by opinion leaders within an industry causes innovation within
that industry to concentrate on a desired attribute or set of attributes called core
competitive dimensions
(Christensen, 1997).
Successful firms within an industry anticipate the peak performance of existing
options and introduce successor innovations accordingly. These successor products or
services innovate along the core competitive dimension with each product outperforming
the previous product in that dimension (Christensen, 1997). Over time, succeeding
6
innovations produce an upward rising performance curve along the core competitive
dimension (See Figure 1.1).
Figure 1.1 Innovation in Video Software (Shapiro, 2014).
Using an illustration from the film industry, innovation in video software
developed along the core competitive dimension of portability, referring to the ability to
take your movies with you. 16 mm film, with its clunky projectors, large reels, and need
for a screen were not very portable. The VHS cassette provided much more portability
and the DVD, with its thin, durable disc, was even more portable than the VHS (Shapiro,
2014).
From the 16 mm film to the DVD, existing firms and content owners within the
film industry, motivated by the needs of their core consumers, drove innovation towards
ever-increasing portability (Shapiro, 2014). Christensen (1997) defined this type of
innovation as sustaining innovation. Christensen et al. (2015) stated that sustaining
innovations “make good products better in the eyes of one’s existing customers” (p. 5).
7
Where disruptive innovations depart from sustaining innovations is that they are
initially inferior in regards to performance in the core competitive dimension, innovating
instead along some new or overlooked secondary competitive dimension or through the
development of a unique business model (Danneels, 2004; Schmidt & Druehl, 2008).
This inferiority in regards to performance in the core competitive dimension causes
opinion leaders to reject the innovation (Schmidt & Druehl, 2008). Ignored and rejected
by the mainstream, these innovations still manage to survive. Schmidt and Druehl (2008)
elaborate, “While existing high-end customers dislike the new product (they despise its
poor performance along the first dimension), a new market segment (or the existing low-
end segment) gladly accepts the de-rated performance along the first dimension in favor
of lower cost or the enhanced performance along the second dimension” (p. 352).
Disruptive innovations develop on the fringes of a market, or create a new market,
slowly evolving and improving performance over time. In the meantime, incumbent firms
innovating along the core dimension eventually overshoot the performance needs of the
market in the core dimension to the point that the market begins to shift their attention to
the previously undervalued secondary dimension or to the newly introduced business
model (Christensen, 1997). It is this shift in the entire basis of competition within a
market that is the hallmark of a disruptive innovation (Danneels, 2004).
When a disruptive innovation succeeds, it begins to take mainstream customers
away from incumbent firms, a process termed encroachment by Schmidt and Druehl
(2008). By the time incumbent firms realize what is happening, it is often too late to
respond. Before long, the disruptors have dominated the new market and the incumbents
are displaced (Christensen, 1997).
8
For example, take the case of Netflix, the disruptive innovation that displaced
video rental stores such as Blockbuster. When Netflix first appeared in 1998, its mail-
based delivery system, built on the newly introduced DVD and a monthly subscription
model, had little appeal to mainstream video rental customers who largely rented VHS
tapes on impulse. However, with its unique monthly subscription business model and no
due dates, late fees, or shipping costs, Netflix appealed to a fringe market including early
adopters of DVD players, people who liked the convenience of ordering from home, and
video rental customers sick of exorbitant late fees (Auletta, 2014). Over time, Netflix
gradually and then increasingly encroached upon brick-and-mortar video rental stores.
Then, in 2007, when Netflix launched their on-demand streaming service, mainstream
video rental customers poured into Netflix to the degree that, by 2013, Blockbuster, the
largest video rental chain in the United States, declared bankruptcy (Christensen et al.,
2015).
The “innovator’s dilemma,” according to Christensen (1997) is that, based on
convention, ignoring innovations that are inferior in the core competitive dimension is the
right response. It made perfect sense for Blockbuster to ignore Netflix and focus instead
on convenience and selection, those dimensions most valued by their existing consumers.
Yet, as Christensen (1997) points out, this strategy often leads to disruption, or, in the
case of Blockbuster, bankruptcy.
In the wake of the publication of The Innovator’s Dilemma, much of the
discussion in the academic community centered on strategies for responding to disruptive
innovations when they appeared. However, because all innovation is to some greater or
lesser degree disruptive, there began to be a lot of misapplication of disruption theory,
9
particularly among practitioners. Scott Anthony (2005) highlighted this new dilemma in
his article in the journal Strategy and Innovation, “Do You Really Know What You Are
Talking About?”:
The word disruption…has become loaded with meanings and
connotations at odds with the concept put forth by Clayton Christensen in
The Innovator’s Dilemma and highlighted in a 1999 Forbes magazine
cover story. As the term has increased in popularity, confusion about the
exact definition of disruption has increased as well, creating challenges for
companies seeking to grow through disruptive innovation.
Indeed, as the concept has seeped into the mainstream, this
language disconnect has generated confusion and led to the occasional
misallocation of resources. (p. 3, emphasis in original)
Anthony (2005) went on to state that confusion over what actually constitutes a
disruptive innovation is often due to three common mistakes: “1) mistaking disruptive
innovation for breakthrough innovation; 2) defining disruptive innovations against the
wrong parameters; and 3) forgetting that disruption involves more than technology” (p.
3). This confusion has led many to a call for further clarification of exactly what
constitutes a disruptive innovation (Danneels, 2004; Schmidt & Druehl, 2008). Schmidt
and Druehl stated, “(A) firm must be able to clearly delineate between what is a
disruptive innovation and what Christensen and Raynor (2003) and Christensen et al.
(2004) define as its converse: a sustaining innovation” (p. 347).
Disruptive innovation theory, like all theories, needs to be continually tested using
sets of historical data different from those already examined. As Danneels (2004) wrote,
10
“(A) reconsideration of the nature of disruptive technological change and its
consequences for firms and industries is in order” (p. 257). Testing a theory provides
opportunity for anomalies not explained by the theory to emerge. Christensen (2006), in
an article on improving theories, stated, “The primary purpose of the deductive half of the
theory-building cycle is to seek anomalies, not to avoid them” (p. 45). It is by testing a
theory that the theory becomes stronger.
The rationale, therefore, for testing the theory of disruptive innovation against the
data surrounding the introduction and rise of on-demand music streaming within the
United States was to determine if the data aligned with the theory or if anomalies would
emerge. As stated previously, while music streaming in the United States has received
much coverage in the popular press, there is not much literature within the academic
community, due in part to a lack of access to the raw sales data necessary for industry-
level analysis. However, through a unique arrangement, The Nielsen Company, the
primary compiler and reporter of marketing information in the entertainment industry,
released complete historical sales data from 2008 through 2017 for the purpose of this
study, making industry-level analysis a possibility.
To set the context for the rest of this study, it is necessary to summarize the
history of online digital music. Online music services first appeared in the late 1990s,
almost exclusively through illegal file-sharing websites like Napster and Pirate Radio.
Because the great majority of early online music activity was illegal, it was hard to
measure the degree of disruption for existing music formats. Although there was much
speculation at the time as to the impact of illegal streaming on music purchases, lack of
reliable data made scientific inquiry impossible. In addition, what data there was came
11
from a variety of sources with one source often contradicting another (Stevans &
Sessions, 2005). That said, all sources seemed to agree that illegal online activity
involved billions of downloads and streams (Auiar & Martens, 2013).
Researchers have taken every imaginable position as to the impact of illegal
activity upon legal options. Some claimed illegal activity killed legal purchases; others
posited there had been no impact at all because illegal users were never purchasers in the
first place; while still others stated the illegal activity actually increased legal purchases
of music (Stevans & Sessions, 2005; Auiar & Martens, 2013).
Regardless of the impact on sales, illegal downloading and streaming were not
without risks and inconveniences. Exposure to malware, viruses, and the potential
compromise of network security were all risks to file sharing. In addition, the activity was
illegal and thus subject to prosecution or penalty. Illegal music sites were also
cumbersome to use and often carried only a small portion of the titles available through
legal means (Machay,
2018).
In 2001, Apple, Inc. released the first iteration of iTunes, a music playback
software platform, initially only available for their own Macintosh computers, but, soon
after, available for all computer systems. In 2003, Apple released the iTunes digital music
store providing the first high profile, commercially viable, legal music download system
compatible with all major platforms. With licenses in place with practically every content
owner in the United States, the iTunes store gave consumers a legal way to purchase
digital files of the music they wanted (McElhearn, 2016). From 2001 through 2011,
digital purchases of albums and individual songs through iTunes and other sources such
as Google, Amazon, and Rhapsody, dominated online music activity, at least for those
12
wishing to obtain digital music legally. In that period, Apple’s iTunes platform was the
clear leader with a market share of online music purchases that reached beyond 60%
(Bostic, 2013).
At about the same time as Apple was launching iTunes in 2001, Rhapsody, best
known for its desktop digital music player, launched the first legal on-demand streaming
platform (Evangelista, 2002). On-demand streaming differed from downloads in that,
instead of purchasing individual tracks and owning them, listeners paid a monthly fee for
access to a catalog of music. In effect, listeners were renting music versus owning it.
In its first iteration, Rhapsody offered subscribers access to thousands of songs,
many of which were from small independent labels, however, by 2002, Rhapsody had
licenses in place with all of the majors labels and offered over 175,000 songs for instant
on-demand streaming (Evangelista, 2002). While there was a fringe market interested in
the Rhapsody model, Apple’s iTunes dominated the digital music market leading Steve
Jobs to famously quote, “People have told us over and over and over again, they don’t
want to rent their music” (Ricker,
2015, para. 3).
In 2011, a new take on the streaming model emerged as Spotify, previously only
available in Europe, launched in the United States. Spotify was the first significant online
platform to provide free on-demand streaming. Using an ad-supported model, Spotify
offered consumers free access to over 15,000,000 songs. While there were significant
restrictions on the free version, Spotify’s subscription model offered a viable legal option
to listeners who were largely using illegal streaming services (Sisario,
2011).
From 2011 to the present day, on-demand streaming has experienced exponential
growth and paid billions of dollars in royalties to artists and content owners for activity
13
that had largely been illegal and unpaid before the launch of Spotify. At the same time,
sales of physical formats and digital downloads have plummeted (Nielsen, 2018).
However, this does not automatically mean on-demand music streaming was a disruptive
innovation as defined by disruption theory (Christensen, 1997). As has been previously
stated in this chapter, disruption of previous activity within an industry does not
necessarily constitute a disruptive innovation. In order for on-demand music streaming to
have been a disruptive innovation, certain events must have occurred. The present study
addresses this issue in detail throughout the following pages.
Research Questions
The following questions were the focal point of the current study and formed the
overall process of investigation into whether or not on-demand music streaming
performed as a disruptive innovation as defined by current disruption theory.
1. Was there a core competitive dimension along which innovation occurred in
music playback formats prior to the introduction of on-demand streaming?
2. Was on-demand streaming initially inferior to existing music playback formats in
the core competitive dimension along which previous innovation had
occurred?
3. Did the mainstream music market initially reject on-demand music streaming as a
music playback format?
4. Did on-demand music streaming find acceptance among the low-end consumers
of the existing market or create a new market due to its superior performance in a
secondary competitive dimension or through the introduction of a unique business
model?
14
5. Did on-demand music streaming improve over time in the core competitive
dimension while maintaining its superiority in some secondary competitive
dimension or through its unique business
model?
6. Did on-demand music streaming eventually encroach upon sales of existing music
playback formats resulting in a shift in the competitive landscape?
Description of Terms
Core Competitive Dimension. A core competitive dimension refers to the attribute
of a product or service most valued by the existing mainstream market of an industry
(Christensen, 1997).
Low-End Consumer. For the purpose of disruption theory, the term low-end refers
to “willingness to pay.” Low-end consumers, then, would be existing consumers with the
lowest price threshold (Schmidt & Druehl, 2008).
On-Demand Music Streaming. On-demand music streaming refers to streaming
platforms where the consumer may choose the exact song they wish to hear. On-demand
streaming is the only type of streaming included in music industry sales reporting
(Passman, 2015).
Portability. Portability refers to the degree to which a music platform allows
listeners to take their music with them (Gopinath & Stanyek, 2014).
Programmed Streaming. Programmed streaming refers to streaming platforms
where the consumer can only select the style of music they wish to hear, but cannot select
individual songs (Passman, 2015).
Secondary Competitive Dimension. A secondary competitive dimension refers to
an attribute of a product or service of less importance to the mainstream market of an
15
industry but which has appeal to the low-end market or a new market for the product or
service (Christensen, 1997).
Contribution of the Study
Determining whether music streaming behaved as a disruptive innovation
according to current disruption theory benefitted three distinct populations: disruptive
innovation theorists, music distributors and content owners, and music business students.
Innovation theorists, including academics and practitioners, have been engaged in
an ongoing dialogue over the last twenty years, refining and adjusting the theory of
disruptive innovation. The goal of all of these endeavors has been to aid business leaders
in innovating successfully as well as in responding to disruption from others. While
views differ over many aspects of the theory, most theorists agree that analyzing data sets
from previously unexamined industries moves the theory forward by either confirming or
challenging its core principles. This study contributed to that effort.
An in-depth analysis of the data surrounding on-demand music streaming
identified consumer segments who were early adopters of the platform. The ability to
describe adopters of new music technology was of great benefit to music distributors and
content owners. First, it enabled the identification of consumers who have yet to adopt
streaming which can aid in future marketing efforts. Second, this study provided insight
into potential early adopters of future innovations in music distribution including better
understanding of those performance dimensions valued by the new opinion leaders.
Finally, relatively little data exists within academic literature regarding the
business side of music. This study codified and collected important data related to the
16
introduction of music formats, the key competitive dimensions that have traditionally
shaped innovation in music distribution, and the introduction of music streaming. Having
these data appear in academic literature aided music business scholars looking for
background for future research.
Process to Accomplish
This study utilized a mixed-methods research design using quantitative data in the
form of archived sales records as well as qualitative data in the form of historical
documents, including press releases, news items, interviews, and journal articles. The
Nielsen Company, the primary collector and reporter of music sales information in the
United States, provided access to archived sales data under a special license for the
purpose of this study. The sales data included every legal transaction of music in the
United States from 2008 through 2017, including Compact Discs (CDs), digital
downloads, and on-demand streaming. Historical documents used in the study came from
library and public search engines, as well as proprietary documents available to the
researcher as an employee of Sony Music, Inc.
In many ways, this particular study was like a court case. Application of
disruption theory required a very specific sequence of events to occur (Christensen,
2006). This study divided those events into six research questions, each of which required
specific data. As a result, each question was its own miniature study requiring, in many
cases, its own set of data pulled from the various sources listed above.
The data provided by Nielsen was in the form of a massive website containing
data related to music transactions in the United States organized by artist, album, genre,
and format (CDs, digital downloads, on-demand streaming). The data used in this study
17
included sales of CDs, digital downloads, and on-demand streams at the national and
genre levels for the years 2008 through 2017. Genres included in the study consisted of
Pop,
R&B/Hip-Hop, Rock, Country, Latin, Christian/Gospel, Jazz, Dance/Electronic, and
World Music.
Qualitative data used in the study consisted of historical documents, including
press releases, news items, interviews, and journal articles from the period examined.
Because of the broad availability of reliable sources of archival material through the
internet, it has become more common for researchers to use historical data analysis as a
source of primary research (Fischer & Parmentier, 2010). One of the key components of
disruptive innovation theory has to do with consumer opinions and reactions at the time
of the introduction of the innovation (Schmidt & Druehl, 2008). Primary research in the
form of a new study would have required participants to recall how they initially felt
about an innovation introduced almost two decades ago. Archival documents from the
period, which captured the immediate impressions of consumers, the media, established
firms, and entrant firms at that time, provided a more reliable source of opinions, beliefs,
and reactions.
According to MusicWatch (2018), a marketing research and analysis firm focused
on the recording industry, the overall population of music listeners in the United States at
the time of the study consisted of 221 billion people, 55% of which were female and 45%
male. Whites made up 73% of the market, Blacks represented 13%, and other ethnicities
constituted the remaining 14%. MusicWatch tracked music consumption activity for
consumers aged 13 and older. Based on their research, 31% of music consumers were
between the age of 13 and 24. Consumers between 25 and 34 made up 28% of the market
18
and those ages 35-44 were responsible for 18% of the market. The final 23% of the
market consisted of adults 45 and older. MusicWatch (2018) reported these estimates had
a +/- 1.75% margin of error. Because the information supplied by Nielsen for this study
was comprehensive of all music transactions in the United States, the quantitative data
used in this study encompassed the entire population of music listeners in the United
States as opposed to a sample.
Answering the first and second research questions, both of which examined the
performance of on-demand music streaming relative to existing formats in the various
competitive dimensions, required analysis of historical documents related to the features
of each of the primary music playback formats including LPs, cassettes, CDs, digital
downloads, and on-demand streaming. Documents analyzed included official press
releases from entrant firms, journal articles, and media reports from major trade and
consumer publications between 2001 and 2011. The researcher recorded, analyzed and
compared comments and opinions related to the various features and functions of each
format including portability, sound quality, depth-of-offering, price, and the overall
business model.
The third question, which addressed whether the existing mainstream music
market rejected on-demand music streaming, required quantitative analysis of archival
sales data. If the majority of existing music consumers rejected on-demand music
streaming, the introduction of on-demand music streaming would have had little to no
impact on consumption of existing formats. Graphical analysis of sales by format
including CDs, Digital Downloads, and On-Demand Streams for 2008 through 2010
illustrated changes in the relative positions of each format during the period. A Pearson r
19
correlation then demonstrated any possible relationship between changes in consumer
activity relative to existing music formats and the introduction of on-demand music
streaming.
Addressing the fourth question, which asked whether existing low-end consumers
and/or non-music consumers embraced on-demand music streaming in its early stages,
required quantitative analysis of the sales data, this time organized by genre (Pop, Rock,
etc.). The researcher sorted data for each genre according to music format with each
format calculated as a percentage of total music consumption for the genre. The
researcher then identified genres with on-demand streaming as a higher percentage of
overall consumption than that of the overall market as early adopters. Finally, the
researcher examined early adopting genre for any patterns or trends that would indicate
whether on-demand streaming was coming from existing consumers, indicated by a low
willingness to buy, or new consumers, indicated by a low paid-to-non-paid activity ratio.
In addition to quantitative analysis, the researcher also examined qualitative data in the
form of historical media reports and journal articles for any evidence that might have
indicated why consumers were adopting on-demand streaming early.
Examining the fifth question, which asked whether on-demand music streaming
improved over time in the core dimension of portability, required similar analysis to that
needed for questions one and two. Historical documents dated from 2011 and later,
tracked consumer reactions to changes in on-demand streaming platforms. The researcher
then sorted and examined comments and opinions related to the competitive dimensions
previously examined in questions one and two.
20
The sixth and final question, which inquired as to the encroachment of on-demand
music streaming upon the sales of existing formats, required quantitative analysis of
archival sales data from 2011 through 2017. Again, the researcher used graphical analysis
to illustrate weekly sales by format for the period of 2011 through 2017 to examine
changes between the relative positions of each format. A Pearson r correlation analysis
provided evidence of any possible relation between changes in on-demand music
streaming and changes in other formats.
21
CHAPTER TWO
REVIEW OF THE LITERATURE
…innovation is the real driver of progress—Bill Gates
“The enterprise that does not innovate ages and declines. And in a period of rapid
change such as the present, the decline will be fast,” wrote management consultant,
educator, and distinguished author Peter Drucker (1985, p. 183). Managers of firms have
never been under more pressure to successfully introduce new products and ideas into the
marketplace as well as respond quickly to potentially disruptive moves from competitors.
As the common saying goes, “Innovate or die.”
Because of the pressure on business leaders to innovate, much research has gone
into how to innovate successfully. Since the mid-nineties, one of the key voices on the
topic of innovation has been Clayton Christensen, the Kim B. Clarke Professor of
Business Administration at Harvard Business School (Christensen, Raynor, & McDonald,
2015). Christensen’s book, The Innovator’s Dilemma (1997), and supporting articles are
mandatory reading for executive education programs across the United States including
MIT Sloan, Harvard Business School, and the Stanford Graduate School of Business
(Schmidt & Druehl, 2008).
In The Innovator’s Dilemma, Christensen (1997) introduced the theory of
disruptive innovation. A disruptive innovation is one in which the innovation is initially
inferior to existing options in the core competitive dimension most valued by opinion
22
leaders and mainstream consumers. However, the innovation survives in spite of this
inferiority because it is able to attract low-end consumers or brand new consumers by
excelling in a secondary competitive dimension or unique business model. Over time, the
disruptive innovation improves in the core competitive dimension while maintaining its
other advantages until it eventually becomes acceptable to mainstream consumers and, as
a result, encroaches on existing options, often displacing them entirely (Christensen et al.,
2015).
The popularity of Christensen’s work and the provocative nature of the term
disruptive innovation has had the adverse effect of diluting the actual theory as business
leaders and scholars adopted the term without truly understanding the related theory
(Christensen et al., 2015). This has led to an ongoing debate over what actually
constitutes a disruptive innovation as well as a consistent appeal throughout the literature
for more industries to apply disruption theory to specific innovations to clarify and
improve upon the theory (Christensen, 2006; Christensen et al., 2015; Danneels, 2004;
Schmidt & Druehl, 2008). The present study was in response to that appeal.
Determining whether the introduction and rise of on-demand music streaming in
the United States followed the pattern predicted by disruption theory required
examination of certain criteria. This study systematically compared each of those criteria
to archival sales data and historical documents. Specifically, this study tested the
following key questions:
1. Was there a core competitive dimension along which innovation occurred in
music playback formats prior to the introduction of on-demand
streaming?
23
2. Was on-demand streaming initially inferior to existing music playback formats in
the core competitive dimension along which previous innovation had occurred?
3. Did the mainstream music market initially reject on-demand music streaming as a
music playback format?
4. Did on-demand music streaming find acceptance among the low-end consumers
of the existing market or create a new market due to its superior performance in a
secondary competitive dimension or through the introduction of a unique business
model?
5. Did on-demand music streaming improve over time in the core competitive
dimension while maintaining its superiority in some secondary competitive
dimension or through its
unique business model?
6. Did on-demand music streaming eventually encroach upon sales of existing music
playback formats resulting in a shift in the competitive landscape?
Historical Perspective
Innovation, as a business concept, has its roots in the work of economist Joseph
Schumpeter (1928), who, in 1928, began to differentiate between invention and
innovation; concepts previously treated synonymously. In his article, The Instability of
Capitalism, Schumpeter (1928) defined innovation as “putting productive resources to
uses hitherto untried in practice” (p. 378, emphasis in original). Innovation, as
understood by Schumpeter, was a completely separate process from invention and served
purely economic purposes. According to Schumpeter, “It is quite immaterial whether this
[the process of innovation] is done by making use of a new invention or not”. He
continues, “…and even if it [an invention] be involved, this does not make any difference
24
to the nature of the process” (p.378). By 1942, Schumpeter, in his work Capitalism,
Socialism and Democracy (1942), went so far as to describe innovators and innovations
as forces of “creative destruction” (p. 82-83) disrupting stabilized markets and spurring
economic growth.
Schumpeter’s view of innovation changed the way academics, business leaders,
and even governments understood change and competition (Nicholas, 2003). For the first
time, innovation was “understood as a process” (Green, 2013, p. 3) as opposed to the
work of individual geniuses operating alone. Firms began to create research and
development departments, and innovation became a natural part of the business cycle
(Green, 2013).
Firms, focusing on innovation as a process of methodical research and
development, soon realized that “no matter what their advantages, newer technologies are
not adopted by all potential buyers immediately. Rather, a diffusion process is set into
motion” (Norton & Bass, 1987, p. 1069). The fact that innovations spread or diffused
through a population as opposed to achieving simultaneous adoption expanded the
conversation around innovation to include marketing and communications. Everett
Rogers, a specialist in communications and assistant professor of rural sociology at Ohio
State University, began to study how innovations spread through rural communities
leading to the groundbreaking work, Diffusion of Innovations (Dearing & Singhal, 2006).
While not the first to study how innovations spread through social networks, Rogers was
the first to develop a full theory on the subject, a theory still embraced by academics and
practitioners today (Dearing & Singhal, 2006).
25
Through multiple field studies, Rogers (2006) discovered that the diffusion of an
innovation, when plotted cumulatively over time, formed an S-curve. He also found that
diffusion, when plotted on a frequency basis, formed a bell curve. Using the properties of
a normally distributed bell curve, Rogers (2006) divided social networks into five
categories relative to the timing and motivation of their adoption of innovations within a
network: innovators, early adopters, early majority, late majority, and laggards (see
Figure 2.1).
Figure 2.1 Diffusion of innovations over time (s-curve) and by frequency (bell curve)
(Rogers, 2006).
Alkamade and Castaldi (2005) summarized the process, “The diffusion of a new
product, or innovation in a network, often follows a gradual pattern. In the first stage a
few consumers (the innovators or early adopters) adopt, then consumers in contact with
them adopt, then consumers in contact with those consumers adopt, and so forth until the
innovation possibly spreads throughout the network” (p.4). In multiple studies, Rogers
(2003) found the key to the diffusion of an innovation was the area in the bell curve
26
between 10-20% of the population, where the new idea spread from early adopters to
early majority consumers. He realized that early adopters often serve as opinion leaders
within a social network determining what ideas would reach the majority market. “[O]nce
opinion leaders adopt and begin telling others about an innovation, the number of
adopters per unit of time takes off in an exponential curve” (Rogers, 2003, p. 300).
In Diffusion of Innovations, Rogers (2003) identified five attributes of innovations
that influenced whether or not an innovation would successfully diffuse through a
population: 1) relative advantage, or “the degree to which an innovation is perceived as
being better than the idea it supersedes” (p. 229); 2) compatibility, or “the degree to
which an innovation is perceived as consistent with the existing values, past experiences,
and needs of potential adopters” (p. 240); 3) complexity, or “the degree to which an
innovation is perceived as relatively difficult to understand and use” (p. 257); 4)
trialability, or “the degree to which an innovation may be experimented with on a limited
basis” (p. 258); and, 5) observability, or “the degree to which the results of an innovation
are visible to others” (p. 258). The common thread that ran through all of these attributes
was the idea that innovations must somehow tie to the familiar. “Old ideas are the mental
tools that individuals utilize to assess new ideas and give them meaning” (Rogers, 2003,
p. 243).
Ideas that are incompatible with existing practices are often rejected (Rogers,
2003). As researchers and practitioners alike began to perceive the importance of opinion
leaders in getting their innovations to diffuse among their most profitable customers, they
began requiring research and development (R&D) departments to, as Leonard (2006)
wrote, “tie their inventions to the bottom line of the company.” Leonard continued,
27
“Relevance became the mantra for research” (p. 88). In many industries, this resulted in
customer-driven innovation where established firms became captive to the needs of their
largest, most profitable customers; a condition that left established firms blind to all else
and open to attack from new market entrants willing to innovate in other ways
(Christensen, 1997).
Disruptive Innovations
In 1997, Clayton Christensen published a series of articles as well as the book,
The Innovator’s Dilemma, which introduced the idea of disruptive innovation theory. The
innovator’s dilemma, according to Christensen (1997), is that established firms in an
industry often restrict innovation only to those products or services that best serve
existing high-end customers, called mainstream customers. However, “in trying to please
high-end customers with regard to a key performance dimension, an incumbent
eventually develops a product that ‘overshoots’ the performance needs of mid to low-end
customers along that key dimension” (Schmidt & Druehl, 2008, p. 352). This over-
performance leads mid to low-end customers to look to other performance dimensions for
improvement, which opens the door for disruption (Adner, 2002).
Christensen (1997) offered that the reason leading firms often fall to new market
entrants is not due to bad management, lack of technical expertise, or shortsighted vision.
He stated, rather, their failure is due to their ties to an existing value network, or, “the
context within which a firm identifies and responds to customer’s needs, solves problems,
procures input, reacts to competitors, and strives for profit” (p. 32). In other words, they
become captive to their existing customers (Christensen, 1997). The dilemma for
established firms, then, is that evaluating each new innovation against the needs of
28
mainstream customers is exactly what established firms should do, but, in doing so, they
open themselves up to disruption (Christensen, 1997).
Focused on the needs of their mainstream customers, established firms reject or
ignore innovations that are inferior to existing options in the core competitive dimensions
preferred by opinion leaders within the existing market. Sometimes these innovations
survive, though, because they find a place among low-end consumers or an entirely new
market due to a unique business model or to their performance in some secondary
competitive dimension overlooked by established firms. Disruption occurs when these
innovations improve to the point that, in addition to the low-end or new market
customers, the high-end customers start to adopt them as well (Schmidt & Druehl, 2008).
The simple fact that an innovation disrupts or encroaches upon existing products
does not make the innovation a disruptive innovation, as defined by the theory (Schmidt
& Druehl, 2008). It is the path the innovation takes to disruption that matters (Christensen
et al., 2015). Most innovations replace the previous generation of products because they
improve upon those things most valued by existing customers. For example, DVD
players completely displaced VCR players because existing customers preferred the
higher resolution, smaller size, and greater durability of DVDs to VHS tapes (Caron,
2004). However, the DVD player, while disruptive, was not a disruptive innovation. It
was what Christensen terms a “sustaining innovation” (Christensen, 1997, p. 10).
The key difference between a disruptive innovation and a sustaining innovation is
in the adoption process. With a sustaining innovation, the mainstream customers of the
existing market adopt the new product or service from the start because it improves upon
29
the core competitive dimension they prefer. However, with a disruptive innovation, the
early adopters come from the fringes of the existing market or an entirely new market.
Christensen’s (1997) main illustration of a disruptive innovation used throughout
The Innovator’s Dilemma comes from the computer disk drive industry. In the late 1970s
and early 1980s, the core performance dimension most desired in disk drives by the
mainstream market was capacity (Christensen, 1997). And so, during that period, most
innovation driven by established firms focused on improving capacity, sometimes
through incremental improvement, at other times through “radically new technology” (p.
11). Christensen (1997) points out that every major technological innovation in disk drive
technology that progressed along the core dimension of capacity came from established
firms.
The story changed, though, when innovation turned a different direction. Entrant
firms new to the disk drive market began innovating along a different trajectory, the size
of the disk drive. These smaller drives had less capacity than existing models and the
mainstream market had no use for smaller drives with less capacity than their current
options. However, the smaller drives soon found a place in the nascent desktop personal
computer market (Christensen, 1997). Over time, these smaller drives increased in
capacity until the mainstream computer market began adopting them as well
(Christensen, 1997).
The disk drive illustration highlights several key attributes of disruptive
innovations. First, part of what creates the opportunity for a disruptive innovation is
performance oversupply. Adner (2001) states, “Christensen introduces the idea of
‘performance oversupply’ to explain the mainstream consumers’ decision to adopt
30
disruptive technology in the face of superior incumbent technology. The principle of
performance oversupply states that once consumers’ requirements for a specific
functional attribute are met, evaluation shifts to place greater emphasis on attributes that
were initially secondary or tertiary” (Adner, 2001, p. 669). In the case of disk drives, the
capacity of drives eventually exceeded what the existing market needed, which led
consumers to shift their focus from capacity to size (Schmidt & Druehl, 2008).
Second, entrant firms are usually responsible for introducing disruptive
innovations. According to Hunt (2013), “To an increasing degree, there has come to be a
tendency to bifurcate innovation into two contrasting sources: revolutionary
breakthroughs emanating from entrepreneurial firms and incremental enhancements
emanating from large, established incumbents” (p. 151). Because established firms focus
on the immediate needs of their high-end customers, they leave room for smaller new
entrant firms to target the fringes of the market or develop new markets (Christensen,
1997). Hunt (2013) continues, “The essence of this argument rests upon the belief that
innovation stemming from existing sources of knowledge will favor large incumbents.
Meanwhile, nascent-stage firms are expected to excel under circumstances that neither
require nor benefit from established organizational routines” (p. 151).
Third, most disruptive innovations only succeed in moving into the mainstream
market once they have improved in the core dimension sufficient to meet the needs of
mainstream consumers (Christensen et al., 2015). “[W]ith continual upgrading, the new
product eventually becomes acceptable even to the high-end customers of the old
product” (Schmidt & Druehl, 2008). While it is true that established firms, through
performance oversupply, exceed the mainstream market’s needs in the core performance
31
dimension causing the market to shift its focus, there is still a minimum performance
threshold in the core dimension that must be met by a new product or service before the
mainstream can accept it (Christensen, 1997).
Fourth, disruption is usually not the result of a single entrant firm. In a 2015
article for the Harvard Business Review, Christensen et al. (2015) wrote, “What we’ve
realized is that, very often, low-end and new-market footholds are populated not by a
lone would-be disrupter, but by several comparable entrant firms whose products are
simpler, more convenient, or less costly than those sold by incumbents” (p. 10). This was
certainly true in Christensen’s disk drive illustration where multiple entrant firms were
competing in the smaller disk drive space.
Finally, in the same Harvard Business Review article, Christensen et al. (2015)
stated, “Disrupters often build business models that are very different from those of
incumbents” (p. 7). Innovative products and services disrupt the market share of
established firms, but innovative business models erode the established firms’
profitability. For example, the ability to order and stream movies in your home allowed
Netflix to take away market share from video rental companies like Blockbuster.
However, Netflix’s monthly subscription model with no due dates or late fees destroyed
Blockbuster’s entire business model eventually leading to its bankruptcy (Christensen et
al., 2015).
While these attributes of disruptive innovations are often present, they are not all
present in every case. This begs the question, what are the essential elements of a
disruptive innovation that must be present in order for the theory to apply? In other
words, what, exactly, is disruption innovation theory and why does it matter?
32
Disruptive Innovation Theory: An Outline
According to Danneels (2004), “A disruptive technology is a technology that
changes the bases of competition by changing the performance metrics along which firms
compete” (p.249). To elaborate, according to disruption theory, an innovation that fits the
theory not only disrupts existing products and firms, but it does so in a very specific
manner. It changes the dimension along which innovation occurs, for example, shifting
innovation in disk drives from a focus on building capacity to a focus on size
(Christensen, 1997).
For an innovation to be a true disruptive innovation, it must initially be inferior in
the core competitive dimension preferred by the industry’s existing mainstream
consumers (Schmidt & Druehl, 2008). At a time when the core competitive dimensions in
video rental were convenience and selection, Netflix only offered a handful of titles and
required you to wait several days to get a DVD in the mail (Satell, 2014). Netflix was the
opposite of what mainstream video rental customers wanted making Netflix a classic
example of how disruptive innovations seem inferior to existing mainstream customers
when they first enter the market
(Christensen et al., 2015).
Because it is inferior in the core competitive dimension, existing opinion leaders
and mainstream customers reject the disruptive innovation. As a result, established firms
ignore the innovation seeing it as no threat to their business (Schmidt & Druehl, 2008).
Video rental chains and movie studios alike ignored Netflix due to its inferiority to
existing options in convenience and selection. Jeff Bewkes, then CEO of entertainment
giant Time Warner, was quoted by the New York Times as saying, in reference to Netflix
33
posing a threat to existing film distribution models, “It’s a little bit like, is the Albanian
army going to take over the world? I don’t think so” (Arango, 2010, para 3). Reed
Hastings, president and founder of Netflix, in an interview for New Yorker Magazine,
said he wore Albanian Army dog tags for a whole year after the New York Times article
came out (Auletta, 2014).
In spite of rejection by mainstream consumers, a disruptive innovation survives
because it appeals to consumers on the fringes of the existing market or creates a new
market by appealing to them through some overlooked secondary dimension or unique
business model (Schmidt & Druehl, 2008). Netflix found an early base among classic
movie enthusiasts, early adopters of DVDs, and online shoppers who liked choosing their
movies from home. In addition, Netflix’s subscription model, which avoided due dates
and late fees, contributed to its early success as well (Satell, 2014). These marginal
consumers were less valuable to established firms like Blockbuster and so they continued
to ignore Netflix (Auletta, 2014).
Over time, a disruptive innovation improves along the existing market’s core
competitive dimension while maintaining its unique attributes with which it attracted its
initial customers (Schmidt & Druehl, 2008). As broadband internet expanded across the
country, Netflix was able to move into on-demand video streaming giving customers a
broader selection and greater convenience than the local video rental store. Now, with
Netflix, consumers could choose and watch new-release movies immediately without
leaving their home (Christensen et al., 2015). At the same time, Netflix, no longer tied
exclusively to physical DVDs, was able to expand its offering of classic movies and
television programs to satisfy their existing customers as well (Satell, 2014). Finally, the
34
established video market responded, and Blockbuster, seeing the increased threat from
Netflix, tried to enter the online market in 2004 spending over $500 million in an attempt
to dislodge Netflix (Abkowitz, 2009).
Due to the increased performance in both dimensions, the existing market’s high-
end consumers, who had originally rejected the innovation, begin to adopt. The disruptive
innovation, then, encroaches upon the market share of established firms (Schmidt &
Druehl, 2008). Upon introduction of on-demand streaming, mainstream video rental
customers began using Netflix causing a major shift in market share away from
Blockbuster and other video rental chains (Christensen et al., 2015). In addition, Netflix’s
subscription model changed the way consumers thought about paying for video content.
As traditional video rental stores, including Blockbuster, attempted to emulate the
subscription model by modifying existing rental and late fee structures, their overall
business model became untenable leading Blockbuster to declare bankruptcy in 2010
(Satell, 2014).
Netflix was a disruptive innovation, not simply because it led to the bankruptcy of
Blockbuster, but through the means by which it disrupted the video rental market. Netflix
found a fringe market that valued its online interface and its subscription business model.
It then innovated its way into the core of the market through improvement in the core
competitive dimensions of convenience and selection. Most importantly, it shifted the
existing market to a new subscription-based business model, which the established
market tried, but could not imitate without destroying their existing value network
(Christensen, et al., 2015).
35
Redbox, the kiosk-oriented video rental business, however, was not a disruptive
innovation, but a sustaining innovation. Founded in 2002, Redbox was initially a
subsidiary of McDonalds Corporation. Unlike Netflix, Redbox had immediate appeal to
mainstream video renters in the core competitive dimension of convenience. Redbox
offered consumers a relatively small selection of the most popular movie titles through
kiosks located conveniently at local McDonalds, grocery stores, and drug stores. Using
Redbox was fast, simple and, at $1 per night, far cheaper than existing options (Tryon,
2011).
Redbox, while disruptive to traditional video rental stores, was a sustaining
innovation because it immediately improved upon a competitive dimension important to
mainstream video rental customers (Christensen et al., 2015). It was a cheaper, easier,
faster way for mainstream video rental customers to get what they wanted (Tryon, 2011).
Netflix, by contrast, was less convenient at first and, therefore, posed less of a threat to
established firms when first introduced (Christensen et al., 2015).
In addition to disruptive and sustaining innovations, you also have failed
innovations. Keeping with movies, in 2007, Walgreens announced they would be
installing DVD-burning kiosks in their stores. These kiosks would allow consumers to
select older, out-of-print movies and burn them directly to a blank DVD while they
waited (Zeidler, 2007). DVD-burning kiosks were inferior to existing options, as was
Netflix, but, unlike Netflix, they never found a fringe market nor did they create a new
market. In 2008, Redbox announced they had negotiated a major distribution deal with
Walgreens who dropped the kiosk idea before full implementation in favor of the more
proven Redbox model (Reuters, 2008).
36
Disruptive innovation theory does more than simply identify what constitutes a
disruptive innovation. Christensen’s (1997) purpose in The Innovator’s Dilemma was to
help established firms better identify and respond to disruption. The three-fold illustration
of Netflix, Redbox, and DVD-burning kiosks highlights why Christensen’s (1997) theory
of disruptive innovations matters. Christensen (2015) wrote, “The problem with
conflating a disruptive innovation with any breakthrough that changes an industry’s
competitive patterns is that different types of innovation require different strategic
approaches” (p. 4). Each of the above examples required a different response from
market leaders.
According to the theory of disruptive innovations, sustaining innovations require
immediate response from established firms. Sustaining innovations improve upon the
core competitive dimension preferred by high-end consumers and so immediately
threaten the market share and profitability of established firms (Christensen et al., 2015).
Looking at innovation in video rental services through the lens of Blockbuster, Redbox
was the innovation that required the fastest response. While Redbox kiosks had relatively
few titles, the $1 price and convenient locations made them an immediate threat (Tryon,
2011). However, Blockbuster ignored them. As late as 2008, the newly appointed
Blockbuster CEO, Jim Keyes, stated in an interview, “Neither Redbox nor Netflix are
even on the radar screen in terms of competition” (Munarriz, 2008, paragraph 13).
While the ignoring of Netflix by Blockbuster at least made sense, Redbox was a
direct and immediate threat to their core business and should have elicited a response.
Blockbuster was right, though, in ignoring DVD-burning kiosks. The kiosks, with their
slow burn time and old titles, did nothing to encroach upon Blockbuster’s core business
37
of front-list movies and, therefore, were inferior in the core competitive dimension.
Knowing how to respond to various types of innovation is a critical part of strategy and
requires leaders to be able to recognize what sort of threat they are facing (Christensen et
al., 2015).
Innovation in Music Playback Media: A Brief History
Turning from movies to music, in order to examine on-demand streaming of
music in the United States to determine if it performed in accordance with disruptive
innovation theory, it is necessary to recount the evolution of music playback media. If on-
demand music streaming constituted a disruptive innovation, the only way to demonstrate
that fact would be to show how it performed within the context of the history of music
playback media. That being the case, a brief account of that history is included here.
The modern era of sound recording began in 1948 with the introduction of the 12-
inch, 33 1/3 rpm Long-Play Record (LP) by Columbia Records. Prior to the LP, there
were multiple playback media vying to become the standard for all recordings (Shayo &
Guthrie, 2005). Not only did the LP become the standard for its day, it is still the choice
today among a small number of music enthusiasts who prefer the sound quality of the LP
to modern playback options (Coleman, 2003).
For almost twenty years the LP reigned unchallenged until 1964 when Philips
introduced the Compact Cassette (cassette) soon followed in 1965 by the 8-Track Stereo
8 (8-track), created by William Lear, the inventor of the Lear Jet (Coleman, 2003). The
cassette and 8-track formats battled it out for over a decade with 8-tracks initially
winning due to support from the auto industry. Car manufacturers, looking for an audio
system that would allow drivers to bring their own music with them to the car, heavily
38
backed the 8-track format (Coleman, 2003). Eventually, though, the cassette’s advantages
in sound quality, playback length, smaller size, and recordability combined with the
introduction of the highly portable Sony Walkman cassette player, allowed the cassette to
win out over the 8-track even among car manufacturers (Coleman, 2003).
In 1982, through a partnership between Sony and Phillips, the Compact Disc (CD)
entered the market, bringing music playback media into the digital era (Shayo & Guthrie,
2005). The CD was the full package with high-fidelity sound rivalling the LP and greater
portability and durability than the cassette. While playback systems were initially
expensive, as prices dropped and car manufacturers began replacing cassette decks with
CD players, CDs eventually displaced cassettes entirely and relegated LPs to novelty
status (Coleman, 2003).
The next chapter in music playback history took place in Germany where Karl-
Heinz Brandenburg, working for the Fraunhofer Institute, invented the technology that
would eventually make MPEG Audio Layer III (MP3) files possible (Bellis, 2017).
Utilizing digital compression technology, the MP3 allowed the direct exchange of digital
music files without the aid of any physical playback media. Although Brandenburg
introduced the technology in 1991, it took several years to standardize and commercialize
the process. By 1996, standards were set, and the Fraunhofer Institute received a United
States patent for the MP3 (Bellis, 2017). However, in 1997, the core software for MP3
technology was stolen by an Australian college student unleashing a wave of illegal
activity (Albright, 2015).
Following the rise of several unlicensed peer-to-peer MP3 trading sites such as
Napster and MP3.com, United States courts sided with record labels and publishers in
39
declaring such trading of MP3 files as violations of copyright law, however, for every site
taken down, several more took their place (Shayo & Guthrie, 2005). At least partial relief
from this flood of illegal file trading came in 2003, when Apple, Inc. (Apple) opened its
iTunes Music Store; the first commercially viable site for the direct legal sale of digital
music files to consumers, called digital downloads (Coleman, 2003).
Legal digital downloads proved immensely popular, particularly as the iPod
music player became more affordable (Albright, 2015). Even early MP3 players allowed
users to store up to 60 hours of digital music in a player the size of a deck of cards
(Coleman, 2003). Between 2003 and 2013, Apple’s iTunes website dominated the sales
of digital downloads with market share exceeding 60% at times (Bostic, 2013).
Simultaneous with the invention of the MP3, other online music models were also
in development including music streaming. In general, music streaming involved the
playing of music directly from a website as opposed to downloading a file to a hard drive.
From the outset, music streaming developed along two different paths: programmed
streaming and on-demand streaming (Passman, 2015).
Programmed streaming, introduced by sites such as Last.FM (2003) and Pandora
(2005), allowed listeners to choose the style of the music they wanted, but not individual
songs (Passman, 2015). Considered legally as broadcasts, similar to radio, these sites did
not require direct licenses with content owners but paid for songs and recordings through
Performing Rights Organizations (PROs) such as ASCAP, BMI, SESAC and Sound
Exchange. The fact that programmed streaming sites did not require direct licenses with
content providers meant they could offer hundreds of thousands of songs without having
to negotiate with content owners (Passman, 2015). Considered a broadcast as opposed to
40
paid consumption, programmed streaming falls outside the scope of the present study
except as a free alternative to paid platforms much in the same manner as radio airplay.
On-demand streaming, unlike programmed streaming, allowed consumers to
choose the exact artist and song they wished to hear. On-demand sites were required to
secure direct licenses with content owners (Keene, 2016). The need for direct licenses
with content owners created a significant barrier to entry for on-demand streaming sites
and limited the number of titles such sites could initially offer. Because on-demand
streaming was a direct substitute for other music playback media, Nielsen tracked and
reported on-demand streaming as a part of total music consumption in the US (Nielsen,
2018).
The first legal on-demand streaming site to appear in the US was Rhapsody,
introduced by Listen.com in 2001 (Passman, 2015). As opposed to the purchase model
offered by iTunes, where consumers paid for and owned individual song and album files,
Rhapsody users paid a monthly subscription fee to have instant access to all of the songs
in the Rhapsody catalog. While Rhapsody initially only offered subscribers access to
several thousand songs, many of which were from smaller, independent artists and labels,
by 2002, Rhapsody had secured direct licenses with all of the major labels and offered a
catalog of 175,000 songs available for immediate access to subscribers (Evangelista,
2002).
While content owners, including most of the major labels and publishers,
provided licenses to Rhapsody and other on-demand streaming sites, they did so with
several restrictions. Concerned about providing a too-convenient alternative to music
purchases, content owners only allowed sites such as Rhapsody to offer on-demand
41
streaming through a computer connected to the internet. In addition, because on-demand
streaming sites played music directly from a website as opposed to a file stored on the
listener’s device, the speed at which data transferred from the website to the listener’s
computer had a big impact on the music streaming experience. In 2001, at a time when
most online users were restricted to dial-up access to the internet, on-demand streaming
could often prove very frustrating.
On-demand streaming, in its initial format as offered by platforms like Rhapsody,
could not compete with the dramatic launch of Apple’s iTunes platform. The iTunes
digital download model, with its emphasis on music ownership, ease of use, and ability to
play on multiple devices both online and offline, proved more popular than the more
innovative but restrictive on-demand streaming approach. While there was a fringe
market of consumers interested in the subscription model, the on-demand streaming
platform remained a novelty leading Apple founder Steve Jobs to famously state, “People
have told us over and over and over again, they don’t want to rent their music” (Ricker,
2015, para. 3).
By 2008, seven years after the launch of Rhapsody and the first year in which
Nielsen began tracking and reporting on-demand streaming as a form of music
consumption, on-demand streaming accounted for less than 1% of total music
consumption in the US (Nielsen, 2018). In that same year, digital downloads accounted
for 31% of total music consumption. In the first round between digital downloads versus
on-demand streaming, the digital download was the clear winner.
In 2011, a new wave of on-demand streaming sites entered the US, led by Spotify,
a Sweden-based company that had been making headlines in Europe since 2008. Through
42
a lengthy negotiation process with US music content owners, Spotify was able to offer
something no other on-demand streaming platform had before, a free on-demand
streaming subscription. Utilizing a business model built on the sharing of advertising
revenue with content owners, called ad-supported streaming, Spotify provided users with
free access to a catalog of 15,000,000 songs (Sisario, 2011).
While content owners insisted upon several restrictions on Spotify’s free ad-
supported subscription users, including limiting the service to computer use only and a
monthly 10-hour cap on free listening, the service grew rapidly surpassing 20,000,000
users by 2013 (Spotify, 2013). Not only did free on-demand subscriptions grow rapidly,
Spotify’s paid subscriptions grew as well, mostly through free users converting to a paid
subscription in order to take advantage of additional features such as ad-free, unlimited
access on computers and mobile devices both online and offline (Ingham, 2016).
Spotify’s entry into the market was soon followed by on-demand streaming
platforms from Tidal, Amazon, Apple, Pandora and YouTube. By the end of 2017, on-
demand music streaming accounted for 65% of all music consumption in the US. In that
same period, digital downloads fell to less than 20% of consumption. On-demand
streaming is now the most dominant music playback platform since the peak of the CD at
the beginning of the 2000s (Nielsen, 2018).
Digital Downloads: A Sustaining Innovation
While it falls outside the present study to conduct an in-depth inquiry into the
introduction and rise of digital downloads as a music playback format, an abbreviated
analysis of digital downloads provides an excellent background for the inquiry into on-
demand music streaming. As was stated earlier, the exchange of digital downloads began
43
in the late 1990s, mostly through illegal websites such as Napster. As this study is
concerned only with legal transactions, this discussion regarding digital downloads
begins with the launch of iTunes in 2003.
The key concept in disruptive innovation theory is that a disruptive innovation is
initially inferior in the core competitive dimension preferred by mainstream consumers.
When innovations are inferior in the core competitive dimension, mainstream consumers
reject them (Schmidt & Druehl, 2008). However, mainstream consumers embrace
innovations that outperform existing options in the core competitive dimension.
Christensen (1997) labels these latter forms of innovation as sustaining innovations.
As this study will later demonstrate, the core competitive dimension in music
playback formats was portability with each successive innovation, from the LP to the CD,
improving upon the portability of music (Albright, 2015). For digital downloads to have
been a disruptive innovation, they would have been less portable than the CD, the
standard playback format at the time. However, that was not the case as digital
downloads were more portable than any format at the time (Lazrus, 2016).
Digital downloads, as electronic files, were easier to store, easier to transport, and,
to the chagrin of content owners, easier to transfer to others (Plambeck, 2010). A simple
MP3 player could store the entire music collection of an average listener in a device the
size of a deck of cards (Flanagan, 2013). Digital downloads were, and many would say
still are, the most portable music playback format available.
That said, while digital downloads were superior in the core competitive
dimension, they never fully replaced the CD. In fact, by 2011, a full decade after the
appearance of iTunes, digital downloads, whether in the form of full albums or individual
44
songs, only accounted for 49% of the sales of music (Nielsen, 2018). To understand why
this might have been the case requires a return to diffusion theory.
In his classic work, Diffusion of Innovations, Rogers (2003), identified five
attributes of innovations that influence diffusion. While disruption theory largely emerges
from the attribute of compatibility, the remaining attributes are always at play as well.
One of the attributes at play with the introduction of digital downloads was that of
complexity.
Complexity, defined by Rogers (2003) as “the degree to which an innovation is
perceived as relatively difficult to understand and use” (p. 257), was an important
element in the adoption of digital downloads due to the shift from tangible to intangible
media. From the LP to the CD, consumers purchased a tangible item they could hold,
carry, lend, and store; something they could possess. With the digital download,
consumers received an intangible computer file without packaging, album art, or liner
notes. There were also limitations on the rights of ownership, particularly in the area of
sharing music with others (Lazrus, 2016). This shift from the tangible to the intangible
added complexity to the adoption process for the average music consumer.
In addition, while the free iTunes software made it easy to download, play, and
store digital downloads using a typical computer or laptop, fully taking advantage of the
enhanced portability features offered by digital downloads required the purchase of a
designated MP3 player. As with the adoption of any new media format, adoption of
digital downloads required adoption of the necessary hardware. Initial attempts at
creating portable MP3 players were clunky, expensive, and difficult to use. However,
45
with the introduction of the iPod in 2001, Apple once again smoothed the way for the
adoption of digital downloads (Flanagan, 2013).
Another critical factor to consider in the introduction of digital downloads was the
unbundling of the album. Prior to digital downloads, most music playback formats
consisted of bundles of ten to twelve songs referred to as albums. If consumers heard a
song they liked, they were often required to buy the full album in order to get the song.
While some single songs were available on CD, by the time digital downloads released,
there were very few on the market.
One of the key features of the iTunes store, when it first launched, was the
unbundling of albums, which allowed consumers to buy only the song they wanted as
opposed to being required to purchase the full album (Elberse, 2010). This ability for
consumers to purchase one song at a time somewhat masked the adoption of digital
downloads due to the way Nielsen accounted for music sales. Music sales charts created
by Nielsen relied on album equivalents with ten single-song digital downloads counting
as an album. While this method allowed comparison between CDs and digital downloads,
it distorted consumer activity in that one consumer buying a CD had the same impact on
the charts as ten consumers buying individual song digital downloads. What appeared as
a 1:1 relationship between CDs and digital downloads was actually a 1:10 relationship
when examined at the consumer behavior level. This distortion of consumer behavior
somewhat obscured the rise of downloads as a music playback format (Nielsen, 2018).
One final factor in looking at digital downloads is the underlying business model.
The business model offered through iTunes was a direct continuation of the retail
purchase model used in the sales of music media since recorded music first became
46
commercially available. This continuation of the purchase model was a critical factor in
calling digital downloads a sustaining innovation as opposed to a disruptive innovation.
The critical point to take away from this brief look at digital downloads is that
when an innovation enters an existing market; there are many factors at play (Rogers,
2003). Conditions apparently caused by a lack of continuity may actually be a
combination of complexity, trialability, or one of the other attributes of innovations that
affect adoption. This is partially why identifying disruptive innovations is so difficult
(Christensen et al., 2015).
Conclusion
In analyzing whether an innovation is a true disruptive innovation, the main thing
to keep in mind is that a disruptive innovation shifts the basis of competition (Danneels,
2004). Netflix shifted the video rental business to a monthly subscription model
(Christensen et al., 2015). Digital downloads, though, competed and survived by
excelling in the existing core competitive dimension of portability. That is why Netflix
was considered a disruptive innovation, and digital downloads were not.
In studying on-demand music streaming, it is important to isolate consumer
behavior using both quantitative and qualitative data to uncover not only what happened,
but also why it happened. Was on-demand music streaming inferior in regards to
portability? Did the mainstream consumers initially ignore on-demand music streaming?
Finally, has on-demand music streaming caused a shift in the competitive dimension
most favored by consumers? These are the questions this study hopes to address.
47
CHAPTER THREE
METHODOLOGY
It is critically important that managers be able to recognize a disruptive
innovation when they see one. –Glenn Schmidt
A review of the literature surrounding innovation in general and, more
specifically, disruptive innovation, reveals much confusion and misinformation regarding
what constitutes a true disruptive innovation. While all innovation is to some extent
disruptive, a disruptive innovation, as defined by the theory put forward by Clayton
Christensen in his book, The Innovator’s Dilemma (1997), follows a very specific
pattern. It is how an innovation disrupts, not just the simple fact that disruption occurred
that matters (Schmidt & Druehl, 2008).
A true disruptive innovation, according to Daneels (2004), “changes the bases of
competition by changing the performance metrics along which firms compete” (p. 249).
Netflix shifted the home video market from a purchase model to a subscription model
(Satell, 2014). Airbnb provided low-cost accommodations for low-end travelers through a
network of private homeowners and then advanced into high-end tourist rentals
(Guttentag, 2015). Online greeting cards, digital photography, smart phones, and
ultrasound technology were all disruptive innovations when introduced (Christensen,
1997). In short, disruptive innovations shift the competitive basis within an industry from
one competitive dimension to another or from one business model to another by
48
innovating along a different trajectory than that upon which previous innovation had
occurred (Schmidt & Druehl, 2008).
The present study is concerned with the question as to whether on-demand music
streaming performed in accordance with Christensen’s (1997) theory of disruptive
innovation. In order for on-demand music streaming to be a disruptive innovation, certain
criteria must apply. The following research questions summarize those criteria:
1. Was there a core competitive dimension along which innovation occurred in
music playback formats prior to the introduction of on-demand streaming?
2. Was on-demand streaming initially inferior to existing music playback formats in
the core competitive dimension along which previous innovation had occurred?
3. Did the mainstream music market initially reject on-demand music streaming as a
music playback format?
4. Did on-demand music streaming find acceptance among the low-end consumers
of the existing market or create a new market due to its superior performance in a
secondary competitive dimension or through the introduction of a unique business
model?
5. Did on-demand music streaming improve over time in the core competitive
dimension while maintaining its superiority in some secondary competitive
dimension or through its unique business model?
6. Did on-demand music streaming eventually encroach upon sales of existing music
playback formats resulting in a shift in the competitive landscape?
49
Research Design
This study utilized a mixed-methods research design, drawing from both
quantitative and qualitative archival documents and data sets. Examining the patterns of
innovation within an industry requires, as Christensen (1997) stated, “carefully
reconstructing the history of each technological change in the industry [so that] the
changes that catapulted entrants to success or that precipitated the failure of established
leaders could be identified” (p. 8). In order to reconstruct both what happened and why it
happened with the introduction and rise of on-demand music streaming, the researcher
required quantitative analysis of archival sales data as well as qualitative analysis of
historical documents in the form of corporate reports, official press releases, marketing
materials, news items, journal articles, interviews, and surveys.
Analysis of archival data is a common means of quantitative research. Fielding
(2004) pointed out that such analysis “informs many academic debates, much policy
analysis, and, though largely unpublished, the business decisions of many companies” (p.
98). In order to know what happened with an innovation, one needs to examine the
evidence left behind by archival records of transactional data.
The challenge in using this approach was gaining access to industry-wide
transactional data as opposed to data from only one company, or relying on second-hand
summary data from trade organizations or the media (Hand, 2018). This was because
private reporting agencies that archive industry-wide transactional sales data usually only
grant access to such data to qualified insiders within the related industry. In most cases,
only researchers willing to pay considerable fees can access transactional data of the kind
necessary for detailed analysis (Hand, 2018).
50
For the purpose of this study, The Nielsen Company (Nielsen), the primary
collector and distributor of entertainment marketing data, granted the researcher a special
license allowing access to their industry-wide sales data archives for music within the
United States (US). This data included weekly sales of every music playback format,
including physical albums (CDs, LPS), digital downloads, and on-demand music streams
from 2008 through 2017. This rare level of access to Nielsen’s archives for the purpose of
research allowed this study to examine on-demand music streaming using industry-wide,
weekly sales data of the kind necessary for the type of study recommended by disruptive
innovation scholars.
While the use of archival data was what Fielding (2004) referred to as “a well-
established practice in quantitative social research,” (p. 98), the use of archival
documents as a primary source for qualitative research was not as common. Historically,
in qualitative research, document analysis served only a supporting role to primary
research collected through interviews, focus groups, and observational data. However,
there has been a growing acceptance of archival material as a primary source due to, as
stated by Fischer and Parmentier (2010), “increasingly sophisticated critiques emerging
of interview data as a primary resource in qualitative research” (p. 799).
Interviews and surveys, long the preferred approach by most qualitative
researchers, have come into question by studies that show “disjunctions between what
people say and what people do” (Fischer & Parmentier, 2010, p. 799), or, in reference to
past events, what people remember and what actually occurred. When looking at past
events, such as consumer attitudes regarding an innovation introduced two decades
earlier, secondary analysis of archival documents may have, as Fielding (2004) described,
51
“a claim to greater plausibility since it is less likely that the analytic interests employed
will have played a part in the interactional field from which they were derived” (p. 100).
In other words, archival documents capturing consumer attitudes regarding on-demand
music streaming at the time of its introduction were not as subject to researcher bias.
Bowen (2009) defined document analysis as “a systematic procedure for
reviewing or evaluating documents—both printed and electronic (computer-based and
Internet transmitted) material” (p. 27). He went on to state that “documents may be the
most effective means of gathering data when events can no longer be observed or when
informants have forgotten the details” (p. 31). In attempting to reconstruct the history of
on-demand streaming, archival documents were a necessity. Corporate annual reports,
press releases, advertisements, magazine articles, archived corporate research, news
items, and journal articles all provided needed information towards answering the
research questions.
One final argument for the use of archival documents as a primary source for
research had to do with triangulation. Referring to the use of archival documents from
various sources, Fielding (2004) commented, “The activity may be useful in evaluating
the generalizability of findings from qualitative research by different researchers on
similar populations” (p. 98). By using archival documents, the researcher was able to
assemble multiple views generating a broader picture than would be possible by focusing
solely on one’s own qualitative research.
The overall design of this study followed, in many ways, the structure of a court
case, in which distinct criteria needed to be present in order for circumstances to fit the
theory of disruptive innovations. Each of the research questions represented one of those
52
criteria. As such, the researcher approached each question individually using different
combinations of archival data as needed. This method deviated from the typical research
project in which one master set of data is gathered and used to address a group of
questions. In other words, the overall question of whether on-demand music streaming
was a disruptive innovation was addressed one criteria at a time through the individual
research questions, each contributing to a final analysis in the end.
Participants
As stated earlier, Christensen (1997) described his method for examining
disruption within an industry as “reconstructing the history of each technological change
in the industry” (p. 8). This approach necessitated assembling data at the industry level as
opposed to a subset population within an industry. The data made available for the
present study included every commercial transaction of music within the US from 2008
to 2017. As global information was not available, this study was limited to examining
data for the US only.
The population for this study included all music consumers within the US from
2008 through 2017. MusicWatch (2018), a research company dedicated to music industry
market research and analysis, described the average overall music buying population
during the period as consisting of 221,000,000 people, 56% of which were female.
Whites made up 72% of the market, while Blacks represented 12%, and other ethnicities
constituted the remaining 16%.
MusicWatch (2018) only collects information on consumers older than 13. Based
on their research, 34% of music consumers tracked were between the ages of 13-24, 28%
53
were 25-34, 16% were 35-44, and the final 22% consisted of adults 45 and older. These
age estimates were reported with a +/- 1.75% margin of error. (See Table 3.1)
Table 3.1
U.S. Music Consumers
Age 13-24 34%
25-34 28%
35-44 16%
45+ 22%
Gender Male 44%
Female 56%
Ethnicity White 72%
Black 12%
Other 16%
Note . Reported accuracy of +/- 1.75
(MusicWatch, 2019)
Because reconstructing the history of technological change within the music
industry required examining not only what happened but also why it happened, the
researcher needed to find a way to examine consumer behavior at the transactional level.
To answer the research questions, it was necessary to know what kind of consumer first
adopted on-demand music streaming. Because Nielsen does not capture individual
consumer data at the time of transaction, another means of identifying consumer behavior
had to be established.
54
A further complication in analyzing consumer behavior related to music
consumption was the fact that the various music playback options were not mutually
exclusive. An individual consumer could purchase physical albums, buy downloads, and
use on-demand streaming all at once. This made isolating individual consumer behavior a
challenge.
Given these same challenges, previous researchers have turned to music genre as
a means of grouping and analyzing consumer behavior (Montoro-Pons & Cuadrado-
Garcia, 2016). Genre in music has been defined by Lena and Peterson (2008) as,
“systems of orientations, expectations, and conventions that bind together an industry,
performers, critics, and fans in making what they identify as a distinctive sort of music”
(p. 698). A genre, then, includes not only those who produce the music, but also the fans
who listen (Lena & Peterson, 2008). As such, genres, as Montoro-Pons and Cuadrado-
Garcia (2016) stated, “help in identifying specific preferences…that could be useful in
clustering individuals in specific social spaces, from which a deeper understanding of
consumer habits and behavior can be derived” (p. 3).
The data provided by Nielsen included not only industry-wide transaction data but
also breakout data by genre for the period examined. While some scholars account for as
many as 60 different music genres (Lena & Peterson, 2008), Nielsen tracks 16 core
genres. Of those 16 genres, this study examined nine core genres, including Pop,
R&B/Hip-Hop, Rock, Country, Latin, Christian/Gospel, Jazz, Dance/Electronic, and
World Music. This study excluded genres considered historical, non-commercial, or
novelty in nature, including Blues, Classical, Children, Comedy, Holiday/Seasonal, and
New Age. This study also excluded the Other category, a collection of uncategorized
55
artists and songs unlikely to exhibit the group behavior upon which the concept of genre
is based (Lena & Peterson, 2008). For future reference, the omission of these genres
meant that later genre-level analyses related to market share resulted in numbers that
equaled less than 100% of the total market.
To summarize, questions dealing with what happened in the introduction of on-
demand music streaming used comprehensive industry-wide data encompassing every
music consumer in the US from 2008-2017. Questions dealing with consumer attitudes
and behaviors regarding on-demand music streaming examined genre-specific
transactional data. The combination of the two approaches provided a broad picture of
consumer actions and attitudes.
Data Collection
Nielsen collects weekly transactional sales data for all music purchases within the
US. Retail outlets and online music providers such as Spotify, Apple, Amazon, and
Pandora report transaction data to Nielsen on a weekly basis through electronic
transmission. Nielsen converts the raw data and compiles various sales charts and
analytical tools available to their clients through their proprietary website and database.
Through a special license, the researcher accessed Nielsen’s archival database that
included weekly music transactions in the US from 2008 through 2017. The site featured
a report generation tool that provided access to industry-wide data or subsets of data
based on music genre. Table 3.2 demonstrates the format of the raw data before
preparation.
56
Table 3.2
Raw National Sales Data for the first two weeks of 2017
Week 1 2017 Week 2 2017
Total
Albums w/TEA w/SEA On-Demand 11,032,793 10,343,343
By Format
Physical Albums Sales 1,910,260 1,586,799
Digital Albums Sales 1,525,309 1,223,565
Digital Song Sales 14,261,486 12,649,377
Streaming On-Demand 9,256,613,749 9,402,061,633
(Nielsen, 2018)
Lucko and Mitchell (2010) referred to the process of data preparation as
“systematically collating and transforming unformatted, unconnected, or otherwise
initially unusable data into a consistent and coherent data set” (p. 49). The data initially
provided by Nielsen reported non-equivalent sales units that needed conversion into
equivalent units for comparison purposes. As the primary generator of sales charts and
marketing data, Nielsen established a standard for converting non-album purchases into
album equivalents. As would be expected, downloaded albums convert into albums at a
1:1 ratio. Downloaded individual songs convert into albums at a 10:1 ratio (Track
Equivalent Albums-TEA) with on-demand streams converting into albums at a 1,500:1
ratio (Stream Equivalent Albums-SEA) (Nielsen, 2018).
Converting all sales data into album equivalents allowed comparative analysis of
nonequivalent measures. The converted data, when added together, produced an album-
57
equivalent total permitting comparative analysis among the different music playback
formats. Table 3.3 shows what the data above looked like after conversion.
Table 3.3
Converted National Sales Data for the first two weeks of 2017
Week 1 2017 Week 2 2017
Total
Albums w/TEA w/SEA On-Demand 11,032,793 10,343,343
By Format
Physical Albums Sales 1,910,260 1,586,799
Digital Albums Sales 1,525,309 1,223,565
Digital Song Sales (converted) 1,426,149 1,264,938
Streaming On-Demand (converted) 6,171,076 6,268,041
(Nielsen, 2018)
Collection of qualitative data in the form of historical documents took place
through various library search engines as well as online search engines such as Google
Scholar. Historical documents generated for analysis included corporate reports, official
company press releases, archived marketing materials, journal articles, news reports, and
archived interviews with representatives of the industry. Where possible, peer reviewed
sources were used to triangulate data from non-academic sources such as new articles and
media interviews.
58
Analytical Methods
Quantitative analysis of archival sales records started with a graphical analysis of
sales by music format. Variables included sales of physical albums (CDs, LPs), full
digital albums, digital songs, and on-demand streams. The researcher graphed the data
using both the actual numbers as well as by percent of the weekly total (i.e., physical
albums as a percent of the album equivalent total for the week, etc.).
In addition to graphing the data, the researcher conducted a Pearson r correlation
analysis among the various playback formats to determine if there was any relationship
between shifts in activity between CDs, downloads, or on-demand streams. The
researcher used weekly transaction data sorted by playback format for the period
examined. The researcher experimented with more sophisticated time series analysis, but
none enhanced the results beyond what the Pearson r correlation revealed.
Qualitative analysis involved historical document analysis of both public and
private records, including corporate reports, official company press releases, archived
marketing materials, journal articles, news reports, and archived interviews with
representatives of the industry. The researcher collected documents specific to each
research question, compiling, classifying, and analyzing them in order to address the
problem stated within each question. The data collected allowed the researcher to, as
Bowen (2009) stated, “Understand the historical roots of specific issues and…indicate the
conditions that impinge upon the phenomena currently under investigation” (p. 30).
To answer questions one and two, the researcher needed to compare music
playback formats across several key competitive dimensions, including sound quality,
price, depth of offering, portability, as well as the relative business models involved. The
59
researcher examined documents such as official press releases, new reports, and platform
reviews in the media related to the various formats including LPs, cassettes, CDs, digital
downloads, and on-demand streaming. The researcher then compared the performance of
each format in the various competitive dimensions.
Question three, which addressed whether mainstream consumers rejected on-
demand streaming when it first appeared, required a mixed-method approach.
Quantitatively, the researcher pulled archived sales data sorted by playback format,
including CDs, digital download albums, digital download songs, and on-demand music
streaming, for the year 2008, the earliest period for which Nielsen tracked on-demand
music streaming. The researcher graphed and analyzed the transactional data for the
period. The researcher also conducted a Pearson r correlation analysis among the various
formats to determine the potential relationship between changes in on-demand streaming
and the other formats. The researcher also collected and analyzed qualitative archival
data related to early consumer opinions to triangulate conclusions drawn from the
quantitative analysis.
Question four was a two-part question, the first part of which examined whether
on-demand music streaming survived due to acceptance by the low-end of the existing
market or the establishment of a new market. To answer this part of the question required
the identification of low-end and non-consumers of music. In the literature surrounding
the theory of disruptive innovations, researchers seeking to identify low-end consumers
typically created a continuum based on a consumer’s willingness to pay for a product or
service (Schmidt & Druehl, 2008). High-end or mainstream consumers were those with
the highest willingness to pay for a product or service, while low-end consumers
60
exhibited less of a willingness to pay. Non-consumers, by extension, were those
unwilling to pay for a product or service (Schmidt & Druehl, 2008).
There were several challenges in examining willingness to pay when applied to
music consumers. First, there was a lack of data at the individual consumer level. In
addition, in music, willingness to pay could shift among individuals based on the specific
artist or song involved. Finally, music formats were not mutually exclusive, so the same
music consumer could purchase physical albums or downloads, stream songs, or listen to
the radio all at once (Montoro-Pons & Cuadrado-Garcia, 2016). Because of these
challenges, academic studies involving the music industry used genres as a means of
grouping and analyzing consumer behavior
(Montoro-Pons & Cuadrado-Garcia, 2016).
Using sales data provided by Nielsen sorted by genre, the researcher calculated
the average percent of total consumption by music format including CDs, downloaded
albums, downloaded songs, and on-demand music streaming for the years 2008-2010.
Averaging data across multiple years minimized the potential skewing of data by major
blockbuster releases in any one year. By calculating the ratio of full album purchases
relative to total consumption, the researcher identified those genres with a full-album
purchase ratio lower than that of the total market. Because full-album purchases cost
considerably more than individual song purchases or streaming, the researcher considered
genres with a low album-purchase ratio to be the low-end of the existing market
(Montoro-Pons & Cuadrado-Garcia, 2016).
Establishing non-consumers of music required an examination of free media in
relation to paid activity by genre. The two main legal free choices for music listeners
were radio and programmed streaming, both of which Nielsen tracked and reported for
61
the period. The researcher used quantitative analysis of radio audience by genre as well
as programmed streaming transactions, again averaged across 2008-2010, to create a
paid-to non-paid activity ratio. Because radio audience numbers (r) and programmed
streaming numbers (ps) were several multiples higher than consumption numbers (c), the
formula c ÷ ((r ÷ 100) + (ps ÷ 100)) was used to develop a paid-to-non-paid activity ratio.
Genres with a lower paid-to-non-paid activity ratio than the total market represented
behavior reflective of non-consumers of music.
The final step in answering the first part of question four required identifying
genres that adopted on-demand music streaming at a rate faster than the overall market.
To do this, the researcher calculated streaming as a percentage of overall consumption by
genre. Once again, the researcher averaged the data across the years 2008-2010 to
minimize the impact of any one major blockbuster release. The researcher then compared
those genres with an adoption rate higher than the overall market with those genres
earlier identified as low-end or non-music consumers to determine if early adopters were,
in fact, low-end or non-music consumers.
The second part of question four asked whether early adopters switched to on-
demand streaming due to superiority in some secondary competitive dimension or
through the introduction of a new business model. The researcher used qualitative
analysis of historical documents to identify why early adopters chose on-demand music
streaming. This included an analysis of on-demand music streaming in each of the
competitive dimensions, including portability, sound quality, price, and depth of
selection. In addition, the researcher examined the underlying business model of on-
demand music streaming relative to existing formats.
62
Question five, which examined ongoing performance of on-demand streaming
platforms in relation to the key competitive dimensions of sound quality, price, depth of
offering, and portability, required qualitative analysis similar to that used for the first two
questions. Archived records, marketing materials, news stories, and consumer reports
provided insight into changes in the platforms over time as well as consumer reactions to
those changes.
Also used in addressing question five was a survey by MusicWatch (2018), a
music industry market research and analysis company that examined consumer
preferences among various features of on-demand music streaming services. Conducted
in February of 2018, MusicWatch surveyed a random selection of 2,495 on-demand
streaming service users, examining consumer preferences among various features of the
top on-demand music streaming services such as Spotify, Apple Music, Amazon
Unlimited, and Google Play Music among others (MusicWatch, 2018). In addressing
question five, the researcher identified specific issues related to the core competitive
dimensions from the MusicWatch report for analysis.
Finally, question six, which examined potential encroachment on existing formats
by on-demand streaming, required quantitative analysis of archived sales records. The
researcher graphed and analyzed the relative performance of each format for the period
between 2011 and 2017. The researcher also conducted a Pearson r correlation analysis
among the various formats for the period, including CDs, downloaded albums,
downloaded songs, and on-demand music streaming to examine and analyze changes
between the formats.
63
CHAPTER FOUR
FINDINGS AND CONCLUSIONS
Learning and innovation go hand in hand. The arrogance of success is to think that what
you did yesterday will be sufficient for tomorrow.—William Pollard
As business theories go, the theory of disruptive innovations is relatively new,
having emerged in the mid-1990s through the work of Clayton Christensen (Christensen,
1997). Christensen (2006), in his article “The ongoing process of building a theory of
disruption,” suggested that the testing of a theory improves the theory. Christensen stated,
“The deductive portion of a complete theory-building cycle can be completed by using
the model to predict ex post what will be seen in other sets of historical data” (p. 45). The
purpose of this study, therefore, has been to contribute to the field of disruption theory
through the testing of the model against the historical data related to the introduction and
subsequent rise of on-demand music streaming.
In short, disruptive innovations are inferior to existing options in the core
competitive dimension upon which previous innovation has occurred. The innovation
survives, however, because the low-end of the existing market or an entirely new market
prefers its superiority in some secondary competitive dimension or its unique business
model. Over time, the innovation improves in the core competitive dimension while
maintaining its other advantages to the degree that the mainstream of the existing market
begins to shift to the innovation thus disrupting existing options.
64
In order to test the model of disruptive innovation against the introduction and rise of
on-demand music streaming, the researcher addressed a series of questions, namely:
1. Was there a core competitive dimension along which innovation occurred in
music playback formats prior to the introduction of on-demand streaming?
2. Was on-demand streaming initially inferior to existing music playback formats in
the core competitive dimension along which previous innovation had occurred?
3. Did the mainstream music market initially reject on-demand music streaming as a
music playback format?
4. Did on-demand music streaming find acceptance among the low-end consumers
of the existing market or create a new market due to its superior performance in a
secondary competitive dimension or through the introduction of a unique business
model?
5. Did on-demand music streaming improve over time in the core competitive
dimension while maintaining its superiority in some secondary competitive
dimension or through its unique business model?
6. Did on-demand music streaming eventually encroach upon sales of existing music
playback formats resulting in a shift in the competitive landscape?
Findings
Using both quantitative and qualitative research methods, the researcher examined
archival records as well as historical sales data for the period between 2001 and 2017 in
order to reconstruct what happened through the introduction of on-demand music
streaming in the United States and how music consumers adopted on-demand music
streaming. The following presents the findings from this research.
65
Question 1: Was there a core competitive dimension along which innovation
occurred in music playback formats prior to the introduction of on-demand
streaming?
Sound quality, price, depth of offering, and portability have all been key
competitive dimensions in music playback media. In an examination of archival
documents relating to various media, from the LP to the digital download, only
portability improved with each successive innovation Table 4.1 illustrates how each
playback media compared along the various performance dimensions.
Table 4.1
Playback Media Performance
LPs Cassettes CDs Downloads
Sound Quality Strong Weak Strong Moderate
Depth of Offering Weak Weak Weak Strong
Price High Medium High Low
Portability Weak Strong Strong Strong
Sound quality would have seemed an obvious choice as a preference among
opinion leaders and high-end consumers, but most experts agree that cassettes and digital
downloads were actually inferior to either the LP or CD (Plambeck, 2010). Moreover,
while digital downloads were available in a high quality format that many experts say
rivalled the CD or LP, most consumers opted for smaller MP3 files or the cheaper version
of the Advanced Audio Coding (AAC) files used by iTunes, which have been sonically
proven to be inferior to both the CD and LP (Plambeck, 2010).
66
Another competitive dimension that often drives innovation is price. Retailers
priced cassettes similarly to LPs, but CDs were more expensive than cassettes and LPs
(Hochman, 1990). Digital downloads, though, when first introduced, were significantly
cheaper than CDs. When Apple launched the iTunes store in 2003 with single-song
downloads at 99 cents and full albums for $9.99, the average new release on CD cost $18
(Griggs & Leopold, 2013). Price as a core competitive dimension would have required
each innovation to drive prices downwards, and that was not the case.
Depth of offering refers to the amount of content available in each new playback
media upon introduction. With each successive innovation, the number of titles available
was relatively small with only 49 titles debuting on cassette when it first emerged
(Billboard, 1966) and 50 titles on CD for its initial launch (LEM, 2014). However, as
each new format gained popularity and playback hardware supporting the format diffused
through the population, labels released more and more content. Digital downloads were
the exception. According to an official press release from Apple, there were over 200,000
songs available when it launched the iTunes store in 2003 (Apple, 2003). While this
created an advantage for digital downloads, for depth of offering to be the core
competitive dimension, each successive innovation (cassette, CD, and digital download)
would have shown improvement along this dimension and, again, that was not the case.
The final competitive dimension considered was portability, or the ability for the
consumer to take their music with them. While there were attempts to make the LP
portable, the resulting technology was impractical and only adopted by a fringe market of
consumers (Gopinath & Stanyek, 2014). Cassettes (and 8-tracks), however, were
specifically created to provide the portability LPs lacked (Coleman, 2003). CDs provided
67
even more portability due to their thinness and greater durability over the cassette. CDs
also provided a significant increase in sound quality, which allowed them to become the
dominant music playback format by the late 1980s (Coleman, 2003). Digital downloads
provided a vast improvement in portability over all previous formats. With digital
downloads, you could carry your entire music catalog with you in the palm of your hand.
Music easily moved from desktop, to laptop, to portable player, and a download could
not be scratched or lost (Plambeck, 2010).
Portability, the ability to listen to music anywhere and anytime, was the core
competitive dimension driving innovation in music playback media since the 1960s
(Gopinath & Stanyek, 2014). Looking at the LP, cassette, CD, and digital download, each
showed an immediate improvement in performance over its predecessor in regards to
portability, often at the expense of sound quality, price or depth of offering. Plambeck
(2010) stated, “In one way, the music business has been the victim of its own
technological success: the ease of loading songs onto a computer or an iPod has meant
that a generation of fans has happily traded fidelity for portability and convenience”
(paragraph 6).
Question 2: Was on-demand streaming initially inferior to existing music playback
formats in the core competitive dimension along which previous innovation had
occurred?
The core competitive dimension along which music playback formats historically
innovated was portability. Consumers continually favored and supported playback
options that provided the greatest freedom to take their music with them (Gopinath &
Stanyek, 2014). In examining on-demand music streaming, then, it was necessary to
68
determine the degree of portability of streaming services relative to the existing options at
the time, the most popular of which were CDs and digital downloads.
The first legal on-demand streaming platform to enter the US market was
Rhapsody, introduced by Listen.com in December of 2001. Rhapsody’s subscription-
based streaming platform offered consumers unlimited access to their catalog of songs,
which by 2002 consisted of over 175,000 titles, including popular tracks from every
major record label at the time, for $10 a month. However, users of the service could only
listen on a computer connected to the internet (Evangelista, 2002).
This restriction to online computer-based listening placed Rhapsody, as well as
other emergent on-demand streaming platforms such as Napster and Zune, at a
considerable disadvantage to CDs, the dominant playback format at the time. CDs were
compatible with home entertainment systems, portable players, car audio systems, and
computers whether they were online or offline (Coleman, 2003). As a result, on-demand
streaming was significantly inferior to CDs in the competitive dimension of portability.
Soon after the introduction of Rhapsody, Apple rolled out its digital download
store, iTunes, as well as the portable digital music player, the iPod (Apple, 2003). Digital
downloads, supported by low-cost digital music players from a variety of manufacturers,
offered greater portability than on-demand streaming or CDs. Digital downloads were
playable on portable digital players as well as computers, whether online or offline. In
addition, digital downloads were easily transferred from a computer to a portable device
and could be stored in multiple places at the same time (Coleman, 2003). On-demand
streaming, then, was also inferior to digital downloads in the competitive dimension of
portability.
69
Question 3: Did the mainstream music market initially reject on-demand music
streaming as a music playback format?
Initial response to on-demand music streaming was so modest, particularly
compared to the strong response to digital downloads, that Nielsen, the primary collector
and disseminator of market data for the entertainment industry, did not even start
reporting on-demand music streaming numbers until 2008, seven years after the launch of
Rhapsody. As shown in Table 4.2, in 2008, on-demand music streaming accounted for
less than .1% of total music consumption, generating only 352,755,259 streams or
235,170 stream equivalent albums (SEA) compared to an industry-wide 546,667,315
albums sold that year. Digital downloads for the same period, however, accounted for
31.5% of music consumption, generating 65,770,119 full digital albums and 106,939,845
track equivalent albums (TEA) (Nielsen, 2018).
Table 4.2
2008 Total Music Consumption by Format
2008 Raw Data 2008 Converted % of Total
Total Market (In Album Equivalents) 546,667,315 546,667,315
Physical Albums 373,722,181 373,722,181 68.36%
Digital Albums 65,770,119 65,770,119 12.03%
Digital Songs 1,069,398,449 106,939,845 19.56%
On-Demand Streams 352,755,259 235,170 0.04%
(Nielsen, 2018)
70
Figure 4.1 shows sales by format for the period 2008-2010. The regular spikes in
the chart represented seasonal increases in the month of December for each year.
Attempts to remove seasonal irregularities did not change the overall data, so it was
determined to leave all data in the chart.
There was a slight downward trend for the entire chart; however, the relative
position of the formats to one another remained unchanged for the period. The downward
trend was consistent with the overall decline of music consumption for the period
examined, as total consumption in 2008 was 546,667,315 album equivalents compared to
2010 consumption of 455,852,222 album equivalents, a decrease of 16.6% for the period.
Figure 4.1 2008-2010 Weekly US Consumption by Format (Nielsen, 2018).
A Pearson r correlation was run to determine if there was a significant
relationship between changes in on-demand streaming and the other formats including
CDs, download albums, and download songs between 2008 and 2010. Table 4.3 shows
71
the results of the analysis. There was no significant relationship between on-demand
streaming and any of the other formats.
Table 4.3
2008-2010 Correlation Analysis by Format
CDs DL Albums DL Songs Streams
CDs 1.00
DL Albums .28 1.00
DL Songs .32 .73 1.00
Streams -.17 .15 .08 1.00
N = 156
Question 4: Did on-demand music streaming find acceptance among the low-end
consumers of the existing market or create a new market due to its superior
performance in a secondary competitive dimension or through the introduction of a
unique business model?
While on-demand music streaming services struggled to gain customers, they did
survive in spite of the lack of portability. While the numbers were very small, the overall
concept of paying for access to thousands of songs, as opposed to buying only a few,
appealed to a niche group of consumers, referred to by Bott (2010) as, “one of the
world’s smallest cults” (Paragraph 1). To answer whether or not on-demand music
streaming was a disruptive innovation required the researcher to determine the identity of
these consumers and why they chose to use on-demand streaming.
72
As outlined in Chapter Three, addressing question four required a four-step
process. First, the researcher created a genre-level willingness to pay continuum in order
to identify low-end consumers. Second, the researcher identified genres with the highest
number of non-consumers of music by establishing a ratio of paid to non-paid activity.
The next step was to identify which genre, if any, adopted on-demand streaming at a
faster rate than the overall market and look for any connection to genres identified as
having a low willingness to pay or a high percentage of non-music consumers. Finally,
the researcher examined archival documents for evidence of why early adopters might
have chosen to use on-demand music streaming.
To establish a willingness to pay continuum required the researcher to examine
sales by format at the genre level with the assumption that consumers within a genre with
a low percentage of full album activity relative to the overall market had a lower
willingness to pay, full albums costing far more than individual songs. In order to
neutralize the potential skewing effect of large blockbuster releases, the researcher
averaged data across the years 2008-2010. Table 4.4 shows percentage of consumption
by format for the entire market and the nine genres included in the study.
Average consumption for the total market between 2008 and 2010 provided a
baseline willingness to pay of 76.96%. Genres with a lower willingness to pay than the
total market included Pop (51.53%), R&B/Hip-Hop (73.45%), and Dance/Electronic
(68.08%). These three genres represented the low-end of the market as defined by
Schmidt and Druehl (2008) in their work on disruptive innovation.
73
Table 4.4
2008-2010 Average Consumption by Format
Full Albums Digital Songs
Streaming
Total Market 76.96% 22.78% 0.26%
Pop 51.53% 47.90% 0.57%
R&B/Hip-Hop 73.45% 26.17% 0.38%
Rock 79.20% 20.53% 0.27%
Country 80.59% 19.22% 0.19%
Latin 88.77% 10.96% 0.27%
Christian/Gospel 86.10% 12.65% 0.11%
Jazz 88.92% 11.00% 0.09%
Dance/Electronic 68.08% 28.07% 0.31%
World Music 85.20% 14.65% 0.18%
(Nielsen, 2018)
Some disruptive innovations survive rejection by mainstream consumers by
creating a new market among people not currently participating in the existing market
(Christensen, 1997). In music, few people remain completely outside of the market;
however, many people opt not to pay for music, choosing instead to listen through free
options such as radio, programmed streaming, or illegal music download sites. For the
purpose of this study, these people represent non-consumers of music.
In addition to recording paid music consumption, Nielsen also tracks free activity,
including radio airplay and programmed streaming. Identifying genres with a low ratio of
paid activity to free activity is one way to isolate non-music consumers. Because radio
audience numbers (r) and programmed streaming numbers (ps) are several multiples
74
higher than consumption numbers (c), the formula c ÷ ((r ÷ 100) + (ps ÷ 100)) was used
to develop a paid-to-non-paid activity ratio. Table 4.5 shows the percentage of paid
activity relative to adjusted free activity resulting from the above formula.
Table 4.5
2008-2010 Paid-to-Non-Paid Ratio
Paid Consumption Airplay Audience Programmed Streams Ratio
Total Market 501,632,816 874,769,375,700 5,215,662,531 0.057
Pop 51,982,254 140,665,270,066 1,564,282,454 0.037
R&B/Hip-Hop 89,468,866 172,886,520,133 1,058,169,410 0.051
Rock 161,896,173 299,560,309,433 1,664,454,356 0.054
Country 52,690,740 113,678,967,266 597,768,313 0.046
Latin 19,212,333 60,601,299,566 135,199,298 0.032
Christian/Gospel 22,694,968 26,814,388,466 140,790,698 0.084
Jazz 11,183,407 8,690,875,933 180,792,281 0.126
Dance/Electronic 10,035,897 8,223,785,466 132,913,755 0.120
World Music 3,862,957 481,647,833 50,461,845 0.726
(Nielsen, 2018)
The baseline paid-to-non-paid ratio as determined by the total market was .057.
Genres with a lower paid-to-non-paid ratio included Latin, Pop, Country, R&B/Hip-Hop,
and Rock. These genres had more non-paid activity relative to the amount of paid activity
than the total market and, therefore, a higher number of non-consumers of music.
Table 4.6 shows the percentage of total consumption through on-demand music
streaming by genre. The benchmark for on-demand streaming as a percentage of total
75
consumption as determined by the average for the total market between 2008 and 2010
was .26%. Genres with a higher percentage of total consumption from streaming were
early adopters of on-demand music streaming and included Pop, R&B/Hip-Hop, and
Dance/Electronic.
Table 4.6
2008-2010 Ave % from Streaming
Streaming
Total Market 0.26%
Pop 0.57%
R&B/Hip-Hop 0.38%
Rock 0.27%
Country 0.19%
Latin 0.27%
Christian/Gospel 0.11%
Jazz 0.09%
Dance/Electronic 0.31%
World Music 0.18%
(Nielsen, 2018)
In comparing early adopters of on-demand music streaming with those genres
identified with the low-end of the market, there is perfect alignment with all three early
adopter genres. However, in comparing early adopters with the genres shown as having a
higher number of non-consumers, the connection is not as clear. At face value, this would
76
seem to indicate that early adopters of on-demand music streaming largely came from the
low-end of the existing market.
In examining why early adopters chose subscription services like Rhapsody, the
most common factor to appear was the subscription-based business model (Bott, 2010).
Early subscribers of on-demand streaming services could access tens of thousands of
songs for one monthly price. While traditional download sites such as iTunes offered
those same songs and more, users of those sites could only listen to those specific titles
they purchased. Access to all proved more appealing than ownership of some for these
early adopters (Gopinath & Stanyek, 2014).
Another factor mentioned in the literature regarding the appeal of early on-
demand streaming services was new music discovery, a factor of the competitive
dimension of depth of offering (Cesareo & Pastore, 2014). Curated playlists, user-
generated playlists, song recommendation features, and genre-based browsing allowed
users to discover new artists and songs outside of the relatively limited mainstream media
spotlight. Also mentioned was the fact that on-demand streaming more closely imitated
the experience offered by illegal streaming sites, providing a legal way for users to
continue browsing and trying music in their accustomed fashion absent the guilt of using
illegal sites (Cesareo & Pastore, 2014).
In the competitive dimension of sound quality, the average on-demand music
stream from sites like Rhapsody varied from 96 to 192 kbps depending on quality of
internet speed, configuration of the computer, and overall traffic on the site (Bullen,
2017). By comparison, CDs have always had a bit rate of 1,411 kbps (Coleman, 2003),
and the average digital download from iTunes has had a bit rate of 256 kbps (Sony,
77
2018). Clearly, on-demand streaming did not exhibit superior performance in sound
quality.
In looking at price, the radically different approach between a purchase and a
subscription model made price a difficult comparison because the models represent
different ways of viewing music consumption. The average music consumer in the US
spent $40 a year on music purchases in 2002 (Michel, 2006), whereas a Rhapsody
monthly subscription cost $120 per year. However, those subscribers were paying for
access to all of the songs on the platform, not just the content of three CDs, which is what
$40 would have purchased in 2002. It was a different model that appealed to a consumer
more concerned with access than ownership (Bott, 2010).
In summary, the appeal of the subscription-based business model combined with
superior performance in the competitive dimension of depth of offering appeared to drive
much of the early adoption of on-demand music streaming. In this instance, depth of
offering referred not just to the amount of content available, but also the degree to which
listeners could access the content. However, while early adopters loved the idea of access
over ownership, their numbers were few, prompting technology writer Ed Bott (2010) to
quip, “These services are still the stuff that cults are made of” (p. 3).
Question 5: Did on-demand music streaming improve over time in the core
competitive dimension while maintaining its superiority in some secondary
competitive dimension or through its unique business model?
In the decade from 2001 until 2010, on-demand streaming languished on the
sidelines of recorded music with annual consumption never exceeding 1% of total music
consumption for the US market (Nielsen, 2018). Despite several services joining
78
Rhapsody in offering on-demand music streaming, most notably Napster (2005), Zune
(2008), and Rdio (2010), on-demand streaming remained relatively static, making only
incremental changes to the basic services first available in 2001 (Bott, 2010). Attempts to
make on-demand music streaming more portable faced major roadblocks, particularly
from Apple, whose portable digital music player, the iPod, had a dominant market share
among digital music players just short of 75% and a closed operating system, which
blocked the various streaming services (Delahunty, 2009).
The appearance of smartphones in the mid-2000s broke Apple’s chokehold on
music portability and allowed consumers more access to streaming websites through
mobile devices. Competition between the various on-demand streaming platforms,
though, kept any one platform from being able to offer access to all of the providers
(Bott, 2010). However, everything changed in 2011 with the introduction in the US of a
new on-demand streaming service, Spotify.
Spotify was a true second generation innovation entering the US market with a lot
of fanfare in July of 2011 with several distinct advantages over existing on-demand
music streaming platforms. Hailed by New York Times columnist Ben Sisario (2011) as,
“the world’s most celebrated new digital music service” (para. 3), Spotify entered the
market with significant financing and sophisticated partnership deals with all major
content owners (Catalano, 2018). In addition, Spotify was compatible with all of the
major smartphones, including the iPhone, and offered an impressive array of features that
drew the attention of mainstream consumers, such as increased portability, the ability to
incorporate downloaded files, offline caching of playlists, and, possibly the biggest
innovation, a free, ad-supported subscription option (Sutter, 2011).
79
Covering the Spotify launch for the New York Times, technology writer Sisario
(2011) wrote, “[Spotify’s] crucial selling point has been its free access, which the
company believes can lure in new users” (para. 11). Spotify’s belief was not just wishful
thinking. In Europe, between 2008 and 2011, Spotify had already signed up 10 million
users, 1.6 million of which were paying for a subscription (Sisario, 2011). The early free
version of Spotify had several restrictions, including limits on time spent listening, a cap
on the number of times a free user could listen to an individual song, and no mobile
access (Sisario, 2011). However, those restrictions were gradually removed so, that by
2014, even free users had unlimited access to Spotify’s entire catalog while on a
computer, and significant access through all mobile devices (Morris, 2014).
In a survey conducted in February 2018 by MusicWatch (2018), a company
dedicated to music industry market research and industry analysis, on-demand streaming
customers listed portability-related features as those most important to them as users.
Portability features mentioned by the majority of users included “easy to use in the car”
and “easy to use on my phone” (MusicWatch, 2018, p. 27). These responses seem to
indicate that improvements in on-demand streaming in the core competitive dimension of
portability reached the minimum threshold required by mainstream music consumers.
The free, ad-supported subscription model, combined with increased portability,
removed many of the barriers to on-demand streaming for mainstream users, leading to a
wave of experimentation. By August of 2011, only a month after their US launch, Spotify
reported 1.4 million US users with 175,000 of those using the paid version (Statista,
2012). Prior to Spotify’s launch, in January of 2011, services estimated the entire US
market for on-demand music streaming to be 1.5 million users (Bylin, 2011). This means
80
Spotify doubled the market for on-demand music streaming, accomplishing in one month
what it took multiple platforms combined almost a decade to achieve.
Spotify’s success was quickly followed by on-demand streaming platforms
introduced by Google Play Music (2011), Beats (2014), Tidal (2015), Apple Music
(2015), and Amazon Unlimited (2016). While each of these formats experimented with
free trial periods, only Spotify offered a permanently free subscription. However, all of
the services offered greater portability and mobility than the previous generation of on-
demand service providers.
Question 6: Did on-demand music streaming eventually encroach upon sales of
existing music playback formats resulting in a shift in the competitive landscape?
Figure 4.2 shows sales by format for the period of 2011-2017. As before, the
regular spikes of activity represented seasonal fluctuations around December. The
random spike in the middle of 2015 was due to a shift in reporting where Nielsen moved
the beginning and end reporting dates due to an industry-wide decision to release new
recordings on Fridays as opposed to Tuesdays, which resulted in one week (7/9/15) with
11 days of activity instead of 7. As noted earlier, attempts to remove seasonal fluctuations
did not materially affect the outcome of the study.
The data for 2011-2012 remained similar to that examined for 2008-2011 with
only a slight increase in streaming. However, in 2013, streaming activity took a
noticeable jump, and an exponential increase began in 2015 continuing through 2017. In
2011, on-demand streaming finished the year with 4,928,087 in stream equivalent albums
(SEA), which represented approximately 1% of total consumption for the year. By 2017,
on-demand streaming contributed 412,038,975 SEA, which accounted for 64.72% of total
81
consumption for the year. Over that same period, overall consumption of music grew
from 471,591,738 total consumption in 2011 to 636,633,549 in total consumption in
2017, a 35% increase during the period.
Figure 4.2 2011-2017 Weekly US Consumption by Format (Nielsen, 2018).
A Pearson r correlation analysis was run to determine if these changes in on-
demand streaming and corresponding changes to the other formats represented a
significant relationship. Table 4.7 shows the results of the analysis. Unlike the 2008-2010
analysis, this time there was a significant negative relationship between changes in on-
demand streaming and the other formats.
The strongest negative relationship was between on-demand streaming and
downloaded songs, r (362) = -.83, p < .001. This was consistent with the raw data, which
showed digital songs, or track equivalent albums (TEA) dropping from 127,111,330 TEA
in 2011 to 55,481,541 TEA in 2017, a 56.4% decrease over the period. This was also
82
consistent with the findings in question four, which presented evidence that the fastest
adopters of on-demand music streaming were those with a lower willingness to pay.
Table 4.7
2011-2017 Correlation Analysis by Format
CDs DL Albums DL Songs Streams
CDs 1.00
DL Albums .58 1.00
DL Songs .50 .84 1.00
Streams -.49* -.67* -.83* 1.00
*p < .001
N = 364
Source: (Nielsen, 2019).
The next strongest negative relationship was between on-demand streaming and
downloaded albums, r (362) = -.67, p < .001. This was also consistent with the raw data,
which showed digital albums dropping from 103,091,988 albums in 2011 to 66,212,5
91
albums in 2017, a 35.7% decrease over the period. Overall, digital downloads in the form
of songs and albums dropped from a combined share of 48.8% of total consumption of
2011 to a combined share of only 19.11% in 2017.
While the negative relationship between on-demand streaming and CDs was
moderate when compared to downloaded songs and albums, it was still a significant
relationship, r (362) = -.49, p < .001. CDs dropped from 236,460,333 in 2011 to
102,900,442 in 2017, a 56.4% drop over the period. Again, this is consistent with the
83
findings of question four where consumers with the highest willingness to pay were
slowest to adopt on-demand music streaming.
Evidence is very strong that on-demand streaming encroached heavily upon
existing formats, going from 1.04% of total consumption in 2011 to 64.72% of total
consumption in 2017. However, the growth in streaming also resulted in a 35% increase
in overall consumption of music during the period, offsetting a trend of decline that dated
back to the mid-2000s. This would seem to indicate that streaming did not just shift
consumers from one format to another, but actually brought in a significant number of
non-consumers from non-paid platforms.
Summary of Findings
The quantitative and qualitative data gathered from archival sales records and
historical documents related to the introduction and rise of on-demand streaming appear
to confirm that on-demand music streaming was in fact a disruptive innovation. A careful
comparison of Christensen’s (1997) theory side-by-side with a summary of the facts
illustrates that the theory correctly predicted the path on-demand music streams would
take. A detailed comparison follows.
According to Christensen’s (1997) theory, the first characteristic of a disruptive
innovation is that it is initially inferior to existing options in the core competitive
dimension preferred by the market. A survey of the history of music playback formats
shows that consumers have consistently preferred portability over any other competitive
dimension in music (Gopinath & Stanyek, 2014). Likewise, documents related to the
launch of on-demand music streaming, most notably the launch of Rhapsody in 2001,
demonstrate that early versions of the format only allowed consumers to stream music
84
from a computer connected to the internet (Sisario, 2011). This restriction to online
computers meant early services provided far less portability than either CDs or digital
downloads, both of which played on mobile devices, in cars, and on offline computers.
According to the theory, because the innovation is inferior in the core competitive
dimension, the mainstream consumers of the market reject it (Christensen, 1997). An
examination of archived sales records related to on-demand music streaming make it
clear the mainstream market soundly rejected on-demand streaming in its initial form.
Even though on-demand streaming was available as early as 2001, consumer activity was
so small that Nielsen, the primary collector and distributor of entertainment industry
activity, did not even start tracking on-demand streaming numbers until 2008. At that
time, almost a decade after the introduction of on-demand streaming, consumer streaming
accounted for only .1% of music activity in the United States.
The third key characteristic of a disruptive innovation is that, in spite of its
rejection by the mainstream market due to its inferiority in the core competitive
dimension, the innovation survives because it appeals to the low-end of the existing
market or a new market through superiority in a secondary competitive dimension or
through a unique business model (Christensen, 1997). On-demand music streaming did
survive, largely due to its unique subscription business model, which allowed access to
all songs as opposed to ownership of some (Bott, 2010). Quantitative analysis of early
on-demand streaming numbers from 2008-2010, sorted by genre and organized into a
willingness-to-pay continuum showed early adopters to have come from the low-end of
the existing market, namely the Pop, R&B/Hip-Hop, and Dance/Electronic genres
(Nielsen, 2018, Schmidt & Druehl, 2008). Qualitative analysis of historical documents
85
from the period showed early adopters preferred the depth of offering and unique
subscription-based business model offered by on-demand streaming sites (Bott, 2010).
The theory of disruptive innovations states that, over time, the innovation, which
has managed to find a small following, begins to improve in the core competitive
dimension while also maintaining its advantages in other dimensions or through its
unique business model (Christensen, 1997). Starting in 2011, with the launch of Spotify
in the US, on-demand music streaming rapidly improved in the core competitive
dimension of portability. Through unique licensing arrangements with content owners,
stronger broadband internet support, and deals with all of the major mobile carriers
including Apple, Spotify was a true second-wave innovation (Sisario, 2011). In addition
to improved portability, Spotify not only continued the innovative subscription-based
business model introduced by Rhapsody a decade earlier, but improved upon it by adding
a free version (Sisario, 2011). Free access to millions of songs proved tempting enough to
entice trial from mainstream music consumers with 1.4 million users signing up within
the first month of Spotify’s launch (Statista, 2012).
The final characteristic of a disruptive innovation is that it eventually encroaches
upon existing options (Christensen, 1997; Schmidt & Druehl, 2008). In the first decade of
on-demand music streaming, from 2001-2010, the innovative format, championed by
Rhapsody, had no effect on the existing formats of CDs or digital downloads, never
contributing even 1% of total music consumption at any point in the period (Nielsen,
2018). However, with the launch of Spotify in 2011, the on-demand streaming format
began to grow rapidly so that by the end of 2017, on-demand streaming accounted for
65% of total music consumption (Nielsen, 2018). During that same period of 2011-2017,
86
CDs fell from 50% of total consumption to 16% of consumption and digital downloads
dropped from 49% of total consumption to 19% of consumption; a fall from which
neither is likely to recover.
An old idiom states, “A picture paints a thousand words,” which bears true in this
instance. Figure 4.3 demonstrates a classic disruptive innovation pattern with on-demand
streaming going from a flat-line trend from 2008 through 2012, to gradual upward
movement starting in 2013, followed by a steeply rising curve in 2015, accompanied by
dramatic declines in all other formats. On-demand music streaming moved from 1.04%
of total consumption to 65% of total consumption in the US in the seven-year period of
2011-2017 after practically no growth in the prior decade. It appears to have been a
classic example of Christensen’s (1997) theory of disruptive innovation.
Figure 4.3 2008-2017 Weekly US Consumption by Format (Nielsen, 2018).
87
Limitations
As with any research project, the present study had several limitations, some of
which may have had an impact on the degree to which the data represented in the study
reflected the actions of the market. First, two variables had the potential to produce a
confounding effect on the results of the study. However, the actual effect of these
variables, in reference to the actual question as to whether on-demand music streaming
was a disruptive innovation, was negligible as is explained below. In addition, the lack of
prior research on the adoption of music formats in the US presented limitations as well.
Finally, the lack of data related to individual, consumer-level behavior led to the less
reliable use of genre-level data. A brief discussion of these limitations follows.
One potential confounding variable in this study was access to the internet speeds
required for on-demand music streaming. On-demand music streaming sites
recommended internet speeds of 256 kilobytes per second (kps) (Dilley, 2017). In the
early 2000s, when Rhapsody first introduced on-demand music streaming, roughly 92%
of internet-using households were still accessing the internet through a dial-up modem
with average speeds of 56 kbs (Kleinbard, 2000). Lack of access to the necessary internet
speeds for on-demand streaming could have been a variable in the early rejection of the
format. This limitation was largely offset, however, by the fact that in the decade between
2001 and 2010, access to broadband, high-speed internet as well as mobile web access,
referred to as 3G, exponentially increased (Statista, 2019), and yet adoption of on-
demand music streaming remained negligible (Nielsen, 2018). For this reason, the
researcher opted not to include an in-depth examination of internet usage as part of this
study. Certainly, lack of access to adequate internet speeds could have been a factor in
88
the initial rejection of on-demand music streaming, however, if it had been a primary
factor, adoption would have increased along with increasing internet speeds, and that did
not occur.
Another potential confounding variable was the presence of illegal music
streaming websites as a substitute for legal music playback formats. A major part of this
study involved the interplay between the various music playback formats, most notably
CDs, digital downloads, and on-demand streaming. The presence of illegal substitutes for
these formats influenced consumer behavior, but in a manner hard to measure, as such
activity lay outside of the visible market. While several organizations, most notably the
Recording Industry Association of America (RIAA), have attempted to quantify the use
of illegal online music options, in the opinion of the researcher, they failed to isolate the
impact of illegal activity to the degree that the data could have been used in this study
(Siwek, 2007). As a result, the researcher limited this study to legal formats only.
The absence of illegal activity in the numbers related to the adoption of on-
demand music streaming potentially skewed the overall picture related to the rate at
which consumers adopted on-demand streaming. Research offered by Spotify (Spotify,
2013) indicated a decline in the use of illegal websites because of consumers choosing to
use Spotify instead. For the most part, though, such considerations fell outside the scope
of the question as to whether on-demand streaming constituted a disruptive innovation.
Examination of the activity on paid platforms more than adequately addressed the overall
question of the study and the inclusion of data related to illegal activity or programmed
streaming would not have changed the outcome, only the degree of interplay among CDs,
digital downloads, and streaming.
89
Another limitation of the present study was the lack of prior research related to
the adoption of on-demand streaming. As a relatively recent phenomenon, not enough
time has passed for studies on on-demand streaming to make their way through academic
channels. This required the researcher to rely on non-academic sources such as historical
press reports, news articles, corporate press releases and other primary sources. I hope
that this study has been able to introduce data related to on-demand music streaming in
such a manner as to stimulate academic conversation around the issue and aid future
researchers.
One last limitation was the fact that consumer behavior in relation to music was
non-exclusive and, as a result, largely hidden. The same consumer could purchase a CD,
download a song, and listen to a stream all at once, not to mention listen to the radio,
stream from a programmed site or even use an illegal website. A consumer choosing one
format did not necessarily do so at the expense of another format. This lack of exclusivity
made it difficult to isolate consumer behavior and preference. As has been explained,
because of this lack of transparency of individual consumer behavior, the common
practice in the music industry has been to use genre as a means of examining consumer
behavior (Lena & Peterson, 2008). However, genre preference itself was not mutually
exclusive; a consumer could easily be a fan of Pop and Country music at the same time.
Using genre as a means of isolating consumer behavior was the best option available, but
less than ideal. Future researchers may consider taking the time to develop longitudinal
studies of consumer preference and behavior at the individual consumer level.
90
Implications and Recommendations
Economic theorist, Joseph Schumpeter, referred to innovation as a process of
creative destruction, a necessary displacement of the old in order to bring about growth
and health within a market (Mee, 2009). Prior to the rise of on-demand music streaming
as the leading music playback format, the US music industry had experienced a fourteen-
year decline in revenue going from a peak of $14.38 billion in 2000 to a low of $6.6
95
billion in 2014 (RIAA, 2017). While the move from a purchase model to a subscription
model brought about by on-demand music streaming was quite disruptive for music
industry stakeholders, the result has been a major market turnaround resulting in $8.723
billion in revenues in 2017, a 30% increase over the low point of 2014.
The music industry took a risk by negotiating licenses with Spotify that allowed
them to offer a free, ad-supported subscription. The data examined in this study would
seem to indicate, though, that the model of using free subscriptions to move people into
paid subscriptions not only paid off, but also led to a dramatic recovery for recorded
music. Disruptive innovations can be painful, but they are often necessary for an industry
to survive.
Not only have recorded music revenues grown, several of the champions of on-
demand music streaming have claimed that this growth has been at the expense of illegal
online music activity. According to Spotify’s website, there has been a dramatic decrease
in illegal online activity since Spotify began operations in the US (Spotify, 2013). Their
research shows that young people aged 18-29 are 55% less likely to use illegal file
sharing when offered a free legal alternative (Spotify, 2013). Future research should
concentrate on validating these claims.
91
One interesting pattern that emerged through the research was the fact that not all
genres adopted on-demand streaming at the same rate. Table 4.8 shows on-demand
streaming as a percent of total consumption for the years 2011 and 2017 by genre. The
figure also shows the market share for each genre for the two periods.
Table 4.8
Growth in Streaming and Market Share by Genre: 2011 and 2017
2011 Stream % 2011 Market Share 2017 Stream % 2017 Market Share
Total Market 1.04% 100.00% 64.72% 100.00%
Pop 1.86% 13.60% 65.86% 12.69%
R&B/Hip-Hop 1.76% 17.20% 76.73% 24.56%
Rock 0.63% 29.00% 47.05% 20.77%
Country 0.74% 11.70% 46.40% 7.70%
Latin 1.35% 2.95% 88.41% 5.87%
Christian/Gospel 0.17% 4.49% 43.99% 2.65%
Jazz 0.23% 1.93% 38.83% 1.03%
Dance/Electronic 1.17% 2.70% 76.92% 3.67%
World Music 0.32% 0.65% 72.63% 1.00%
(Nielsen, 2018)
Genres who adopted streaming at a rate significantly faster than the total market,
indicated in italics in Figure 4.6, showed market share growth between 2011 and 2017.
Of particular note was the growth of R&B/Hip-Hop from 17.2% market share to 24.56%
market share during the period examined. Not only did R&B/Hip-Hop grow in market
92
share, it grew in raw numbers going from 81,115,806 in total consumption in 2011 to
156,386,485 total consumption in 2017, a 93% increase over the period.
Similarly, those genres that adopted at a rate significantly slower than the overall
market experienced a decline in market share. Most notable among this group was Rock,
which went from 29% of the total market to 20.77% of the total market. Country music
also experienced a significant decline in market share during the period dropping from
11.7% to 7.7% during the period.
Future research into on-demand streaming should examine the connection
between early adoption of on-demand streaming and growth in market share. The
substantial growth in total volume among early adopting genres would indicate the
change in market share was due to more than a shift in consumer taste. The data would
seem to indicate that those genres were able to grow by moving non-paying consumers
into payed services. This would be worth future study and particularly useful for genres
that have experienced declines during the period.
Another interesting pattern that emerged was the lack of seasonality in the
streaming data. As noted in the various sales charts, particularly Figure 4.3, there was a
spike in activity each year in the month of December. This spike in activity was most
noticeable in the line indicating physical album purchases.
In examining the other formats, the seasonal trend was less remarkable among
digital downloads and entirely absent in the on-demand streams. This absence of
seasonality was an important element in understanding the shift from purchase to
subscription behavior. The lines indicating purchase behavior, including physical albums,
digital albums, and digital songs, are measuring a single transaction where a consumer
93
purchases a piece of product. The seasonality is a result of retail patterns because the
chart is measuring purchase activity.
In on-demand streaming, the transaction represented by the chart was not a one-
time purchase, but an individual stream of a song; it reflects listening behavior, not
purchase behavior. The lack of seasonality, then, was due to the measurement of listening
activity, which does not fluctuate greatly based on seasons. The lack of seasonality in on-
demand music streaming activity has interesting implications and is worth future study.
One last observation from the data was that digital downloads, once expected to
become the dominant music playback format (Coleman, 2003), never actually displaced
the CD. According to the Nielsen (2018) data, digital downloads reached their peak in
2012, almost a decade after the launch of the iTunes store, with full album downloads
generating 117,582,197 in consumption, and single song downloads generating
133,588,041 in TEA for a combined 53.4% of total consumption. That same year, CDs
alone sold 206,752,704 copies, representing 43.97% of total consumption. Digital
downloads encroached upon, but never replaced CDs. On-demand streaming, though, has
displaced CDs to the point that, by 2017, CDs only made up 16% of total consumption.
Digital downloads appear to have been a bridge technology leading from the
physical CD format to the on-demand music stream. Digital downloads represented a
comfortable transition for a music industry and consumer base used to the concept of
owning music and a business model built on purchases. In addition, 2001 copyright law,
internet speeds, and mobile hardware were simply not ready to provide the on-demand
streaming experience most mainstream consumers expected. It would take a decade for
the stage to be set for on-demand streaming to thrive.
94
The idea of technological innovations that are only partial moves into the digital
space is not restricted to music. Netflix started with distribution of DVDs and moved to
on-demand streams. Amazon started out as an online book distributor and became a
distributor of online books, and then a distributor of everything. In future studies on
disruptive innovations, it would be useful to look at this phenomenon of second wave
innovations that take advantage of fully developed technology not available at the time of
an initial innovation. There could be industries that believe they have fully transitioned
into the online space that are actually in a bridge period with an even more disruptive
innovation waiting in the wings.
Twenty years after first proposing the model that would form the basis for
disruptive innovation theory, Clayton Christensen, along with Michael Raynor and Rory
McDonald (2015), warned in a Harvard Business Review article that incumbent firms
must, “Disrupt or be disrupted” (p. 8). For industry leaders to continue to lead, they must
become better at innovation. However, as Schmidt and Druehl (2008) point out,
“sustaining innovations have more often been associated with incumbents and disruptive
innovations with entrants” (p. 349).
After a decade and a half of decline, the music industry experienced a turnaround.
A major part of that shift in momentum was due to the industry finally recognizing the
need to embrace new music formats and the subscription business model. This required
rethinking music licenses, copyright payments, and a shift from ownership to access in
regards to music.
While innovation around on-demand streaming involved several entrants, most
notably, Rhapsody, Spotify, and Amazon, many incumbent firms were able to make the
95
transition into on-demand music streaming, including major record labels, publishers,
artists, songwriters, and other content owners. They did this by embracing innovation,
creating partnerships that took advantage of new revenue streams, and imagining new
ways of doing business. Disruption does not always have to lead to destruction.
Christensen, Raynor, and McDonald (2015) state at the end of their Harvard
Business Review article, “As an ever-growing community of researchers and practitioners
continues to build on disruption theory and integrate it with other perspectives, we will
come to an even better understanding of what helps firms innovate successfully” (p. 11).
This study of on-demand music streaming offered yet another look at a successful
disruption, one in which incumbent firms survived for the most part. Perhaps this is
evidence that Christensen’s work has made a difference in helping businesses cope with
disruption and harness its creative force for growth.
96
REFERENCES
Abkowitz, A. (2009, January 28). How Netflix got started. Retrieved from Fortune:
http://archive.fortune.com/2009/01/27/news/newsmakers/hastings_netflix.fortune/
index.htm
Adner, R. (2002). When are technologies disruptive? A demand-based view of the
emergence of competition. Strategic Management Journal, 23(8), 667-688.
Albright, D. (2015, April 30). The evolution of music consumption: How we got here.
Retrieved from MakeUseOf: https://www.makeuseof.com/tag/the-evolution-of-
music-consumption-how-we-got-here/
Alkamade, F., & Castaldi, C. (2005). Strategies for the diffusion of innovations on social
networks. Computational Economics, 25, 3-23
Anthony, S. D. (2005, May-June). Do you really know what you are talking about?
Strategy & Innovation, 3 (3), pp. 3-6.
Apple (2003, April 28). Apple launches the iTunes music store. Retrieved from
Apple.com: https://www.apple.com/newsroom/2003/04/28Apple-Launches-the-
iTunes-Music-Store/
Arango, T. (2010, December 12). Time Warner views Netflix as a fading star. Retrieved
from The New York Times: https:nytimes.com/2010/12/13/business/media/
13bewkes.html
97
Auiar, L., & Martens, B. (2013). Digital music consumption on the Internet: Evidence
from Clickstream data. Seville: Institute for Prospective Technological Studies.
Auletta, K. (2014, February 3). Outside the box: Netflix and the future of television. The
New Yorker, pp. 1-19. Retrieved from https://www.newyorker.com/magazine/
2014/02/03outside-the-box-2
Bellis, M. (2017, April 19). The history of MP3. Retrieved from ThoughtCo:
https://www.thoughtco.com/history-of-mp4-1992132
Billboard (1966, July 16). Mercury issues 49 ‘cassettes’. Billboard Magazine, p. 69.
Bostic, K. (2013, April 16). Apple’s iTunes rules digital music market with 63% share.
Retrieved from AppleInsider.com: http://appleinsider.com/articles/13/04/16/
apples-itunes-rules-digital-music-market-with-63-share
Bott, E. (2010, December 10). More iTunes alternatives: Can a subscription music
service ever succeed? Retrieved from ZD Net: https://www.zdnet.com/article/
more-itunes -alternatives-can-a-subscription-music-service-ever-succeed/
Bowen, G. A. (2009) Document analysis as qualitative research method. Qualitative
Research Journal, 9 (2), 27-40.
Bullen, J. (2017, November 20). Vinyl, Streaming, and CDs: Maybe it’s not just about the
audio quality. Retrieved from ABC News: https://www.abc.net.au/news/
science/2017-11-21/vinyl-streaming-cd-what-is-best-audio-quality/86934
98
Bylin, K. (2011, January 21). Rhapsody adds over 100,000 subscribers. Retrieved from
Hypebot: https:www.hypebot.com/hypbot/2011/01/rhapsody-adds-over-100000-
new-subscribers.html
98
Caron, S. (2004). Transitioning to DVD. Library Journal, 129 (5), 4-5.
Catalano, F. (2018, April 14). From Rhapsody to Napster: How this pioneering music
service coulda been Spotify—and why it isn’t. Retrieved from Geek Wire:
https://www.geekwire.com/2018/rhapsody-napster-pioneering-music-service-
coulda-spotify-isnt/
Cesareo, L., & Pastore, A. (2014) Consumers’ attitude and behavior towards online
music piracy and subscription-based services. Journal of Consumer Marketing, 31
(6), 515-525.
Christensen, C. M. (1997). The Innovator’s Dilemma. Boston: Harvard Business Review
Press.
Christensen, C. M. (2006). The ongoing process of building a theory of disruption. The
Journal of Product Innovation Management, 23, 39-55.
Christensen, C. M., Raynor, M., & McDonald, R. (2015, December). What is disruptive
innovation? Harvard Business Review, 93 (6), 1-11.
Coleman, M. (2003). Playback: From the Victrola to MP3: 100 Years of Music,
Machines, and Money. Cambridge: DaCapo Press.
Daneels, E. (2004). Disruptive technology reconsidered: A critique and research agenda.
The Journal of Product Innovation Management, 21, 246-258.
Dearing, J. W., & Singhal, A. (2006). Communication of innovations: A journey with Ev
Rogers. In J. W. Dearing, Communication of Innovations (pp. 15-28). Thousand
Oaks: Sage.
99
Delahunty, J. (2009, September 9). iPod market share at 73.8 percent. Retrieved from
AfterDawn: https://www.afterdawn.come/news/article/cfm/2009/09/09/
ipod_market_share_at_73_8_percent_225_million_ipods_sold_more_games_for_
touch_than_psp_nds_apple
Dilley, J. (2017, November 20). How much speed to I need to stream music? Retrieved
from High Speed Internet: https://www.highspeedinternet.com/resources/how-
much-speed-do-I-need-for-pandora-and-spotify/
Downes, L., & Nunes, P. F. (2013, March). Big-Band Disruption. Harvard Business
Review, 91 (2), 1-12.
Drucker, P. F. (1985). Innovation and Entrepreneurship. New York: Harper Business.
Elberse, A. (2010). Bye-Bye bundles: The unbundling of music in digital channels.
Journal of Marketing, 74 (3), 107-123.
Evangelista, B. (2002, July 1). Industry starting to endorse Net music/Listen.com to offer
songs from all five major labels. Retrieved from San Francisco Chronicle:
https://www.sfgate.com/business/article/industry-starting-to-endorse-Net-music-
801248.php
Fielding, N. (2004). Getting the most from archived qualitative data: Epistemological,
practical, and professional obstacles. International Journal of Social Research
Methodology, 7 (1), 97-104.
Fischer, E., & Parmentier, M. A. (2010). Doing qualitative research with archival data:
Making secondary data a primary source. Advances in Consumer Research, 37,
198-199.
100
Flanagan, A. (2013, August 16) When the world first met the LP, Cassettes, CD and other
physical formats. Retrieved from BillboardBiz: https://www.billboard.com/
biz/articles/news/digital-and-mobile/5657586/when-the-world-first-met-the-lp-
cassttes-cd-and-other
Gopinath, S., & Stanyek, J. (2014). Anytime, anywhere? An introduction to the devices,
markets, and theories of mobile music. In S. Gopinath, & J. Stanyek, The Oxford
Handbook of Mobile Music Studies, Vol. 1. Oxford: Oxford University Press.
Green, E. (2013, June 30). Innovation: The history of a buzzword. The Atlantic, pp. 1-5.
Retrieved from https://www.theatlantic.com/business/archive/2013/06/innovation-
the-history-of-a-buzzword/277067/
Griggs, B., & Leopold, T. (2013, April 26). How iTunes changed music, and the world.
Retrieved from CNN: https://www.cnn.com/2013/04/26/tech/web/itunes-10th-
anniversary/index.html
Guttentag, D. (2015). Airbnb: Disruptive innovation and the rise of an informal tourism
accommodation sector. Current Issues in Tourism, 18, 1192-1217.
Hand, D. J. (2018). Statistical challenges of administrative and transaction data. The
Royal Statistical Society, 181, 555-605.
Hochman, S. (1990, June 23). Cassette prices rise, close the gap on Compact Discs.
Retrieved from Los Angeles Times: http://articles.latimes.com/1990-06-
23/entertainment/ca-284_1_cassette-prices
Hunt, R. A. (2013). Entrepreneurial tweaking: An empirical study of technology
diffusion through secondary inventions and design modifications by start-ups.
European Journal of Innovation Management, 16 (2), 148-170.
101
Ingham, T. (2016). Spotify is converting more people into paying subscribers than ever
before. Retrieved from Music Business Worldwide: https://www.musicbusiness
worldwide.com/spotify-is-converting-more-people-into-paying-customers-than-
ever-before/
Keene, P. (2016, November 29). Spotify v. Pandora: Streaming your favorite songs on-
demand. Retrieved from Klemchuck: http://klemchuk.com/spotify-pandora-
streaming-on-demand/
Kleinbard, D. (2000, November 1). Broadband access surges. Retrieved from Money:
https://money.cnn.com/2000/11/01/technology/fcc_dsl/
Lazrus, D. (2016, May 13). When you buy digital content on Amazon or iTunes, you don’t
exactly own it. Retrieved from LA Times: http://www.latimes.com
/business/lazarus/la-fi-lazarus-digital-content-20160513-snap-story.html
LEM. (2014, May 1). History of the compact disc. Retrieved from Low-end Mac:
Lena, J. C., & Peterson, R. A. (2008). Classification as culture: Types and trajectories of
music genres. American Sociological Review, 73, 697-718.
Leonard, D. A. (2006). Innovation as a knowledge generation and transfer process. In A.
Singhal, & J. W. Dearing, Communication of Innovations (pp. 83-110). Thousand
Oaks: Sage.
Lucko, G., & Mitchell, Z. W. (2010, January). Quantitative research: Preparation of
incongruous economic data sets for archival data analysis. Journal of
Construction Engineering and Management, 136 (1), 49-57.
102
Machay, J. (2018). What are the dangers of torrents? Retrieved from Chron.com:
http://www.smallbusiness.chron.com/dangers-torrents-70661.html
McElhearn, K. (2016, January 9). 15 years of iTunes: A look at Apple’s media app and its
influence on an industry. Retrieved from Macworld: https://www.macworld.
com/article/3019878/software/15-years-of-itunes-a-look-at-apples-media-app-
and-its-influence-on-an-industry.html
Mee, S. (2009). Joseph Schumpeter and the business cycle: An historical synthesis.
Student Economic Review, 23, 1-6.
Michel, N. J. (2006). The impact of digital file sharing on the music industry: An
empirical analysis. Topics in Economic Analysis and Policy, 6 (1), 1-22.
Montoro-Pons, J. D., & Cuadrado-Garcia, M. (2016). Genre-breadth, variety-seeking
behavior and preferences in recorded popular music. Valencia: University of
Valencia.
Morris, S. M. (2014, January 15). No more limits: Spotify removes time restrictions on
free desktop player. Retrieved from Independent: https://www.independent.co.uk
/life-style/gadgets-and-tech/spotify-removes-time-restrictions-for-desktop-
listening-9062361.html.
Munarriz, R. (2008, December 10). Blockbuster CEO has answers. Retrieved from The
Motley Fool: https://www.fool.com/investing/general/2008/12/10/blockbuster-
ceo-has-answers.aspx
MusicWatch. (2018). Music Monitor: February 2018. New York: MusicWatch.
Nicholas, T. (2003). Why Schumpeter was right: Innovation, market power, and creative
destruction in 1920s America. The Journal of Economic History, 63, 1,023-1,058.
103
Nielsen. (2018). US Music Consumption by Format: 2008-2017 (Unpublished Raw
Data). New York: Nielsen Entertainment.
Norton, J. A., & Bass, F. (1987). A diffusion theory model of adoption and substitution
for successive generations of high-technology products. Management Science, 33
(9), 1069-1086.
Passman, D. S. (2015). All You Need to Know About the Music Business. New York:
Simon & Schuster.
Plambeck, J. (2010, May 9). In mobile age, sound quality steps back. Retrieved from The
New York Times: https://archive.nytimes.com/www.nytimes.com/2010/05/10/
business/media/10audio.html
Reuters. (20018, February 27). Redbox expands kiosk presence at Walgreens. Retrieved
from Reuters: https://www.reuters.com/article/us-redbox-walgreens/redbox-
expands-kiosk-presence-at-walgreens-idUSN2745885020080227
RIAA. (2017). US Recorded Music Revenues by Format. Washington: Recording
Industry Association of America.
Ricker, T. (2015, June 8). First Click: Remember when Steve Jobs said even Jesus
couldn’t sell music subscriptions? Retrieved from The Verge:
https://www.theverge .com /2015/6/8/8744963/steve-jobs-jesus-people-dont-
want-music-subscriptions
Rogers, E. M. (2003). Diffusion of Innovations (Fifth ed.). New York: Free Press
104
Satell, G. (2014, September 5). A look back at why Blockbuster really failed and why it
didn’t have to. Retrieved from Forbes: https://www.forbes.com/sites/
gregsatell/2014/09/05/a-look-back-at-why-blockbuster-really-failed-and-why-it-
didnt-have-to/#38ffaaf61d64
Schmidt, G. M., & Druehl, C. T. (2008). When is a disruptive innovation disruptive?
Journal of Product Innovation Management, 25, 347-369.
Schumpeter, J. (1928). The instability of capitalism. The Economic Journal, 38, 361-386.
Schumpeter, J. (1942). Capitalism, Socialism and Democracy. London: Routledge.
Shapiro, M. (2014, August 8). The history of camcorders. Retrieved from ECN:
https://www.ecnmag.com/blog/2014/08/history-camcorders
Shayo, C., & Guthrie, r. (2005). From Edison to MP3: A struggle for the future of the
music recording industry. Hershey: Idea Group Publishing.
Sisario, B. (2011, July 13). New service offers music in quantity, not by song. Retrieved
from The New York Times: https://www.nytimes.com/2011/07/14/technology
/spotify-music-streaming-service-comes-to-us.html
Siwek, S. E. (2007). The true cost of sound recording piracy to the U. S. economy.
Washington: RIAA.
Sony. (2018). Battle of the sound. Retrieved from Sony: https://www.sony.com
/electronics /hi-res-audio-mp3-cd-sound-quality-comparison
Spotify. (2013). Spotify’s progress so far. Retrieved from Spotify: http://web.archive.org
/web/20131205225811/http://www.spotifyartists.com/spotify-explained/
105
Statista. (2012, April). Spotify’s paid and overall subscriber numbers in the US from
August 2011 to April 2012. Retrieved from Statista: https://www.statista.com/
statistics/222099/spotifys-paid-and-overall-subscriber-numbers-in-the-us/
Statista. (2019, January). Average internet connection speed in the United States from
2007 to 2017. Retrieved from Statista: https://www.statista.com/statistics
/616210/average-internet-connection-speed-in-the-US/
Stevans, L. K., & Sessions, D. N. (2005). An empirical investigation into the effects of
music downloading on the consumer expenditure of recorded music. Journal of
Consumer Policy, 28, 311-324.
Sutter, J. D. (2011, July 15). What’s this Spotify thing all about? Retrieved from CNN:
http://www.cnn.com/2011/TECH/innovation/07/14/spotify.us.why/index.html
Tryon, C. (2011). Redbox vs Red Envelope, or what happens when the infinite aisle
swings through your grocery store? Canadian Journal of Film Studies, 20 (2), 38-
54.
Valente, T. W. (2006). Communication network analysis and the diffusion of
innovations. In A. Singhal, & J. Dearing, Communication of Innovations (pp. 61-
82). Thousand Oaks: Sage.
Wessel, M. (2016, January). Disruptive innovation: How big data is changing disruptive
innovation. Harvard Business Review, 94 (1), 1-5.
Zeidler, S. (2007, October 28). Walgreens sees movie-burning DVD kiosks at stores.
Retrieved from Reuters: https://www.reuters.com/article/us-walgreens/walgreen-
sees-movie-burning-dvd-kiosks-at-stores-idUSN2840912120071029
2
DR. SUE GREENER & DR. JOE MARTELLI
AN INTRODUCTION TO
BUSINESS RESEARCH
METHODS
3
An Introduction to Business Research Methods
3rd edition
© 2018 Dr. Sue Greener, Dr. Joe Martelli & bookboon.com
ISBN 978-87-403-2045-9
Peer review by Kiefer Lee, Principal Lecturer,
Sheffield Business School, Sheffield Hallam University
http://bookboon.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Contents
44
CONTENTS
Preface 8
1 Research problems and questions and how they relate to debates
in Research Methods 9
1.1 Chapter Overview 9
1.2 Introduction 9
1.3 The nature of business research 10
1.4 What kind of business problems might need a research study? 14
1.5 What are the key issues in research methods we need to understand? 16
1.6 Questions for self review 24
1.7 References 24
2 Putting the problem into context: identifying and critically
reviewing relevant literature 26
2.1 Chapter Overview 26
2.2 How does literature relate to research? 26
2.3 What kind of literature should we search for? 28
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
http://www.deloitte.ca/careers
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Contents
5
2.4 Effective literature searching 31
2.5 Critical analysis of literature 35
2.6 Using Harvard referencing style 41
2.7 Questions for self review 42
2.8 References 43
3 Choosing research approaches and strategies 44
3.1 Chapter overview 44
3.2 Different perspectives of knowledge and research which underpin
research design 44
3.3 Identify differing research paradigms for business 46
3.4 Key differences between qualitative and quantitative research
methods and how and why they may be mixed 47
3.5 Criteria of validity and reliability in the context of business research 49
3.6 Your choice of research strategy or design 51
3.7 Classification of research 52
3.8 The Business Research Process 54
3.9 The Academic business research process 55
3.10 Questions for self review 56
3.11 References 56
4 Ethics in business research 58
4.1 Chapter Overview 58
4.2 Understand how ethical issues arise in business research at every stage 58
4.3 Ethical criteria used in Higher Education business research studies 62
4.4 Strategies to ensure ethical issues in business research are addressed
appropriately 62
4.5 Plagiarism 66
4.6 Questions for self review 67
4.7 References 67
5 Choosing samples from populations 68
5.1 Chapter Overview 68
5.2 Understand how and why sampling relates to business research 68
5.3 Identify and use a range of probability and non-probability
sampling techniques 69
5.4 Selecting the size of your sample 72
5.5 Understand and assess representativeness of samples and
generalisability from samples 75
5.6 Sampling simulation exercise 77
5.7 Questions for self review 77
5.8 References 77
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Contents
6
6 Quantitative research methods: collecting and analysing data 78
6.1 Chapter Overview 78
6.2 Anticipating how the research design is affected by data collection
and analysis tools 79
6.3 Recognising different levels of data for analysis 80
6.4 Coding and entering data for computerized statistical analysis 82
6.5 Choosing appropriate ways to present data through charts, tables
and descriptive statistics 85
6.6 Selecting appropriate statistical tools for the research variables 88
6.7 Families of Statistics 89
6.8 Measures of Correlation – the correlation coefficient 91
6.9 Regression analysis 92
6.10 Statistical significance 93
6.11 Questions for self review 95
6.12 References 96
7 Questionnaire design and testing 97
7.1 Chapter overview 97
7.2 Appreciate and overcome the difficulties associated with
questionnaire design 97
7.3 Choosing from a range of question formats 99
7.4 How to design, pilot and administer questionnaires 101
7.5 Questions for self review 105
7.6 References 106
8 Using secondary data 107
8.1 Chapter Overview 107
8.2 The value of secondary data to business research 107
8.3 What to look for as secondary data and where to find it 110
8.4 The disadvantages of using secondary data in business research 112
8.5 Big Data 114
8.6 Questions for self review 115
8.7 References 115
9 Qualitative research methods: collecting and analysing
qualitative data 116
9.1 Chapter overview 116
9.2 Key issues in qualitative data analysis 116
9.3 The range of qualitative research methods applicable to research topics 118
9.4 How qualitative data can be prepared for analysis 123
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Contents
7
9.5 Computer based methods for qualitative data analysis 124
9.6 Questions for self review 125
9.7 References 125
10 Practical issues in conducting interviews, focus groups,
participant observation 127
10.1 Chapter overview 127
10.2 Practical considerations relating to participant observation 127
10.3 Practical issues relating to interviews 130
10.4 Practical issues relating to focus groups 134
10.5 Questions for self review 136
10.6 References 136
11 Forecasting trends 138
11.1 Chapter overview 138
11.2 Why forecasting is not widely covered in the business research
methods literature 138
11.3 Existing methodologies for forecasting 140
11.4 Basic forecasting tools 144
11.5 Regression and discriminant analysis 145
11.6 Measures commonly used to evaluate forecasts & predictions 148
11.7 Exploring the value of forecasting methods in business practice 150
11.8 Questions for self review 151
11.9 References 151
12 Reporting research results 152
12.1 Chapter overview 152
12.2 Your personal approach to writing a research report 153
12.3 The differences between writing a report for a business audience
and for academic purposes 155
12.4 Producing an oral presentation of key findings 161
12.5 Questions for self review 163
12.6 References 163
Comments from peer reviewer 164
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS PrefaCe
8
PREFACE
Welcome to this research methods guide which aims to introduce students to the main ideas
and issues to consider in conducting rigorous and effective business research. We offer many
links and references to standard works in the field. This book is not a substitute for those
standard works, but a starting point which should help you to understand the terminology
and find what you might need to know.
In this third edition, we have included some updating and improved explanations of families
of statistics, regression analysis, big data, mixed methods, social media and presentation.
We have also tried to improve sections by changing formats to make them easier to read.
If you are new to academic research, keep one thing in mind: there is no one right way
to research and no one right way to write a research methodology chapter. The point is
to look at the alternatives and build a rational case for the path that your research takes.
Sue Greener & Joe Martelli
2018
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
9
1 RESEARCH PROBLEMS AND
QUESTIONS AND HOW
THEY RELATE TO DEBATES
IN RESEARCH METHODS
1.1 CHAPTER OVERVIEW
1.1.1 LEARNING OUTCOMES
By the end of this chapter successful students will be able to:
1. Distinguish business and management research from other kinds of research
2. Understand the issues relating to identifying and reformulating problems for research
3. Identify the key debates in research methods
References, Links and Further Reading
Bryman and Bell (2015) or look for other web resources relating to “problematisation”,
business research and debates in research methods in social sciences.
1.2 INTRODUCTION
1.2.1 RESEARCH METHODS AS AN AREA TO STUDY
As a student of Business Research Methods, you will be wearing two hats. One hat or
role is that of a student who wishes to pass exams in this area, so you will need to learn
enough about research methods to write an assignment of appropriate standard and/or to
pass the examination. This is your academic role, and this means we must look at research
methods from an academic point of view. All academic work, as you already know, must
take account of published literature (textbooks, journal articles, professional articles, relevant
website information, company literature etc.). So we will be looking at research methods
literature, in order that you can use it to help you understand the chapters, and use the
literature in your assessment. You may continue your studies and do further academic work
at a higher level; again you will need to use research methods ideas and theories from the
literature directly in that study.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
10
But there is another hat, that of manager, research consultant or practitioner, for which this
short book aims to prepare you. Sometimes, your academic assignments may require you
to step into the role of consultant. So sometimes in this book, you will need to imagine
yourself in the role of manager or consultant, needing to answer questions in real-time,
carry out research to answer vital questions for the business you are in.
Most of you reading this book may not wind up as researchers in an organization or ever
have the title of “researcher”, but in fact, as a manager or a professional in an organization,
you will be expected to operate in a logical and scientific manner. Most of the research that
is being done in an organization is not in the Research and Development department. In
fact, it’s done throughout the organization.
As an accredited professional in an organization, particularly one with a university or
graduate education, you will be expected to work with sound research-oriented skills. In
most organizations, the responsibility for thinking in a systematic and logical manner is
everyone’s responsibility, rather than being concentrated in just one function of the business
or just being “management’s responsibility”.
Take a moment to think through the differences between these research roles, between your
academic hat and your business hat.
1.2.2 RESEARCH METHODS VERSUS RESEARCH METHODOLOGY
Many authors use these terms interchangeably, but there is a correct way of using them.
As students of “Research Methods”, we must know the difference. What is it? Textbooks
treat this in varying ways but research “methods” usually refers to specific activities designed
to generate data (e.g. questionnaires, interviews, focus groups, observation) and research
“methodology” is more about your attitude to and your understanding of research and the
strategy or approach you choose to answer research questions. This chapter will start with
a good look at research methodology, and then will go on to look at research methods.
1.3 THE NATURE OF BUSINESS RESEARCH
If you have ever used the phrase “research shows that…” in an assignment or conversation,
you will not be doing this again! Understanding Research Methods helps us to be specific
about the research we discuss, and to make sure that research comes from a valid source and
was collected and analysed appropriately. Many surveys are conducted every day throughout
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
11
the world to prove a particular point, to support an ideological argument, or just to sound
authoritative. We hear them and see them in the news media all the time. Some of this
“research” is a “vox pop” where someone, often a journalist, has asked a few people in the
street their view of a Government policy, or a product or service, or a current crisis. This
is quite different from the kind of business research we are discussing on this chapter.
In business, and for academic research, the questions we ask must be valid and fair, relating
directly to our need for information, in other words our research must have a clear objective
purpose, we are not collecting information for its own sake. Survey research software (e.g.,
Survey Monkey, etc.) makes it simple to construct and administer surveys, and many of
these are poorly worded and designed
We must also collect that information (data) in a fair and systematic way. For example,
we should think about who we ask for information, and how they will understand our
questions. If we cannot ask everyone involved, then we must be able to justify why we ask
only a certain section of that population. When using sampling, you must ask “to whom
can the results be statistically generalized?”
We must also analyse our data with great care in a systematic way. The rigour of our analysis
will have a major effect on whether our research results are valid or not. If we are trying to
determine which of a range of new technologies to invest in, then it will be very important
that we don’t skew our results towards a technology or application created by someone we
know, or that we don’t miss out certain relevant technologies, as these inaccuracies will lead
to a poor investment decision.
1.3.1 WHAT MIGHT BE SPECIAL ABOUT BUSINESS RESEARCH?
If we contrast research in business with, for example, research into chemistry, one particular
issue is clear: business research is not a single pure academic discipline like chemistry. If
we conduct research in the field of chemistry, we will certainly have to know a lot about
chemical concepts, the laws of chemistry and the history of scientific development in
chemistry as well as the context of current chemical research. There will be much to learn
about the field before we could become successful researchers in that field, contributing to
new knowledge.
However, in business the issues are not so narrowly focused. We will need to understand
things about a range of stakeholders; for example, managers, staff, customers and owners,
about business entities such as companies and partnerships and co-operatives, about economies
and how they affect business operations, about products and services and how they vary over
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
12
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
12
time, how they can be produced efficiently, about money and what regulates its availability,
how it produces profit, and Governments and how their policy affects business operations,
customers’ income and needs etc.
We can see that business is an umbrella term for many different things, and involves a
number of different academic disciplines, such as mathematics, psychology, sociology, physics,
economics, politics, history and language. So when we research into business or management,
we will be drawing on a number of different disciplines and domains. Business research is
multi- faceted and disciplinary.
Business research can also be conducted at different levels. We may want to find a way to
predict when a particular project might move to the next stage of the product life cycle.
This could involve a substantial piece of work involving customers, competitors and markets
as well as product strategies for resource use, marketing and sales. We could try some trend
analysis and aim to forecast future growth or decline in sales of our product against the
competition, we could do some desk research into government policy affecting this market,
we could interview experienced managers in the field to find out their subjective views about
the product’s predicted life. This is a complex piece of research, since there are so many
variables and stakeholders involved in influencing a product’s life cycle.
TMP PRODUCTION NY026057B
PSTANKIE
gl/rv/rv/baf
ACCCTR0005
Bookboon Ad Creative
4
6 x 4
12/13/2013
Bring your talent and passion to a
global organization at the forefront of
business, technology and innovation.
Discover how great you can be.
Visit accenture.com/bookboon
©
2013 A
ccenture.
A
ll rights reserved.
http://s.bookboon.com/accentureCZintl
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
13
Alternatively, we may want to find out how sales have changed over a period of five years.
This will involve “fact finding”, and may be simple to collect from financial statements,
and be expressed in a clear chart showing sales figures over time. Easy. But what if there
were major changes to products or services during that time? Or a move of premises which
caused a slump in sales during a short period? Or a re-branding exercise? We would have
to decide what depth or what level to use for our research, and for this we would need to
know its purpose.
You might be thinking that this sounds a bit complicated. After all, not every manager
or employee has studied business research methods, yet they still have to make decisions
affecting the business on the basis of what they find out. Fair point. Millions of business
decisions are made daily across the world without detailed research. What we are trying
to do by studying Business Research Methods is to give you the choice to do the research
systematically and rigorously. That way, your decisions will improve, and you won’t be
tempted to go with the first option, which may not be the best one.
Does this mean a lot of theory? Not necessarily a lot, but some will be helpful, in order
to interpret the “facts” that we find. Usually business research will be conducted to achieve
a practical outcome, and that practical outcome will be best understood in a context. A
theoretical context, for example industrial sociology, or economics, may help us to analyse
a situation more effectively and critically. It may even help us to challenge or move that
theory forward. While this book is not about critical thinking skills, it should be clear to
you that that is a fundamental skill to learn in your studies. It does not mean being “critical”
in a negative sense. It means asking searching questions to challenge the assumptions people
make, looking not just for what is said but also for what is not said and considering the
reasoning behind conclusions drawn. For a good presentation further expanding on critical
thinking, watch the following video: https://www.youtube.com/watch?v=oefmPtsV_w4
Bryman and Bell (2015) discuss the distinction between “grand theory” i.e. a theory dealing
with abstract ideas and/or relationships between factors and “middle range” theory which
deals with a more limited context (p. 21–22). Additionally, Saunders, Lewis and Thornhill
(2016) provide a summary of some research on “what theory is not”.
1.3.2 MODES OF KNOWLEDGE
One way of thinking about the knowledge that is created through business research is
provided by Gibbons et al (1994). These researchers talked about “Mode 1 knowledge” as
that which is created by academics for an academic intellectual purpose, to further and add to
what is known. This has to do with basic research and tends to be built on the foundations
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
14
of what was known before, just as in any academic essay, you must discuss what is known
(published) before you start to do your own research or consider how that knowledge might
be further discussed or developed. Who wants Mode 1 knowledge? Usually other academics.
An example of Mode 1 business knowledge could be: the concept of economies of scale.
The researchers distinguish this from “Mode 2 knowledge”, which is practical applied
knowledge and comes from collaborating with practitioners or policy makers, for example
managers in organizations. Who wants Mode 2 knowledge? People making business
decisions or developing policy as well as academics interested in applied research. This
kind of knowledge is much more dependent on an understanding of context because it is
essentially “real world” knowledge. It is no use knowing that generally there are economies
of scale if your business has overstretched itself by investing in a larger factory and profit
has reduced as a result. An example of relevant Mode 2 knowledge here would be: how to
calculate depreciation on capital investment with a particular country’s accounting standards
and how this might be used in conjunction with business strategy objectives for expansion.
Huff and Huff (2001) also suggest a third mode of knowledge. “Mode 3 knowledge”. This
is knowledge, which is neither produced specifically for academic purposes nor for direct
application to practical need, but for understanding the bigger picture in relation to society’s
survival and the “common good”. An example of Mode 3 knowledge might be: the impact
of capitalism on developing countries in the African continent. This kind of information
does not have specific immediate practical value (and would not find a business sponsor),
and it may not result from academic enquiry, yet it could be of profound importance to
international economic and social policy and business organizations in Africa.
Have a look on the web, use Google Scholar or another academic database or search engine,
to find an example of business research and then classify it into Mode 1, 2 or 3 knowledge.
1.4 WHAT KIND OF BUSINESS PROBLEMS
MIGHT NEED A RESEARCH STUDY?
Most work in business organizations, in whatever sector or ownership, will require research
activities. We have already discussed the idea that business research in the context of this
course is likely to involve some theory or concept as well as purely practical questions such
as “how does the product range compare in terms of contribution to profit?” Or “which
method of training has produced more output – coaching or a group training course?”
Both these questions have potential for theory application as well as simple numerical
survey, but some research problems are more obviously underpinned by theoretical ideas.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
15
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
15
For example, those which seek to generalize or to compare one organization with another:
“what are the most effective ways of introducing a new employee to the organization?” or
“how do marketing strategies differ in the aerospace industry?”
When choosing an area for research, we usually start either with a broad area of management,
which particularly interests us e.g. marketing or operations management, or we start with
a very practical question like those in the last paragraph, which need answers to help with
managerial decision-making.
Refining from this point to a researchable question, objective or hypothesis is not easy. We
need to do a number of things:
• Narrow down the study topic to one which we are both interested in and have
the time to investigate thoroughly.
• Choose a topic context where we can find some access to practitioners if possible;
either a direct connection with an organization or professional body, or a context
which is well documented either on the web or in the literature.
• Identify relevant theory or domains of knowledge around the question for reading
and background understanding.
Unlock your potential
eLibrary solutions from bookboon is the key
Interested in how we can help you?
email ban@bookboon.com
eLibrary
http://bookboonglobal.com/en/qualities/corporate-education-elibrary-solution/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
16
• Write and re-write the question or working title, checking thoroughly the
implications of each phrase or word to check assumptions and ensure we really
mean what we write. This is often best done with other people to help us check
assumptions and see the topic more clearly.
• Use the published literature and discussion with others to help us narrow down
firmly to an angle or gap in the business literature, which will be worthwhile
to explore.
• Identify the possible outcomes from this research topic, both theoretical and
practical. If they are not clear, can we refine the topic so that they become clear?
(For example, ask yourself the question, if I find an answer, then what use is it?)
1.5 WHAT ARE THE KEY ISSUES IN RESEARCH
METHODS WE NEED TO UNDERSTAND?
1.5.1 RESEARCH IS A MESSY ACTIVITY!
Saunders, Lewis and Thornhill (2016) provide a flow diagram of the research process. This
helps us to see the process as a logical progression, which has certain stages, and this process
would apply whether your research is for an academic purpose or a business purpose. However,
this model could give a rather misleading impression, as the authors mention. Let’s take just
two of the early stages: formulating the research topic and critically reviewing the literature.
Formulating the research topic, as we have seen above in the previous section, can take quite
a time. We start with a broad idea of an issue or area for research such as the impact of
flexible working on an organization, and this goes through many iterations before it turns
into a working title and clear set of research questions. Often the working title does not
get finalised until very near the end of the research, when the process and outcomes are
clearer, but because this is the first thing which appears in the process, it can seem, often
wrongly, to be a first stage. At best, the first stage is a tentative idea, sometimes a leap in
the dark, an idea we want to test out. All it needs to do at this stage is give us a direction
for research and some ideas about what to read and where to look for information. Much
later, the research topic will be the label given to the completed research and will be how
others navigate their way to our work, so by then it must be clear and precise.
Critically reviewing the literature – this stage seems to come early on in the research, and
that is how it should be, since we must read what is published on a topic before we can
begin to formulate clear ideas about how to proceed with primary research and which
questions still need answers. However there is no one set time period in which we read the
literature. We read as early as possible to get an idea of what is published, but we must
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
17
keep on reading throughout the research as new items may be published in the area, and
the primary research may lead us to form new questions of the literature, which involve
new literature searches.
However, when we write up the research, it is likely that the literature review will appear to
be an early and separate stage in the research process. In reality, it is iterative and “messier”
than this.
1.5.2 THE RESEARCHER AFFECTS THE RESULTS OF RESEARCH.
Researchers try very hard to be objective and balanced in their enquiries and their writing.
However there is no such thing as totally impersonal objective research. Imagine a scientific
model, which sets out a hypothesis or a contention such as “H1: this new computer keyboard
will improve typing speeds” and then seeks evidence to prove or disprove the hypothesis,
(this is usually referred to as deductive research). This could be considered the closest to
“objectivity”, especially when it is possible to experiment on one group and have a “control”
group of similar subjects for comparison. For our hypothesis, we could divide all the keyboard
users in our organization into two groups, time their typing speeds on the old keyboard
on a particular task and then, from the speeds produced, set up two groups, each of which
had a similar profile of typing speeds. Then we give a new task at the same time to each
group, giving one group the new keyboard. Measure the results to test the hypothesis.
This sounds pretty objective.
So in what way could we, as the researchers, influence the results?
Because researchers are people, not machines, not only will their method of research affect
their results, but their values will also affect results. The researcher’s mindset and personal
values and experience will provide a filter for which method they use and what they see in
the research results. This is often a consequence of the classic functional design model, where
employees with similar disciplines (e.g. marketing, accounting, operations etc.) are grouped
together for administrative purposes. An unintended consequence of this is “groupthink”
(Janis 1983). When an employee wears her “researcher hat”, regardless of their discipline,
they try to set aside as much as possible their internal bias and opinions.
Malcolm Gladwell (2000, 2005, 2008) helps readers to understand how we often make
quick decisions based upon subconscious processes working within our own minds. His
words will help you see the world differently and explain bias in everyday decision making.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
18
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
18
In our example about keyboard speeds above, can you see any possibility of bias in the
research method? Can you see any assumptions or values? Can you see any ways in which
we might look for particular results to confirm what we think?
Just to illustrate this idea a little further, imagine a company in which profit levels are
falling. The finance director may see a financial problem here and will research sales and
cost trend data, looking for that financial problem. The marketing director will look for
problems in the marketing strategy, or more likely the way other people in the business
have prevented the marketing strategy from being carried out effectively. The non-executive
director may see an industry trend as the problem, and will research professional literature
to support his or her idea. Each is likely to find the problem they look for, and they may
all be right to some extent.
In business research, we must try our hardest to look for possible bias in both how we
conduct the research and in what we think we have found. But since we cannot eradicate
all bias, we must also be explicit about the perspective which may colour our research, so
that readers of our results can understand we do not find “the truth”, just one version of
that truth in a particular context.
.
http://s.bookboon.com/AlcatelLucent
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
19
1.5.3 THE DIFFERENCE BETWEEN QUALITATIVE AND QUANTITATIVE RESEARCH
As we move through this book, we will be looking at a wide range of ways to approach
business research, especially in the third chapter when we look at research designs. For now,
it is simply important to distinguish two major approaches: qualitative and quantitative.
Of course, by now in your studies, you will have noticed that nothing is really “simple” in
academic work! So in order to talk sensibly about qualitative and quantitative approaches
we also have to introduce a few other ideas.
Deductive versus Inductive
As mentioned above, a deductive approach begins by looking at theory, produces hypotheses
from that theory, which relate to the focus of research, and then proceeds to test that theory.
But that is not the only way to use theory in research. An inductive approach starts by
looking at the focus of research (the organization, a business problem, an economic issue
etc.) and through investigation by various research methods, aims to generate theory from
the research. A simple way to put it would be: deductive reasoning moves from the general
to the specific and inductive reasoning moves from the specific to the general. For deductive
reasoning to work, we need to have confidence in the general premise, or theory, on which
it is based. For inductive reasoning to work, we need to make careful observations of the
specific situation and to consider possible causal links in that situation in order to produce a
reliable idea or theory which relates to other situations too. For a good presentation further
expanding on inductive and deductive reasoning, watch this video: https://www.youtube.
com/watch?v=ZBxE0y7b464
Sherlock Holmes, the infamous detective had the extraordinary ability to use logic and
reasoning to make accurate deductions from the evidence he collected. To most casual
observers, this evidence would seem meaningless and be overlooked or ignored. The process
of deduction requires accumulating evidence, asking the right questions (interrogating),
formulating and then evaluating a hypothesis, and reaching a conclusion (Smith, 2012).
The research process may require a mix of inductive and deductive reasoning in order to
identify and solve business problems.
Divergent and Convergent thinking
Inductive reasoning requires divergent thinking. However, people are better at convergent
thinking. In the diagram below, try to connect all nine dots with four straight unbroken lines.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
20
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
20
Figure 1: Nine dots puzzle
Divergent thinking solves the nine dot puzzle. See below. You have to think “outside the
box” for the solution. This is helpful in the problem identification or “problematisation”
phase of research.
AXA Global
Graduate Program
Find out more and apply
http://s.bookboon.com/AXA
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
21
1 2
3
4
Figure 2: Nine dots solution
In school, we are often taught not to “colour outside the lines”. Even answers to the most
perplexing problems, requiring rigorous research, are often out there in the fringe of our
perception. That’s what makes solving them so tricky. Once we have identified the underlying
problem, having engaged in a creative and divergent process, allowing lots of answers and
problems to be generally possible, then it’s time to move to a more convergent or evaluative
thinking model to complete the research and look for the solution.
Qualitative
methods
Perceived problem or symptom
Solution
GeneralGeneral
Specific
Specific
Quantitative
methods
Divergent thinking
(inductive reasoning)
Convergent thinking
(deductive reasoning)
Evaluative
process
Creative
process
Figure 3: Divergent then Convergent process in problem solving
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
22
Failure to explore the problem thoroughly at the start regularly happens in business. Time
pressures, boss pressures, customer pressures all force people to act quickly, often without
thinking things through. As a researcher, you need to resist the urge to jump to a conclusion;
aim to explore the problem creatively, widening out the possible ways forward, before you
use convergent and evaluative research approaches to reach a solution.
Positivist versus Interpretivist
A positivist approach is usually associated with natural science research and involves
empirical testing. Positivism states that only phenomena which we can know through our
senses (sight, smell, hearing, touch, taste) can really produce “knowledge”. It promotes the
idea of experimentation and testing to prove or disprove hypotheses (deductive) and then
generates new theory by putting facts together to generate “laws” or principles (inductive).
Positivists suggest that this kind of research can be “value free” (but see our discussion on
bias above). Finally, positivist research is about objective rather than subjective (normative)
statements and only the objective statements are seen to be the proper domain of scientists.
You can find examples of this approach in randomized controlled trials used for testing
new medicines. A control group is used where the new medicine is not used, or a placebo
is used, and the test results of this group are compared with results from a group using
the new medicine. This aims to find the truth about the new drug – did it help or didn’t
it. In business research, such trials are rarely possible because of the difficulty of creating
useful control groups, and the difficulty of narrowing down one changed variable (like the
drug in a controlled trial).
We contrast this with the idea of “interpretivism”. This is much more common in the social
sciences, in which business and management belongs. Because business and management
involve people as well as things, the interpretivist argument promotes the idea that
subjective thought and ideas are valid. This idea is broadly based on the work of Max Weber
(1864–1920) who described sociology as a social science “which attempts the interpretive
understanding of social action in order to arrive at a causal explanation of its course and
effects” (1947, p. 88). An interpretivist researcher aims to see the world through the eyes
of the people being studied, allowing them multiple perspectives of reality, rather than the
“one reality” of positivism.
Objectivist versus constructivist
This is a different angle on the ideas above. Objectivism states that social entities (like
organizations, societies, teams) have an existence, which is separate from the people in them.
You will have discussed the company as a legal entity earlier in your studies, we know it
has a legal existence. So from a legal point of view, objectivism is fine. Suppose now we
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
23
consider the idea of a “learning organization” (Senge, 2006). Clearly people in organizations
can learn, but to what extent could we say the organization itself learns? Who teaches it?
Who assesses that learning? This is a big debate, but we are using this idea to show that an
objectivist view would say there definitely is an entity (the organization) independent of
the people in the organization which can learn and foster learning. Constructivists would
say on the contrary that the organization has no independent reality. It is constructed in
the minds of those who think about it. So every time we think about an organization, we
are “constructing” it into some kind of reality. From this perspective, the organization only
has an existence in the minds of people, whether they are the staff or managers, customers,
suppliers, contractors, government, professional bodies or, of course, business researchers.
Quantitative versus qualitative?
So where do these different ideas take us in relation to understanding qualitative and
quantitative research strategies? We can use the other concepts above to help us build a picture:
A quantitative approach to research is likely to be associated with a deductive approach
to testing theory, often using number or fact and therefore a positivist or natural science
model, and an objectivist view of the objects studied.
A qualitative approach to research is likely to be associated with an inductive approach to
generating theory, often using an interpretivist model allowing the existence of multiple
subjective perspectives and constructing knowledge rather than seeking to “find” it in “reality”.
In current business and management research, you are likely to find a mix of both quantitative
and qualitative strategies, looking at observable objective facts where they might be seen to
exist, through the use and manipulation of numbers, and looking also at the perceptions
of those involved with these “facts”. So in a practical sense, we try to use the best of both
worlds to investigate the messy reality of people and organizations. Sound business research
often uses both strategies (“mixed-methods”) in coming to valid and accurate conclusions
for the problem-solving process.
You may wish to search the web for an article in the International Journal of Social Research
Methodology which is the transcript of an interview with the famous social anthropologist
Peter Townsend (Thompson, 2004). Although this is not research directly related to business,
you should be looking in this article to find some understanding of the complexity and
messiness of research, the influence of the researcher on the research and some of the
differences between qualitative and quantitative methods.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
24
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
24
1.6 QUESTIONS FOR SELF REVIEW
1. What is the difference between Mode 1, Mode 2 and Mode 3 knowledge and why
does it matter in business research?
2. What do you think will be the most difficult part of identifying research problems
for study and why?
3. Do you prefer the idea of conducting quantitative or qualitative research? Is this
just about statistics versus interview research methods? Check what each of these
means in terms of what you believe is the nature of knowledge and what you believe
about business organizations.
1.7 REFERENCES
Bryman, A. & Bell, E. 2015, Business research methods, 4th Edn. Oxford University Press,
Oxford, UK.
Gibbons, M.L. & Limoges, H. et al. 1994, The New Production of Knowledge: The Dynamics
of Science and Research in Contemporary Societies. Sage, London.
Gladwell, M. 2000, The tipping point. Little, Brown & Co, New York.
��������
�����
������
��������
���������
�����������
��������
�����
��
���������������������������
������
������
������������������������������ ����� �����
�
����� ������!���”�
������������
�
�
�����#$%����&'())%�*+������
�
���,���������
�-
.�
��������������������
������������
��
��������
��
����
�� ��������
��
����
���������
���
����������
����������
�����
��
���������
The Wake
the only emission we want to leave behind
http://www.mandieselturbo.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
RESEARCH PROBLEMS AND QUESTIONS AND HOW
THEY RELATE TO DEBATES IN RESEARCH METHODS
25
Gladwell, M. 2005, Blink: the power of thinking without thinking. Little, Brown & Co,
New York.
Gladwell, M. 2008, Outliers. Little, Brown & Co., New York.
Huff, AS & Huff, JO 2001, “Re-focusing the business school agenda”, British Journal of
Management, vol. 12, special issue, pp. 49–54.
Janis, I. 1983, Groupthink: Psychological studies of policy decisions and fiascoes. Houghton
Mifflin, Boston.
Saunders, M. Lewis, P. & Thornhill, A 2016, Research methods for business students. 7th Edn.
Pearson Education Limited, Harlow, England.
Senge, P.M. 2006, The fifth discipline: The art and practice of the learning organization, 2nd
Edn. Currency; Doubleday, New York.
Smith, D. 2012, How to think like Sherlock. London, O’Mara Books Limited, London.
Thompson, P. 2004, “Reflections on becoming a researcher: Peter Townsend interviewed by
Paul Thompson.” International Journal of Social Research Methodology, vol. 7, no. pp. 85–95.
Weber, M. 1947, The Theory of social and economic organization. (Translated by Henderson,
AM & Parsons, T.) Free Press, New York.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
26
2 PUTTING THE PROBLEM
INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING
RELEVANT LITERATURE
2.1 CHAPTER OVERVIEW
2.1.1 LEARNING OUTCOMES
By the end of this chapter successful students will be able to:
1. See how literature review relates to research projects
2. Identify literature from primary, secondary and tertiary sources
3. Undertake effective literature searching and become an effective consumer of research
4. Critically analyse literature for a research project
5. Apply Harvard referencing style and understand reference management
2.2 HOW DOES LITERATURE RELATE TO RESEARCH?
In Chapter 1 we discussed superficial research studies and the idea that theory was going to
be relevant to good quality business research, whether or not immediate practical questions
needed an answer. We also talked briefly about what theory was and what it was for. We
identified deductive research, which looked first at theory and identified propositions or
hypotheses, which the research was meant to confirm or disprove, and we found the opposite
direction, inductive research which begins with the study of a situation and then seeks to
generate theory.
Any research study, inductive or deductive, which you undertake for academic purposes,
will always require a review of relevant literature, and that will be a “critical” review, not
just a description of what others have said. When you are working in an organization, you
may find that there is no time to conduct a full literature review, but this chapter will try
to convince you that a clear idea of the theoretical context of a piece of research, helps you
to clarify its purpose and outcomes, and make clear for which situations your findings do
or do not hold. We all need to get into the habit of literature searching before working out
how to research a particular topic.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
27
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
27
At the very minimum, it is desirable to search professional or industry sources of information
before completing a research study of any kind at work. This will demonstrate your
professionalism and the breadth of your understanding of the field. Anyone can ask a few
people to fill in a questionnaire, but not everyone can make sense of the answers!
Robert Sutton presents the case for using scholarly research in business in an article in
Strategy and Leadership (2004). See if you can track down this article on the web and
see what you think of his views. There is a strong case there for what we do when we
search for and review what others have published in the field. Most business problems can
be illuminated by finding out what others have thought before and then trying to apply
some of their ideas – this is a natural response, as we chat with friends or colleagues about
problems. How much better then to look for and use the extensive work researchers have
put in on similar problems and gain the advantage of their scholarly work, provided we
can put it into our own context.
http://s.bookboon.com/elearningforkids
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
28
2.3 WHAT KIND OF LITERATURE SHOULD WE SEARCH FOR?
At an early stage in trying to identify a research project, any kind of literature may help
us. So a Google search (www.google.com) or using Wikipedia (www.wikipedia.com) or any
other general search engine will help us experiment with key words until we begin to find
material which is helpful.
As soon as we get a clearer idea of what is out there, we need to identify specific kinds of
literature, so that we can judge the relative merit of what we find for our research study.
2.3.1 PRIMARY LITERATURE SOURCES
These are the sources, which are the least accessible, often being company literature or
unpublished research, private correspondence and can include conference proceedings. What
is their value? In some cases this is very valuable information, which relates directly to the
research problem in which you are interested. For example, suppose you are researching
corporate advertising to children, an area, which is sensitive. Much information about what
companies decide, and why, will be contained in company documents and emails. However
access to primary sources is becoming easier as the web provides an instant publishing medium.
Blogs and personal websites are able to bring primary literature directly to the public,
however we should bear in mind that in such direct personal publication, there is no vetting
or monitoring process as there usually is in a secondary source. DO NOT confuse primary
literature sources with “primary research”. The latter means research you have conducted
yourself for a specific purpose (which produces more primary literature i.e. yours).
2.3.2 SECONDARY LITERATURE SOURCES
These sources are much more easily available in the public domain. They include published
books and articles in journals, news media and published business, government and
international body publications. Why are they secondary sources? Usually they reproduce
in a different format what was original primary work. For example, a researcher will often
first reveal their findings at a relevant conference and these may get published later in
an academic journal. Similarly, business consultants will report research findings to their
clients – often the company in which or for which the research was completed – but later
may seek permission to disseminate findings more publicly, perhaps in an anonymised or
generalized form, in a professional journal or news report.
www.google.com
http://www.wikipedia.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
29
Value is high but information in these publicly available media is likely to be less current
than primary sources, due to the time it takes to check, review, authorise and publish.
However, the web is making a huge difference here. Already many academic journals and
professional publications are available full text on the web. In some cases, there is no time
difference between primary and secondary sources.
For academic research, peer-reviewed journals, such as the Journal of Management Studies,
are considered more reliable sources than trade magazines or news channels, as the material
will have to be monitored by experts in the relevant field, who are not in the business of
selling publications. The process of peer review is usually rigorous. When someone submits
a paper to an academic journal, the editor will first check it fits with the aims and scope of
the journal, then check the content is relevant and clearly written and well argued. Only
then does the paper normally get sent to other academics not known by the author, who
are asked to read and review the paper and suggest improvements and whether they think
it is fit for publication. At least two, sometimes more, such referees are approached with
the paper. Only when they have sent in their reviews (and these are unpaid and have to be
fitted in around other academic teaching and research so they may take some time), can the
editor consider their suggestions and send them back to the author. Then the author has the
choice to accept and modify their paper, which again takes some time, before resubmitting
the amended paper. This process of peer review may happen two or three times for each
paper until everyone believes the paper is fit for publication. Then it gets into a queue for
the publisher to release, often electronically first, then on paper if it is a print journal. This
whole process can take more than a year, which explains the time lag on published academic
work and the serious rigour introduced by peer review.
Textbooks may also be peer-reviewed to some extent, but due to the time lag of publication,
and the need to reach a wider readership in order to recoup the costs of publication, they
tend to be less specialised than journal articles and possibly less current.
It is also possible to find academic journal articles which are themselves reviews of academic
literature. While most articles will mention and relate to studies to the published field, a
published literature review will provide a deep and wide range critique within a particular
field. Of course, the review will only be useful at a time close to its publication, since there
will usually be additions to the field after that time which are not included. There is a
rigorous method for undertaking such reviews, known as “systematic review”. A systematic
review is a type of literature review which tries to find, evaluate and synthesize all good
quality research evidence on a particular topic or research question. The review includes a
thorough assessment of methodological quality, looking in detail at how research questions
in this literature were connected to existing literature at the time, and how research methods
were chosen and how rigorously they were used.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
30
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
30
All this detail is to help the reader judge the reliability and quality of the research. Such
systematic reviews enable findings to be checked by readers as they show an audit trail of
review, and are usually high quality scholarly works. To undertake a systematic review, we
have to do a thorough search of the literature for relevant papers and show where and how
we searched in the review itself. This could be listing the databases and citation indexes
searched, plus journal websites and hand-searched individual journals and other sources.
Remember that although most literature can be tracked down somehow on the Web, much
literature is not easily found through standard search engines, and some will be inaccessible
online (the content may be subscription only or just the abstract and citation will be there,
or it is unpublished). This thorough search is conducted against clear criteria, determined
by the research question. The outcome is a synthesis of known research on the question
or topic and will be transparent and easy to follow. If there are limitations to the research
review, these will be clearly stated.
Consider doing a brief search using either Google Scholar (https://scholar.google.com/) or
another database or search engine such as Emerald for a “systematic review” of an area of
business literature. Read the abstract, or the full article if you prefer, and familiarise yourself
with a good quality literature review. Moving from a basic Google.com search to using Google
Scholar is an easy transition as they both work in much the same way. Google Scholar is
http://s.bookboon.com/academictransfer
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
31
a free, searchable database providing sources for both digital and hard copies of text from
a variety of publishers, including books, articles, papers, theses and dissertations, abstracts
and technical reports from business, governments, universities and academic publishers.
There is another great reason for using Google Scholar in your literature search: it offers
citations – links to the number of times a particular paper has been cited by others – which
can help us determine the popularity and to some extent the quality of a paper before you
even open it. Do bear in mind though that citations take time to appear. If the paper was
published in the last year or two, it is unlikely that other papers will have been published
citing it given the time it takes to publish after peer review.
2.3.3 TERTIARY LITERATURE SOURCES
These are collections of, or gateways to, secondary sources. They include encyclopaedias,
dictionaries, citation indexes, catalogues and web-based portals, databases and journals’
contents pages. We use tertiary sources to track down secondary literature which is relevant
to our field of study.
Useful lists and details of primary, secondary and tertiary literature sources are given in
most business research methods textbooks; for example (Saunders, Lewis, & Thornhill 2016,
pp. 83–89).
2.4 EFFECTIVE LITERATURE SEARCHING
Most of you will have received guidance on literature searching at some point in your studies.
Just in case you don’t remember it, or you would like a refresher, here are some tips. If you
are comfortable with literature searching, skip this section and go to 2.5.
Sometimes searching for academic literature is simple. You want academic information on
a specific topic or by a particular author. You put the information into a web search engine
and there it is.
But sometimes it can seem like looking for a needle in haystack.
For these times, consider a three stage search:
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
32
Stage 1
1. First, make sure you are using appropriate search terms. Perhaps you don’t know
enough about the topic to use the right vocabulary for searching. Or someone
mentions a theory or idea, which means nothing to you. As a first step, just enter
whatever you do know into Google or Wikipedia. Remember that to narrow a
search engine search you need to lengthen (i.e. make more specific) the search
string. E.g. rather than just looking for “motivation”, try more detail “Herzberg’s
theory of motivation”. Hopefully that will bring you fewer and more relevant results.
Using Google Scholar instead of general Google search will help to eliminate many
of the commercial and marketing oriented “hits” which are associated with your
search terms.
2. Are there American/UK English words and spellings to look out for? Use AND
& OR to refine your search. Use ‘truncation’ (e.g. sociol+ to find sociology or
sociological). Use ‘wild cards’ (e.g. p*diatrics to allow for different spellings of
paediatrics or pediatrics).
3. Once you have some results, scroll through and look for academic domain names
in your results. E.g. …ac.uk or ….edu.au Such academic sites are more likely to
give you reliable general information. There are often course outlines on the web
for Higher Education courses, which give basic information on topics or theories.
Use these academic links to find more vocabulary to describe your topic search. A
little reading at this point will help you narrow your second stage search.
4. Alternatively, you could look in a relevant book for useful keywords and definitions.
Try using the index!
Stage 2
5. Now you have better vocabulary to describe what you are looking for, try a relevant
database or portal (tertiary literature source). Examples for business research are
Emerald (www.emeraldinsight.com good range of academic management journals,
often fulltext), ABI/Inform Global (www.ovid.com wide range of periodicals and
reports), Business Source Premier (http://www.ebscohost.com/academic/business-
source-premier again a wide range of journals but also useful sources such as Harvard
Business Review, Academy of Management Review and professional journals), and
the Social Sciences Citation Index (http://mjl.clarivate.com/cgi-bin/jrnlst/jlresults.
cgi?PC=SS this only has abstracts and titles but gives a wide search of what is
currently being published in the social sciences).
http://www.emeraldinsight.com
http://www.ovid.com
https://www.ebscohost.com/academic/business-source-premieragainawiderangeofjournalsbutalsousefulsourcessuchasHarvardBusinessReview
https://www.ebscohost.com/academic/business-source-premieragainawiderangeofjournalsbutalsousefulsourcessuchasHarvardBusinessReview
http://mjl.clarivate.com/cgi-bin/jrnlst/jlresults.cgi?PC=SS
http://mjl.clarivate.com/cgi-bin/jrnlst/jlresults.cgi?PC=SS
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
33
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
33
6. Within the portal or database, use your more specific search terms and make sure
you are looking in the right place e.g. full text or abstract or keywords rather than
journal title.
7. Hopefully this search will find some useful academic articles. Read the abstracts and
if they look appropriate, try to go to full text if available. If not available see step 9.
8. Consider downloading 3 academic articles, which relate to your chosen topic. If
they are Full text, then scroll straight to the reference list at the end. Compare
them and see which authors and works appear in more than one of the three lists.
If you find some, you have probably found important academic sources on your
topic. Go back to your academic database (such as Emerald full text) to key in
these author names or titles to find good quality information on your topic.
Stage 3
9. Often the full text version of the articles you want is not available. It may ask you
to subscribe or pay, or it may simply not be online as full text. In this case, print
off the abstract and journal details of articles and take them to your library. In
some cases an inter-library loan or a photocopy can be procured for you.
http://thecvagency.co.uk
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
34
10. Don’t give up on important articles just because they aren’t fully online. Physically
going to the library may lead you to other similar information which is also not
online. Also books! Loans of articles and books can take some time so don’t leave
this until the assessment deadline.
11. Finally, remember that searching for relevant literature is just one, quite time-
consuming, stage of research. Leave plenty of time to do this, because much of
what you find and read will not be useable in your final research study, but without
searching and reading a wide range of literature, you are less likely to find the really
appropriate sources that you need.
2.4.1 HOW DO WE KNOW WHEN WE HAVE FOUND ENOUGH?
It is impossible to answer this question accurately. However, when you begin to find
references to the same ideas and authors in several articles you have found, you should
start to feel more comfortable that you have covered a good range of the literature for
this topic. While you are still discovering yet more and more angles to the topic in your
reading, keep on reading.
In most academic domains there are “seminal” articles or books, which are widely cited by
other authors in the field. These are usually important to read, preferably in the original
version if you can get hold of them. They will be the key pieces of literature, which have
shaped the thinking of researchers and practitioners in the field. We had an example of
this in the last chapter when we discussed interpretivist research approaches and mentioned
Weber (1947). Many writers on research methods, and sociology and philosophy, use his
work, so although it was written many years ago, it is still widely cited.
2.4.2 HOW UP TO DATE SHOULD REFERENCES BE?
As just mentioned, seminal works may go back a very long way in time. However, if we
are discussing a relatively modern issue, for example employment protection legislation,
then we need to use absolutely up-to-date references to show we understand current trends.
It is not that older articles are less important, just that they may have been superseded
in the field. Some academic journals regularly invite contributors to critique or respond
to new articles. If you are using one of these journals (an example would be Interactive
Learning Environments on learning technology published by Taylor and Francis), then it
is worth reading through the response articles as they often produce valuable critiques of
the main article. As a general rule, look for academic references within the last three years
for preference, going back further if you cannot find enough useful material.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
35
If you are using professional journal or media information (e.g. in UK Financial Times,
The U.S. based Wall Street Journal or People Management, a professional HRM magazine)
then aim to use very current material, within the last two years if possible. Out of date
news items are rarely useful in academic work, unless you are doing a historical analysis.
2.5 CRITICAL ANALYSIS OF LITERATURE
2.5.1 WHAT DOES “CRITICAL” MEAN?
The following table is an extract from The Study Skills Handbook (Cottrell, S., 2008). You
might consider using this when you are drafting a piece of work. Check for those parts of
your writing, which do the things on the left, and look across to see how you can redraft
them into a critical analytical style.
Descriptive writing Critical analytical writing
States what happened Identifies the significance
States what something is like Evaluates strengths and weaknesses
Gives the story so far Weighs one piece of information against another
States the order in which things happened Makes reasoned judgements
Says how to do something Argues a case according to the evidence
Explains what a theory says Shows why something is relevant or suitable
Explains how something works Indicates why something will work (best)
Notes the method used Identifies whether something is appropriate or suitable
Says when something occurred Identifies why the timing is of importance
States the different components Weighs up the importance of component parts
States options Gives reasons for selecting each option
Lists details Evaluates the relative significance of details
Lists in any order Structures information in order of importance
States links between items
Shows the relevance of links between pieces of
information
Gives information Draws conclusions
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
36
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
36
For much more in depth advice, consider looking at a book on critical thinking (e.g. Browne,
M.N. and Keeley, S.M. 2011)
2.5.2 CRITIQUING A PUBLISHED ARTICLE
Saunders, Lewis & Thornhill (2016) discuss Mingers’ (2000) idea of four aspects of a critique
(pp. 225–6) i.e. critiques of rhetoric, tradition, authority and objectivity. They also classify
the types and purposes of reviews; such as integrative, historical, theoretical, methodological
and systematic (p. 74).
You could add a practical critique to this list, for example ask the question “How does this
article or idea relate to a specific organization, sector or problem?” Could the findings be
generalized to a particular context? If the author did not set out to generalize the findings or
apply them to a particular context, then we cannot be negative about this, since it was not
the author’s purpose. Yet some concepts can be particularly useful in delivering an insight
to a practical business situation.
http://s.bookboon.com/Subscrybe
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
37
For example, Herzberg’s ideas on a two factor theory of motivation (1966) could be perceived
as a universal generalization about how we understand what motivates staff. This finding
could be practically applied by minimising focus on dissatisfiers and maximising the focus on
motivating factors. A narrower context might render the theory less powerful, for example a
workplace where staff delivered a routine public service with few opportunities for intrinsic
motivating factors (no career development or job change possibilities, no reward potential)
where extrinsic dissatisfiers could be more useful in relation at least to staff retention.
If you have difficulty thinking critically about something you are reading, you may wish to
try applying the following set of questions, developed by Professor Tom Bourner (2003).
What explicit assumptions are being made? Can they be challenged?
What implicit/taken-for-granted assumptions are being made? Can they be challenged?
How logical is the reasoning?
How sound is the evidence for the assertion(s)?
Whose interests and what interests are served by the assertions?
What values underpin the reasoning?
What are implications of the conclusions?
What meaning is conveyed by the terminology employed and the language used?
What alternative conclusions can be drawn from the evidence?
What is being privileged and what is off-the-agenda in this discourse?
What is the context of this discourse? From what different perspectives can the discourse
be viewed?
How generalisable are the conclusions?
Source: Bourner 2003
Now we will introduce one more way of critically reviewing academic literature – it is
practical and may save you some time in the long run. This is a method designed by UK
academics Mike Wallace and Alison Wray and it is starting to be widely used in the UK
in universities.
It consists of first producing a synopsis of anything you read, it may be an article or a
chapter of a book. You have to ask Five Critical Synopsis Questions of this article or chapter
as follows (and, of course, note down your answers).
Why am I reading this?
What are the authors trying to do in writing this?
What are the authors saying that’s relevant to what I want to find out?
How convincing is what the authors are saying?
In conclusion, what use can I make of this?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
38
From your answers to these questions, you can write a critical summary through the
following structure:
Title
Introducing the text – use Question 1 to write this
Reporting the content – use Questions 2 and 3 to write this
Evaluating the content – use Question 4 to write this
Drawing your conclusion – use Question 5 to write this.
When you are producing a literature review which will compare a number of articles or
chapters about a subject, if you have completed the synopsis questions, again you have a
ready-made set of information with which to compare articles:
So a comparative critical summary would take this structure:
Title
Introducing the text – use answers to Question 1 for all texts
Reporting the content – use answers to Questions 2 and 3 for all texts to answer this (you can
synthesise the answer rather than dealing with each one in turn)
Evaluating the content – use answers to Question 4 for all texts to answer this (you can easily
compare each text this way)
Drawing your conclusion – use answers to Question 5 to compare how useful each of the texts
is in relation to your research question.
This method is quite simple in structure, but will produce really good academic critical
analysis if you think carefully about your synopsis in the first place.
2.5.3 ARE THERE DIFFERENT WAYS OF READING ACADEMIC LITERATURE?
It is always tempting to read without writing. Reading for academic purposes, however,
invariably means reading with a computer or mobile device to hand, or pen and paper, so
that notes can be made during reading.
Even just highlighting important extracts as you read can be futile if you are not going to
go back over the highlighted text and read it again to make useful notes.
What kind of notes do you make? First it will be vital that you note down bibliographic
details (the information you would include in a reference) if you are making notes outside
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
39
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
39
the text itself (on a separate piece of paper, in a notebook, in a database or citation software).
Always remember key details such as volume and issue numbers of journal articles, access
dates if retrieving articles online, editors if you are reading a contributed chapter in a book.
On top of this, we need to note responses to what you are reading e.g. surprise, disbelief,
admiration, links to other things you have read, questions. Doing this helps to ensure you
don’t just record a description, but that you are starting to respond critically to what you read.
2.5.4 SHOULD I DEAL WITH EACH REFERENCE
SEPARATELY IN THE LITERATURE REVIEW?
It is possible to do this, but it is not best practice. If you look at an article from a peer-
reviewed academic journal such as Personnel Review (published by Emerald Emeraldinsight.
com), you will rarely find a section on the literature, which deals with each piece separately.
Instead you find that authors summarise and synthesise ideas from the literature, listing
references together where they all take a particular perspective, discussing them separately
only when the difference between them is important to the article or research study.
Emeraldinsight.com
Emeraldinsight.com
http://s.bookboon.com/EOT
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
40
This means that we can start to see some stages in preparing a literature review:
• General keyword search to learn about the topic area
• More specific search (online and in libraries) to identify high quality literature
(academic and professional) which relates to the topic area and research questions
• Using really relevant and good quality articles to identify others in the field through
their bibliographies
• Reading as much of what we find as possible until we are not finding new ideas
Noting in a retrievable format not only what these articles and chapters say but
their bibliographical details (including access dates for web material) and your
critical responses to them and links with other literature (similarities, differences)
• Reviewing notes and discarding items which do not fit the research study
• Making new notes of the themes in the relevant literature
• Writing the literature review on the basis of these themes, including appropriate
referencing.
Summarising what you have learned from the literature review relating to your research study.
For example, what gap your primary research needs to fill, or what hypotheses you could
test from the literature.
2.5.5 SHOULD I INCLUDE MY OWN OPINIONS?
Just recording your likes and dislikes about a piece of writing is insufficient, since this
just tells us about you, not about the piece of writing. Often universities spend some
time encouraging students not to include their own opinions in their academic work, and
this is because many students do include very subjective reactions to theories, models or
concepts, without arguing logically for their view or supporting it with evidence from
published literature.
However, your opinion is important. Provided your opinion is based on evidence and
logical reasoning, and is expressed fairly and objectively, it is valuable. You will find that the
simplest place to express your opinions, and develop them, is either in class, if you attend a
class, or online in a discussion forum about this topic or study. A discussion forum thrives
on argument and people expressing ideas and being open to others’ ideas. However, your
academic assignment will need careful and cautious monitoring of how you express your
views, to ensure that you express a balanced view, having weighed up, and referenced where
possible, both sides of an argument.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
41
Consider searching for an academic article, preferably a systematic review as mentioned
above, on a business topic which interests you, follow the general search advice in these notes
until you have tracked one down. Read and make notes on the article and then develop a
250 word critical response to the article, using the technique described above. This should
provide you with useable notes for revision later, as well as good notes on the article if you
are using it for an assignment.
2.6 USING HARVARD REFERENCING STYLE
For any research of professional standard, consistent referencing of all sources of information
used is vital. You will already have been doing this in your degree course, but at this stage
in your studies you will be penalised if the referencing style is not correct. When you have
produced your own research studies and published them, you will want them to be correctly
referenced so that no-one uses your work without attributing it to you.
The Harvard style is the most common referencing style in use in universities across the
world, but other styles do exist. The main point about Harvard style is that it does not
use footnotes, which can interrupt the flow of the text, and its bibliography is ordered
alphabetically by author surname. Most in-text referencing includes simply the author
surname(s) and year of publication, plus page number if a direct quotation is given. This
means it is easy to find that reference in the surname ordered bibliography.
The basic bibliographic style is author, year of publication, title, publisher, so even for web
pages without clear guidance on referencing, we have to look for an author (perhaps the
institution hosting the site – this is called a “corporate author”?), a year of publication (is
there a recent revision or last updated date?), a title (even of the page used) and a publisher
(this could also be the hosting institution).
Managing your references and citations is an important aspect of conducting your literature
search and it is essential in maintaining an accurate and up-to-date bibliography. In the past,
this was an onerous task. Today, modern word-processing software (e.g. Microsoft Word)
have built-in reference management tools. It is important to keep track of your references
as you work on your literature review.
Additionally, Zotero (www.zotero.org) provides a free stand alone online system for collecting,
organizing and maintaining your references and citations. Your academic institution may
also provide you with a paid-for program such as Endnote. All these programs allow for
easy integration into your literature review. They also allow you to choose from the most
popular style guides (which could be Harvard style, or APA, Chicago, MLA or Turabian
for example).
http://www.zotero.org
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
42
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
42
For more detailed guidance, especially on referencing personal correspondence and electronic
sources, try one of the following websites:
http://www.usq.edu.au/library/referencing/apa-referencing-guide
www.purdueowl.com they provide online guides for many of the standard writing styles.
2.7 QUESTIONS FOR SELF REVIEW
1. Why are critical reviews of relevant literature important in research studies?
2. What are the three main types of literature source and what are the key differences
between them?
3. If you were advising a novice researcher, how would you suggest they find useful
published work?
4. What should you include in the bibliographic details of a chapter written by three
contributing authors, within an academic textbook?
5. How can the five critical synopsis questions from Wallace and Wray help you to
avoid “description” in literature reviews?
PDF components for PHP developers
www.setasign.com
SETASIGN
This e-book
is made with
SetaPDF
http://www.usq.edu.au/library/referencing/apa-referencing-guide
http://www.purdueowl.com
http://s.bookboon.com/Setasign
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PUTTING THE PROBLEM INTO CONTEXT: IDENTIFYING
AND CRITICALLY REVIEWING RELEVANT LITERATURE
43
2.8 REFERENCES
Bourner, T. 2003, “Assessing reflective learning” Education and Training, vol. 45 no. 5, pp. 267–272.
Browne, M.N. & Keeley, S.M. 2014, Asking the right questions: A guide to critical thinking.
11th edn. Prentice Hall, Harlow, UK.
Cottrell, S. 2008, The study skills handbook. 3rd edn. Palgrave / Macmillan, London..
Herzberg, F. 1966, Work and the nature of man. World Publishing, Cleveland..
Mingers, J. 2000, “What is it to be critical? Teaching a critical approach to management
undergraduates”, Management Learning, vol. 31, no, 2, pp. 219–237.
Saunders, M. Lewis, P. & Thornhill, A 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
Sutton, R.I. 2004, “Prospecting for valuable evidence: why scholarly research can be a goldmine
for managers”, Strategy and Leadership, vol. 32, no. 1, pp. 27–33.
Wallace, M. & Wray, A. 2011, Critical reading and writing for postgraduates. Sage Publications,
London.
Weber, M. 1947, The Theory of social and economic organization. (Translated by Henderson,
A.M. & Parsons, T.) Free Press, New York.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
44
3 CHOOSING RESEARCH
APPROACHES AND STRATEGIES
3.1 CHAPTER OVERVIEW
3.1.1 LEARNING OUTCOMES FOR THIS TOPIC
By the end of this topic successful students will be able to:
1. Understand their personal position concerning the different perspectives of knowledge
and research which underpin research design
2. Identify differing research paradigms for business
3. Explain the key differences between qualitative and quantitative research methods
and how and why they may be mixed
4. Explain the concepts of validity and reliability in the context of business research
5. Choose a research design for a topic and generate appropriate research questions
3.2 DIFFERENT PERSPECTIVES OF KNOWLEDGE AND
RESEARCH WHICH UNDERPIN RESEARCH DESIGN
“Whether we are considering the physical sciences, the life sciences or the social sciences,
the research process begins with an interesting thought about the world around us. Without
this there is no research. The interesting thought or research question is the common
starting point of all research work in all fields of study. From this point research is always
concerned with the emergence of theory whereby concepts and notions develop through
the application of ideas, the observation of evidence and the evaluation of results. It is
worth always keeping in mind that the final result of research is to add something of value
to the body of theoretical knowledge.”
(Remenyi, D. 2005)
This is a great starting point, because this chapter is about how you start a research study –
and the first step is usually a thought or an idea or an unsupported opinion. But do we
really start there? Or do we have to take account of what is already there in our minds? For
example, we may have strong opinions, or no opinions, about reality, the world, politics,
history, people etc. When that “original thought” occurs to us, it comes up already embedded
in a context of what we already think we know about the world.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
4545
You might like to look online for an article by Bannister (2005) which has an intriguing
discussion of “realities” and the kinds of filters applied by people directly experiencing
an event, researchers looking at the event through their eyes and readers of that research
applying a third set of filters to reality.
In Chapter 1 we discussed ideas about positivist and interpretivist research approaches.
Basically, this is a debate about the nature of knowledge, which is also called “epistemology”.
Questions asked are “to what extent can we know something is true?”. “Does a phenomenon
(e.g. gender discrimination at work) have an objective existence, or is it only existing in
the minds of those who discuss it? Can we investigate it directly, or must be interpret its
meaning from what people say about it?
Then there is “ontology”. We have already begun to look at ontology through the Chapter
1 discussion of objectivism and constructivism. This is like epistemology but deals not with
the nature of what we can know or reveal as “true” but with the nature of social entities
such as organizations. Again the question is how and if they exist. We regularly refer to
teams in business studies. What are teams? An objectivist view of a team is that it exists in
itself, beyond the team members. A constructivist view of a team is that every time team
EXPERIENCE THE POWER OF
FULL ENGAGEMENT…
RUN FASTER.
RUN LONGER..
RUN EASIER…
READ MORE & PRE-ORDER TODAY
WWW.GAITEYE.COM
Challenge the way we run
1349906_A6_4+0.indd 1 22-08-2014 12:56:57
http://s.bookboon.com/Gaiteye
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
46
members interact, they have a concept of team which is there in their minds and which can
alter over time depending on how they interact, but does not have an independent reality.
Questions like this are relevant to business research because they will affect the kind of
research strategy we choose. A further idea which will be helpful in looking at where
researchers are coming from is the idea of paradigm.
3.3 IDENTIFY DIFFERING RESEARCH PARADIGMS FOR BUSINESS
3.3.1 WHAT IS A PARADIGM?
Try web-searching for the word “paradigm”. Is it only researchers and academics who use
this term? Is it helpful – or could you find a better word which is less academic? Kuhn
(1970) describes it as a cluster of beliefs, which guide researchers to decide what should be
studied and how results should be interpreted.
Saunders, Lewis & Thornhill (2016, p. 35) cite research by Burrell and Morgan (1979)
which offers four paradigms for social sciences research, within which we include business
research:
• Functionalist (problem-solving and rational approach to organizations)
• Interpretive (organizations only understood through perceptions of people about
those organizations)
• Radical humanist (organizations are social arrangements and research is about
changing them)
• Radical structuralist (organizations are a product of structural power relations,
where conflict is inherent)
These paradigms are held by the authors to be inconsistent and mutually exclusive with
each other, in other words, if you hold one paradigm, you cannot also hold a different
one. They therefore foster different research methods and focus on different areas for study.
For example, a functionalist paradigm takes a classic survey approach to issues, which are
thought to have objective reality. A climate survey of employees would be an example, made
to assess something “real” such as how employees feel about working in an organization, and
using a questionnaire with both quantitative and qualitative questions to gain descriptive
responses about that “reality”.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
47
An interpretive paradigm uses a qualitative research approach such as discourse analysis,
unstructured interviews to investigate perceptions and constructions of reality by “actors”
in organizations, i.e. employees, managers, shareholders etc.
A radical humanist paradigm would suggest again a qualitative method but looks not necessarily
at the perceptions of social actors in the organization but seeks to probe a deeper level of
values and social definitions, which underpin the organization. A relevant method would
be grounded theory, which looks for theory through a structured method of investigation
of what is said or written (inductive) and produces categories of idea, which can then be
used to characterize, develop or change organizations.
A radical structuralist paradigm may suggest a historical analysis of power in the organization,
by developing case studies or seeking to symbolize transactions between actors in the
organization, for example an analysis of employee relations over time.
This is one attempt to pull together the ontological and epistemological debates about
conducting social science research. It is the ontological and epistemological stance of the
researcher which affects the methodology and specific methods they choose for their research.
Does this make sense to you? We are talking about how you think about the world and the
stuff you find in it; for example, whether you believe in objective truth, or whether you
find all things subjective. What kind of status business organizations have, and the policies
and plans and structures and cultures they develop. As researchers, we have to develop a
clear sense of how we understand the world so that we don’t make the mistake of thinking
everyone else thinks about it the same way. We have to learn to be as objective as possible,
to recognize when our assumptions and philosophies may cloud our thinking and try to
dispel them for the purposes of research.
3.4 KEY DIFFERENCES BETWEEN QUALITATIVE
AND QUANTITATIVE RESEARCH METHODS AND
HOW AND WHY THEY MAY BE MIXED
You can have integrated paradigms as just mentioned, but you can also have a mix of
qualitative data from a case study approach and the perspective of “grounded theory” (Glaser,
B. & Strauss, A., 1967; Locke, K., 2001; Strauss, A. & Corbin, J., 2015) and quantitative
data from a subsequent survey. We will go into detail about grounded theory when we cover
qualitative data analysis. For now, you should know that this approach is interpretive, as
written and verbal data are collected and transcribed so that the texts can be fragmented
into ideas, categories and themes by the researcher. So such a mix involves mixed methods
as well as an integrated paradigm.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
4848
Research approaches or strategies need to be seen as related but distinct from the actual
methods used in research. Make sure you understand what methods are; for example:
experiment, interview, survey, case study, action research, grounded theory, ethnography,
archival research. This is by no means an exhaustive list of research methods, but it is a
useful broad range to keep in mind at this stage.
Why should a business researcher want to mix qualitative and quantitative research methods?
It is increasingly usual for business research to mix methods of data collection and analysis.
This can be done by using different data collection methods which are all either quantitative
or qualitative (e.g. web and paper survey, or interviews and focus groups) (a multi-method
approach), or you can use both qualitative and quantitative data collection and analysis
methods (e.g. survey and interview and action research) (a mixed method approach). One of
the reasons for this is “triangulation” where different methods of data collection and analysis
will both enrich and confirm the picture you collect of a situation. Often survey results are
used to map out a broad view of the research question, and to provide themes or areas for
investigation in more depth through interview. Triangulation can also provide a check on
findings from a particular method. It is worth looking at Dr J.W. Cresswell’s website on
mixed methods to see why the mixing of research methods, particularly combining statistical
How to retain your
top staff
F I N D O U T N OW FO R FR E E
Get your free trial
Because happy staff get more done
What your staff really want?
The top issues troubling them?
How to make staff assessments
work for you & them, painlessly?
DO YOU WANT TO KNOW:
http://s.bookboon.com/performancereviewpro
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
49
results with narratives which can add richness and understanding of the data collected in
social sciences such as business research (http://johnwcreswell.com/).
It will also be important to decide whether research should take a point in time approach,
i.e. look at a phenomenon (a new training course, induction process, technology, product
launch) at a particular time from the perspectives of more than one person – this is cross-
sectional research, or whether you have the opportunity to look at a phenomenon over a time
period (for example tracking a new product from launch to maturity, looking at industry
trends over time, or following cohorts of new employees through their employment over
an extended period) – this is a longitudinal study. Most academic studies for qualifications
tend to be cross-sectional as they are completed in a very limited time period. Longitudinal
studies usually require external funding to protract the period of research.
3.5 CRITERIA OF VALIDITY AND RELIABILITY IN
THE CONTEXT OF BUSINESS RESEARCH
3.5.1 RELIABILITY
Another term for consistency or repeatability over time. Reliability is required of research
studies. We must try to design research which is auditable i.e. transparent and clear so that
the reader can either undertake the same method themselves and produce the same results,
or at least the method is clear enough to instill confidence in the reader that the results
were not fudged in any way. (Triangulation will help here.) People not involved in studying
business research will often confuse reliability with validity, but in fact just because something
is reliable, it does not mean it is also valid. However, measurement and observation cannot
be valid if the study is not reliable.
The concepts of participant error, participant bias, observer error and observer bias are
factors which can affect the reliability and validity of collecting data and information in
survey and experimental research.
3.5.2 VALIDITY
Validity refers to the accuracy of a measurement or observation. There are three main ways
of characterizing validity in research studies. It is important that research methods have “face
validity” and “construct validity” and “internal validity”. Face validity means effectively that
the non-researcher or lay person can broadly see that this is a valid method of researching
http://johnwcreswell.com/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
50
this question; i.e. “on the face of it” it makes sense as a method. Face validity is important
to encourage participation in surveys or interviews, as well as other experimental or research
designs. We want to be able to answer the question “why do you want to know that?”. This
is a bit like your personal credibility in the business world, which is influenced by your
own behaviour and the extent to which it is professional.
Construct validity is a more complex idea and means that the method must actually measure
what you think it measures. There are, for example, statistical ways of checking surveys and
questionnaires to check that the questions are really asking what you think (factor analysis
and item response theory). Construct validity is particularly important in questionnaires
which are not administered face to face by a researcher but sent by post, email or done
online, as there is no chance then to discuss and clarify the meaning of a question. Sometimes
results can be invalidated because respondents have misunderstood a question and answered
in a way which was not intended. This is also referred to as “measurement” validity. We
can illustrate this idea by the famous IQ test which was intended to measure intelligence
(IQ stands for Intelligence Quotient) but includes items which bias towards particular
ethnic groups and educational norms. Or we could ask the question, do examinations test
knowledge? Is their measurement validity strong? Or do they actually test something else,
for example examination technique and memory skills?
Internal validity relates to causality, i.e. does factor X cause factor Y to happen? It is
sometimes easy to assume causality when in fact there is only association of two factors.
For example, does strong motivation cause or lead to effective teamwork, or does effective
teamwork lead to or cause strong motivation? In this case causality can work either way or
may be quite independent concepts. We cannot assume causality either way. In business
research, it is easy to make assumptions about a factor (or “independent variable”) causing
an effect (or “dependent variable). To test internal validity we have to ask the question:
does the independent variable account completely for a change in a dependent variable, or
are other factors affecting this outcome? Usually in business organizations, there are very
few simple cause and effect relationships. Does a performance bonus make someone work
harder? Internal validity essentially checks for the technical soundness of the study.
Other kinds of validity which are sometimes talked about include: external validity (this
is more often called generalisability, i.e. can we generalize the results of our study to other
contexts, situations or populations?) and ecological validity (this relates to whether the act of
researching a situation itself has an effect on that situation; it may be that findings from a
business research study are clear within the study, but when applied to a different “ecology”
i.e. outside the research study in “real life”, they no longer apply). This is discussed further
in Chapter 5: Choosing samples from populations.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
5151
3.6 YOUR CHOICE OF RESEARCH STRATEGY OR DESIGN
A research design is a grand plan of approach to a research topic. It takes quite a lot of
work and reading, as well as simply understanding your views as a researcher. For a start,
there will be no one right way of conducting business research – this will depend on a
number of factors such as research topic, audience for the research (you, your university
tutor or your company for example), time and other resources available to you, and the kind
of study which is considered appropriate for that topic. There will also be other practical
considerations such as access to information and people.
Suppose you wanted to investigate what shoppers thought about a particular marketing
strategy associated with an organization. Can you stand outside its shop and ask passers-by
questions? From an academic perspective, it is never that simple. There are ethical issues
(you would need permission from the retailer to stand outside accosting customers), practical
issues (you may cause an obstruction or even a breach of the peace in a public place!),
sampling issues (which ones do you talk to because you will have to make a choice), what
language will you use for your questions (relevance to the interview subject, their ability to
understand the questions), their motivation to respond (why should they? Do you offer an
incentive? Will that affect results?) and how do you analyse the results (quantitative analysis
of tick box answers? Textual analysis of their comments? Both? Record their body language
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
http://www.deloitte.ca/careers
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
52
as well?). And so on. Many of these questions are practical and detailed, but underpinning
your approach there will be philosophical assumptions which you must make explicit.
So designing your research will be vital and choosing a strategy will mean you have considered
your views on truth and knowledge, social entities, what business research can and cannot
achieve and how all this will affect what you actually do to answer a research question.
We have talked about the underpinning role of philosophy and research strategy, which
then guides your choice of research method (e.g. survey, interview, grounded theory etc.)
and whether they should be mixed, i.e. both qualitative and quantitative. These questions
need settling and justifying before you rush off to ask people questions.
3.7 CLASSIFICATION OF RESEARCH
It is also helpful to think about different models of business research. Most subject disciplines
are represented by various theories and models which give us language and ways of thinking
to describe and understand real situations. When we look at business research, this is no
different. The Figure below offers a way of classifying different approaches to business research.
Classification of business
research approaches
Exploratory
research
Descriptive
research
Correlational
research
Experimental
research
Quasi-
experimental
True
experimental
Figure 4: Classification of approaches to business research
The figure shows four distinct types of business research. First, we have exploratory research
in which we are trying to identify or clarify a problem. Sorting out real problems from
symptoms of the problem happens here. This phase is often qualitative, divergent and
inductive in nature. The outcome is a better definition and understanding of a problem.
This phase of a research project is overlooked, jumping to conclusions about the source of
a problem in order to correct a problem. Exploratory research can be the most important
phase of a research project. But beware: if you study the wrong problem, you will probably
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
53
come up with the wrong solution. Exploratory research can be an end in itself, or serve as
the beginning of further research.
Descriptive research answers research questions which are largely “factual” in nature. These
questions include those which start with “how”, “what”, “where”, “when”, “how much” and
“how often”. In business these questions can often be solved by finding people with the
answers, or by doing some simple digging for information (this kind of question is not
typically suitable for an academic business research project as it can be quite limited in
scope). Research questions are always stated in the form of a question.
Descriptive research typically utilizes descriptive statistics as the statistical method for use
in its analysis (e.g. mean, median, mode, standard deviation, variance, range, frequency
counts, etc.).
Correlational research looks for relationships between variables. These relationships may
be correlational in the statistical sense which means that when one variable varies, another
varies too, though not necessarily in the same direction. Correlation is an association of
variables but that doesn’t mean the relationship is one of cause and effect. For example, it
may be found that more mistakes occur in the office when the boss is present. That doesn’t
necessarily mean the presence of the boss has a direct causal effect on mistakes (it might!
But it is not proven unless a different kind of research is undertaken). Determining cause
and effect among and between variables is the fourth kind of research shown in this figure.
Correlational research uses correlation coefficients as the statistical method of analysis.
There are different types of correlational coefficients which depend upon the level of
data used to calculate the correlation coefficient.
Causal or experimental research tests hypotheses and is designed to explain “why” something
happened, i.e. to show cause and effect relationships. Conducting true experimental research
in a business environment is not an easy thing to do. However, there are variations of
experimental research which do allow for making reasonably accurate “cause and effect”
statements about phenomena and this is called “quasi-experimental” research. Experimental
research uses inferential statistics for analysis. Inferential statistics allows the researcher to make
statistically valid generalizations from a sample to a population. Hypotheses are statements
which are made (prior to conducting the research) and must be tested and then rejected or
“fail to be rejected” (not “accepted”), within statistically set parameters (the probability level).
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
5454
3.8 THE BUSINESS RESEARCH PROCESS
The figure below introduces an outline of the business research process, looking at its stages
systematically. This graphical representation looks like a static 5-step, two-dimensional
process which appears linear. In practice we don’t always start at one end and follow each
step through to the outcome. Remember, business research is often “messy”. So think of
this diagram as the “line of best fit” for how business research is conducted. In the real
world, business research is a dynamic, multi-dimensional and non-linear process. Conducting
business research is a bit like designing a bicycle and riding it at the same time. Downstream
processes can affect previously determined upstream conclusions.
TMP PRODUCTION NY026057B
PSTANKIE
gl/rv/rv/baf
ACCCTR0005
Bookboon Ad Creative
4
6 x 4
12/13/2013
Bring your talent and passion to a
global organization at the forefront of
business, technology and innovation.
Discover how great you can be.
Visit accenture.com/bookboon
©
2013 A
ccenture.
A
ll rights reserved.
http://s.bookboon.com/accentureCZintl
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
55
1. Analyse 2. Design 3. Implement 4. Interpret 5. Act
Recognise
organisational
performance
discrepancy
Develop
research
questions and
hypotheses
Select
appropriate
research
design
Conduct
research and
collect data
Run statistics
Make
appropriate
informed
management
decision
Answer
research
questions
and test
hypotheses
Prepare
research
report
Conduct
front-end
analysis
State the
problem
Figure 5: The business research process
Phase 1 – Analyse – is critical. This phase of the business research process is primarily
qualitative exploratory research as explained above. If you do not separate symptoms from
problem at this stage, you may identify the wrong problem and come up with the wrong
solution. This phase is often missed out or rushed through but that is neither efficient nor
effective for a business organization.
3.9 THE ACADEMIC BUSINESS RESEARCH PROCESS
We began this book explaining that we were talking about two different roles or “hats”.
Figure 5 is addressed to the business role – this is what a manager or consultant is aiming
to do for a business research project. However, this book is also designed to help students
of business, and you may find some small differences here in the process of achieving an
academic business project. The steps are very similar but you need to pay attention to an
assignment brief and the specific requirements of your study institution for the project. You
are also required to produce conclusions in an academic project, not just recommendations
for action. So below is the process to follow when conducting an academic business research
project. Just like the business process, this diagram can be a bit misleading in that it looks
linear, but in practice will be much messier. It is offered as a guide, not a requirement.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
56
Throughout this process, academic students are advised to keep in touch with their tutors/
supervisors and above all, to keep a record of everything you do related to the research
study for easier writing up later.
So what?
• Conclusions
• Recommendations
and action for
action
• Presenting results
Finding out more
• Primary
research
• Surveys,
interviews etc
• Recording data
• Analysing data
• Relating findings
to literature
What is known
• Literature search
• Asking questions
of literature
• Comparing
literature
• Conclusions about
what is known
Plan
• What you know
• What you need to
know
• What the
deliverables are
• What will be done
when and how
Purpose
• Is it an
assignment?
• Is it a problem?
• Is it a knowledge
gap?
• Talk it through to
define clearly
Figure 6: The academic business research process
3.10 QUESTIONS FOR SELF REVIEW
1. Review the ideas of epistemology/ontology, research paradigms, validity and reliability,
mixed and multi-methods and triangulation. How do all these relate to yourself
as a researcher?
2. If you used a mixed method approach, what reasons would you give to justify
this choice?
3. Which classification of research are you most comfortable with? Which do you
think might be your weakness?
4. Are you a convergent thinker or a divergent thinker? What are the advantages and
disadvantages of each?
3.11 REFERENCES
Bannister, F. 2005, “Through a glass darkly: fact and filtration in the interpretation of
evidence”, The Electronic Journal of Business Research Methods, vol. 3, no. 1, pp. 11–24.
Glaser, B. & Strauss, A. 1967, Discovery of grounded theory: strategies for qualitative research,
Alpine Publishing Co, New York.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing researCh aPProaChes and strategies
5757
Kuhn, T.S. 1970, The Structure of Scientific Revolutions, 2nd edn. University of Chicago
Press, Chicago.
Locke, K. 2001, Grounded theory in management research. Sage Publications, London.
Remenyi, D. 2002, “Research Strategies – Beyond the Differences”, The Electronic Journal of
Business Research Methods, vol. 1, no. 1, pp. 38–41.
Remenyi, D. 2005, “Tell me a Story – A way to Knowledge”, The Electronic Journal of
Business Research Methods, vol. 3, no. 2, pp. 133–140.
Saunders, M. Lewis, P. & Thornhill, A 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
Strauss, A. & Corbin, J. 2015, Basics of qualitative research: techniques and procedures for
developing grounded theory. 4th edn. Sage Publications, Thousand Oaks, CA.
Unlock your potential
eLibrary solutions from bookboon is the key
Interested in how we can help you?
email ban@bookboon.com
eLibrary
http://bookboonglobal.com/en/qualities/corporate-education-elibrary-solution/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
58
4 ETHICS IN BUSINESS RESEARCH
4.1 CHAPTER OVERVIEW
4.1.1 LEARNING OUTCOMES
By the end of this chapter, successful students will be able to:
1. understand how ethical issues arise in business research at every stage
2. identify the main ethical criteria used in Higher Education business research studies
3. propose strategies to ensure ethical issues in business research are addressed
appropriately
4. evaluate the ethical properties of existing research
5. understand ethical principles when planning a research endeavour
4.2 UNDERSTAND HOW ETHICAL ISSUES ARISE IN
BUSINESS RESEARCH AT EVERY STAGE
4.2.1 WHAT DO WE MEAN BY ETHICS?
Discussions of Ethics tend to sound worthy, sometimes border on the philosophical, and
occasionally stray right off the point. Why should this be? Ethics relate to moral choices
affecting decisions and standards and behaviour. So it is quite hard to lay down a set of
clear rules, which cover all possible moral choices.
Especially in research, where the practical aspects of a study (e.g. how and when to meet
people for interview, which data to sample, how to deal with someone changing their mind
about being part of a study, coming across information which you aren’t really supposed to
have etc. and the potential isolation of you as the researcher (not being in a group or class
all doing the same thing, but following your own research with your own objectives and
contacts), as well as possible inexperience of research at this stage of your studies, can all
contribute to a feeling of doubt and worry about what to do for the best.
Sometimes it can be quite a shock, when you have been used to being given pretty clear
ideas about how to do something, to find you have to make your own decisions about
how things will be done. Ethical choices we have never imagined can just creep up and
hit us. An obvious example would be when, as a very honest student, we start to collect
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
59
some data together and realize that one source of data is completely out of step with the
rest. As a professional researcher, that is an interesting challenge, which will create its own
new pattern of research and investigation. But as a business student with a fast approaching
hand-in deadline, the temptation to lose the odd data can be great.
We are not suggesting that we have to be great moral advocates here, perhaps that is a matter
for our own consciences, but we must anticipate as much as we can the moral choices and
dilemmas which the practice of research will bring, and try to find appropriate ethical ways
of dealing with them. Fundamental to our understanding of ethics in Business research is
the idea that all research involving human subjects needs to be governed by good ethical
practice, as well as any relevant legislation (e.g. protection of personal data). When we
conduct primary research there are some key principles of good ethical practice to consider:
• Non-maleficence
• Beneficence
• Autonomy
• Justice
4.2.2 HOW ETHICAL ISSUES CAN ARISE RIGHT
THROUGH THE RESEARCH PROCESS
How can these broad philosophical notions affect our research study? Here is a brief list
of the kinds of ethical issues, which can arise at different points in the research process:
• Access – physical, cognitive, continuing – just getting at the appropriate people can
be frustrating and tempt researchers to cut corners. Don’t be tempted.
• Participant acceptance/access (not just those in authority) – for example, you have
permission to ask people in customer-facing positions some questions, but they
don’t know you and are not sure how far to trust you – are you a representative
of management?
• Time – people just don’t respond in time for you to achieve project
• Your identity as researcher – what do they know about your study? And how the
data you collect will be used? And whose data is it, if they spoke or wrote it?
• Re-phrasing research questions on basis of feasibility (not wrong) i.e. you find that
your initial idea won’t work because you cannot gain access to the right people, so
you may need to review your research question to one which is feasible, provided
it is still valid and ethical.
• Convenience sampling – e.g. using people we know to take part, which could
produce participants who simply want to please you with their answers; or excluding
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
6060
troublesome views or statistics. E.g. including a poor sales year in an otherwise rising
trend. Reality is messy – do we want to smooth the mess and create simple answers,
or do we want to understand messy reality in order to change or anticipate it?
• Data recording – what if tape or digital recorder doesn’t work? Can data be recreated
from your notes? Do we pretend it worked?
• Interviewing – e.g. what if the first interview turns up new ideas, which are then
used in subsequent interviews – can you include that first one in your data set? What
if an interviewee starts to see things in a new light and uncovers painful memories
or ideas? Latter can also happen in focus groups- conflict, personal animosity could
develop – how can this be handled?
• Your role in the data – we have already mentioned this, the researcher is not an
object but a human being to whom people will react. What effect does this have
on your data? Does it affect validity of results?
• Transcripts – if you transcribe an interview or conversation, what happens to it?
Whose is it? How do you label it (Jo Bloggs’ interview?). And how exactly do you
transcribe? Do you include repeated phrases or words? Do you attempt to record
body language which may affect the meaning of what is said?
• Cheating in analysis when results don’t fit – this can affect both quantitative and
qualitative research methods. Remember that provided the process was justified
.
http://s.bookboon.com/AlcatelLucent
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
61
and conducted ethically and professionally, then a not very exciting outcome does
not really matter. We cannot all discover gravity or relativity, but we can all design
sound research plans and carry them out professionally.
• Confidentiality in the report of your research – how do you ensure it?
• Anonymity in the report – how do you deal with it?
• Use of research data for new purposes – can you recycle data? How could you get
ethical approval for this?
4.2.3 “HOW TO LIE WITH STATISTICS” ACTIVITY
We have all heard that statistics may not be “true”. However just how untrue may be
surprising. Visit https://www.fastcompany.com/1822354/7-ways-lie-statistics-and-get-away-it.
This link has some great information on statistical analysis, which we will return to later in
the text when we discuss quantitative techniques. However, this particular link shows with
examples how easy it is to present data in such a way that they tell an inaccurate story,
using value judgments, inadequate data summaries, inappropriate graph scales, incorrect
pictogram sizes, coincidences and generalizing from small samples.
Peter Corning (2011) writes in “The Fair Society” about how former British politician
Benjamin Disraeli is noted for the famous quote “…there are three kinds of lies: lies, damned
lies and statistics.” Disraeli’s quote prompted the book “How to lie with Statistics” (Huff,
D. 1954). Corning (2011) expands on this by classifying statistical trickery into several
categories, including statistical cherry picking, the magic of averages, creative graphing,
crunching diversity, small samples and mindless extrapolation. These books illustrate how
facts can be distorted to present any perspective one wants to promote. Being aware of these
techniques sometimes (consciously or subconsciously) used by researchers in presenting their
findings is important in ensuring at least that you are an effective consumer of research.
There is also some useful discussion of ethical research issues in an article by Jane Richardson
and Barry Godfrey (2003) which focuses on ownership and authority to use interview
transcripts which may be in the public domain.
https://www.fastcompany.com/1822354/7-ways-lie-statistics-and-get-away-it
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
62
4.3 ETHICAL CRITERIA USED IN HIGHER EDUCATION
BUSINESS RESEARCH STUDIES
Here are some broad pointers, check they make sense to you:
• Honesty and avoidance of deception (covert studies will rarely be sanctioned within
HE and would always require ethical committee approval)
• Following ethical codes of any professional body involved or associated with this
kind of research
• Full information about the purpose of the study and the researcher’s status and role
• Not to cause harm (including embarrassment, stress, discomfort, pain) by any action
or omission of the research study
• Gaining informed consent to participate in the research study unless this would
both invalidate the research and its absence could be approved by a research
ethics committee
• Respecting participants’ right to refuse to take part (at any stage)
• Respecting participants’ wish or need for anonymity and confidentiality
• Clarifying to participants and gate-keepers potential limits on anonymity
and confidentiality
• Respecting assurances given to participants and gatekeepers concerning anonymity,
confidentiality and use of data
• Maintaining objectivity during data collection, analysis and report stages
• Justifying and offer an audit trail for data collection and analysis
• Where any possible question arises from the above, seeking the advice and authorization
of the university or college Ethics or Human Subjects Research committee (though
this may be a requirement of all research).
4.4 STRATEGIES TO ENSURE ETHICAL ISSUES IN BUSINESS
RESEARCH ARE ADDRESSED APPROPRIATELY
Some key themes and strategies to anticipate and deal with them are given below.
4.4.1 STAKEHOLDER ANALYSIS
For ethics considerations and compliance, in business research, it generally can be helpful
to start by working out who are the stakeholders in your proposed study. This may include
the research participants, their managers and other team members, “gate-keepers” who
may be senior managers or specific post-holders who can authorize your research. It may
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
6363
also include shareholders (in a private quoted company) who may be affected if your
research is detrimental to the company and is leaked. It could include customers, for the
same reason, but it could be any kind of organization, yourself as researcher, yourself as
student, competitors to this organization or activity, suppliers? Can you identify any other
stakeholders? Some will be specific to the kind of research study undertaken; for example,
a study of recruitment practices could affect potential employees.
Once you know who might be affected by your research study, you could design a simple
risk analysis:
– for each stakeholder identify the type of risk from your research, its potential impact
(low, medium or high) and the probability that it will happen (unlikely, possible, probable).
Entering this into a grid, will give you a clear idea of priorities in designing an ethical study,
and should lead you to think about strategies to reduce undesirable impacts.
Participant anonymity is usually a basic requirement in business research, unless using a
research method where a particular identity is relevant to results and participants agree their
association with the research. So what does participant anonymity involve? It is not usually
just a case of not putting their names in the final report, though that is vital. It will be
AXA Global
Graduate Program
Find out more and apply
http://s.bookboon.com/AXA
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
64
important to decide whether you need to devise a code for each participant (so you know
who they are but they cannot be named by others), or whether this is not needed by the
study so no-one will have a code or a name. Can you refer to their title, role, function,
department, site etc.? All these, in conjunction with your results, may reveal identity. Is it
appropriate to record the participant’s names on questionnaires? (this issue is not just to
do with ethics, since anonymity also affects how we answer questionnaires). Can you stop
yourself referring to someone, in your study, to others in their company, who might try to
identify them? If you have, for good reason, collected personal details, have you checked
whether you comply with the requirements of any data protection legislation in your country?
Why do you need to know someone’s age or gender or ethnic group? Does it really affect
the research outcomes and thus will be important data to collect? Or could you redesign
your study so that this kind of data was not important and need not be collected?
4.4.2 INFORMED CONSENT
Once you know where to look for participants in your study, and you have identified
how to achieve ethical involvement for them and their organization, there is the practical
business of achieving their consent. Informed consent requires you to prepare for all research
participants some documentation which shows them what you are doing and why, what
their role in the research is, what will happen to the data you collect from them and what
they are agreeing to do. It will also usually set out how you will keep and dispose of the
data and how the required confidentiality will be ensured. It will also set out how the
participant can withdraw their consent at any time and you will not proceed with their
data/ interview etc. This is very detailed and seems like a lot of work, but in fact a short
text can often achieve all the requirements of informed consent. This, or a brief statement
referring to this documentation, must then be signed by your participants. Remember that
no undue pressure should be brought to bear on any participant or gate-keeper, since this,
however well-intentioned, will influence their involvement in your research and will prove
not only unethical, but may also invalidate results.
4.4.3 OBJECTIVITY
Let us assume that your motives are honest, in which case there are just two issues to tackle
here. The first issue is the way data are collected and recorded. You may be using a specially
designed relational database in which to record observations and related information, or we
may be talking about a highlighter pen and notes in the margin of an interview transcript, or
a clipboard and pencil. Whatever method is used to collect, and transfer data to a retrievable
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
65
record, then it must be designed for purpose, systematic and capable of capturing all relevant
details. Take for example a semi-structured interview method: what kind of system could
be used to record the interview?
Digital audio or video recorder? Notepad and pen? Pro-forma template including main
questions, and adequate space to record answers? Reflect for a moment on what kind of
issues could arise which might affect research objectivity depending on choice of system
used to transcribe or document a session.
Could any of these systems fail? If so, what would you do? How could you ensure continuing
objectivity if a problem arose during the session.
The second issue is when a research study is under way and something unexpected happens
to cause a problem with your data. This might be a rogue result which doesn’t fit the rest
of the data. Or a defective or failed recording. Or a key participant withdrawing from the
study (attrition), as they have a right to do. At this stage of the research, however honest we
are, there will be a temptation to fix the problem. So we should anticipate this temptation
and understand, before it happens, that that is the road to failure in research. Academic
and professional audiences will not be fooled, because they will understand and look for
such issues. The moral responsibility of the researcher is considerable and when researchers
are found to have transgressed, they are likely to be held to be accountable to authorities
and perhaps in the media. To test this, search the web for legal implications and media
coverage of “fixed” or falsified research. Sadly, there are many examples to be found, but
at least these will have been held to account publicly.
4.4.4 PRACTITIONER RESEARCHER OR INTERNAL RESEARCHER
This is an extreme case of having a potential unintended effect on the outcomes of your
research. If you are researching an organization of which you are part, (even if you just have
a casual or part-time job there) then you already have an understood role or status within
this organization. It will be difficult for you suddenly to put on an “objective researcher”
hat, and even if you could do this successfully, how easy would it be for your colleagues, or
subordinates or managers to see you differently in this role? However, an internal researcher
may be in a position to conduct a kind of research, which may be impossible from an
external perspective. Can you think of an example in business research?
It can be very tempting to undertake participant observation in a covert way in your own
organization, but this clearly raises ethical issues and possible bias. Could you possibly find
more useful and reliable data covertly, rather than openly declaring your intention and
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
6666
gaining official agreement for access? In a few cases, the answer may be yes, but if so, there
must be approval from any research ethics committee relating to your studies or research
(or professional body ethics approval e.g. relating to your work function) and in retrospect
you must inform those involved that the study took place and why access was not officially
sought in advance. Assurances must then be given about the use to which the research data
will be put and to what extent it will be anonymised. Spying is not research!
4.5 PLAGIARISM
It has always been possible to copy information from other sources into your work and pass
them off as your own. However, with the rapid expansion of digital media and mobile web
access this process has become much easier. Whenever cases of plagiarism are discovered
in academic work, there are serious implications for the student, resulting in cancelled
marks and sometimes greater sanctions such as forced withdrawal from a course. Digital
applications are not just good for finding and copying someone else’s work as your own,
they are also available to educational institutions to find plagiarism when it occurs. Clearly
plagiarism is an ethical issue. While many excuses, including cross-cultural practice, are
offered, it can never be good ethical practice to use someone else’s work without reference or
��������
�����
������
��������
���������
�����������
��������
�����
��
���������������������������
������
������
������������������������������ ����� �����
�
����� ������!���”�
������������
�
�
�����#$%����&'())%�*+������
�
���,���������
�-
.�
��������������������
������������
��
��������
��
����
�� ��������
��
����
���������
���
����������
����������
�����
��
���������
The Wake
the only emission we want to leave behind
http://www.mandieselturbo.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS ethiCs in business researCh
67
acknowledgement. This includes putting quotations from other people’s work into quotation
marks to demonstrate the existence of another source, as well as “in-text” referencing and
inclusion of full source details in the reference list. This does not just apply to text, but
also of course to videos, music and graphics of any kind. Most business research projects
undertaken for academic credit do not require images etc. but if they are necessary in your
work, always look for graphics licensed by authors under a Creative Commons licence, and
check the particular conditions under which it may be possible to use the graphic or image.
4.6 QUESTIONS FOR SELF REVIEW
1. When should we think about Ethics in a research study?
2. What elements would you include in a consent form for interview based research?
3. In what circumstances might covert research be justified? How would you deal
ethically with this?
4. What practical activities can you suggest to anticipate and prevent unethical
research practice?
5. Does your institution have a research or ethics committee? What roles do they play
in providing guidelines and approval mechanisms which must be followed in order
to conduct a formal or academic research study.
4.7 REFERENCES
Corning, P. 2011, The fair society and the pursuit of social justice, University of Chicago
Press, Chicago.
Huff, D. 1954, How to lie with statistics. W.W. Norton & Company, New York & London.
Richardson, J.C. & Godfrey, B.S., 2003, Towards ethical practice in the use of archived
transcript interviews, International Journal of Social Research Methodology vol. 6, no, 4,
pp. 347–355.
Saunders, M. Lewis, P. & Thornhill, A 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
68
5 CHOOSING SAMPLES
FROM POPULATIONS
5.1 CHAPTER OVERVIEW
5.1.1 LEARNING OUTCOMES BY THE END OF THIS CHAPTER
TOPIC SUCCESSFUL STUDENTS WILL BE ABLE TO:
1. Understand how and why sampling relates to business research
2. Identify and use a range of probability and non-probability sampling techniques
3. Select appropriate techniques for different research studies
4. Understand and assess representativeness of samples and generalisability from samples
5. Define key terms associated with sampling
5.2 UNDERSTAND HOW AND WHY SAMPLING
RELATES TO BUSINESS RESEARCH
Problem 1: the world is large and full of people. To find out things about people we need
to ask (research) them. We usually can’t ask all of them because the numbers make this
impossible. So we ask some of them. We sample from the population.
Problem 2: we wanted to find out things about people, so we researched a sample of them.
To what extent do our results relate to all people, and to what extent do they only relate
to our sample?
Problems 1 and 2 put sampling in a nutshell. Sampling is a practical way of studying people
and their activities, thoughts, attitudes, abilities, relationships etc. in relation to business.
But because we are not asking everyone in the chosen “population” (which could be the
members of a company, or all sales managers in the United States or the UK, or all applicants
for a particular job – any group we define in relation to our research objective), then how
can we have any certainty that our results can be representative of the whole population?
The crunch is that we don’t want just any sample, we usually want a sample to be representative
of a group (population). That would mean that our findings can be generalized to the whole
group. To make this happen, we have to learn about a number of issues and technical words
and phrases in sampling. In the next section there is a brief glossary based on Box 8.1 in
Bryman and Bell (2015):
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
6969
5.3 IDENTIFY AND USE A RANGE OF PROBABILITY AND
NON-PROBABILITY SAMPLING TECHNIQUES
The first table below gives definitions of some commonly used types of sampling.
To learn more about each technique, read the textbook and web search further.
Table of Sampling techniques.
Convenience sampling: Sample is chosen for ease or convenience rather than through random
sampling. This sounds underhand but is often used, at least in
pilot studies or short-term projects where there is insufficient time
to construct a probability sample. Therefore, where this is used,
the results cannot be generalized to the population (though many
newspapers would like you to believe otherwise!).
Multi-stage cluster
sampling:
When drawing a sample from a geographically dispersed population,
the logistics suggest that cluster sampling can help. The sampling
frame is first broken into clusters (e.g. geographic areas), and a
random or systematic sample taken. Then the population of each
cluster is sampled randomly to provide random sampling which is
logistically feasible. This can of course introduce bias, but using both
cluster and systematic sampling can usually produce effective samples.
http://s.bookboon.com/elearningforkids
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
70
Quota sampling: Regularly used in market research and opinion polling. Like a stratified
sample, this sample is chosen to include a certain proportion of
particular variables (e.g. gender, age group, ethnicity, socio-economic
group). Unlike a stratified sample, there is no random sampling stage;
the choice of respondent is up to the interviewer provided the profile/
quota is accurate.
Snowball sampling: Similar idea to convenience sampling, the researcher contacts an
initial group of people relevant to the research topic, and then uses
this group to contact others for the research. There is no sampling
frame here, so it is not random, but sometimes it is difficult to pre-
define the population (e.g. staff in a company who contribute creative
ideas). This technique is often used in qualitative approaches.
Purposive sampling: Using your own judgement to select a sample. Often used with
very small samples and populations within qualitative research,
particularly case studies or grounded theory. This approach cannot
yield any statistical inferences about the population. Cases may be
selected for being unusual or special or particularly related to your
research question.
Stratified sampling: Random samples are just that and they can appear surprisingly
“biased” or unrepresentative of the population (e.g., it would be
possible for a random sample to include only one gender, which might
affect your results). Stratified sampling specifies any characteristics,
which you wish to be equally distributed amongst the sample, e.g.
gender or work department. Provided the sampling frame can
be easily identified by these characteristics, then strata for each
characteristic are identified and within each group, random sampling
or systematic sampling can proceed.
Random sampling: (also called probability sampling – see explanation above). Define
the population. Define the sampling frame (F) (this may be the same
or it may exclude certain groups or individuals as not relevant to the
study). Decide the sample size (Z). Apply consecutive numbers to
the full sampling frame (F=N). Using a table (or computer program)
to generate random numbers, collect Z amount of different random
numbers within the range 1-N. Apply the chosen random numbers
to the sampling frame to identify your random sample.
This table below includes a simple glossary of terms you may come across when reading
about sampling.
Sampling glossary.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
71
Generalisability:
Being able to use sample results as if they applied to the whole population –
this must be based on sound sampling processes
Non-probability
sample:
Random selection was not used so some units in the population may have
had a higher chance of being selected (e.g. pointing to a crowd and saying
“You, you and you!” to the people in front) Non-response: a source of
non-sampling error when someone in the sample does not respond (e.g.
to questionnaire or interview). A fair amount of this is normal and there
are many reasons for it to happen (e.g. away on holiday, lack of time, lack
of interest, doesn’t understand question etc.).
Non-sampling error :
As sampling error, but these differences do not result from the sample
chosen, instead they result from the sampling process (e.g. non-response,
errors in sample frame, wording of questions, data analysis)
Population: The full universe of people or things from which the sample is selected
Probability sample:
A sample selected using random selection (this is not the same as “selected
randomly” – Why?) so that each unit in the population has a known (e.g. a
10% or 50%) chance of selection. Probability samples keep sampling error
low and usually offer a sample which can be seen to be representative
Representative
sample:
One which reflects the population accurately – showing the same
distribution of characteristics or variables as the whole population
Sample: The section of the population chosen for study
Sampling error: The difference of results between a sample and that of the whole population
Sampling frame:
A list of all people or units in the population from which a sample can
be chosen
Systematic sample:
Doing without random numbers in selecting a “random” sample. Sample
is chosen directly from the sampling frame (which ideally should not be
in any specific order except alphabetical). Once you know the sample
proportion required e.g. 1 in 20, start with a random number generated
item in the list, then choose every 20th name until the sample is complete.
Random number
tables:
Lists of numbers which are randomly generated – there is an example of
such a table at Appendix 4 in the textbook. Used in random sampling.
Use whatever digits in the random numbers apply within your sampling
frame total and ignore duplicates. You may find it is simpler to use
Excel spreadsheet function to generate random numbers. Formula to
find a random number between 0 and 100 is =RAND()*100 Use F9 key
to recalculate.
Sampling fraction:
Number required for sample divided by number in total sampling frame
expressed as a fraction or percentage.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
7272
Of all the sampling techniques included in the table, quota sampling and convenience
sampling, to some extent snowball sampling, are the least “statistically accurate” in nature.
These techniques offer varying levels of generalisability but are always less than a random
sampling method. Think about these three techniques and decide how justified you think
each is for conducting business research.
5.4 SELECTING THE SIZE OF YOUR SAMPLE
When we are designing a research study, the most common question about sampling is –
how large should the sample be? In the definitions of random sampling above, we have
ignored this question so it is now time to tackle it. Unfortunately there is no right answer
to sample size. You cannot just apply a consistent proportion to the total sample frame.
Instead the following issues need consideration:
http://s.bookboon.com/academictransfer
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
73
5.4.1 ABSOLUTE SAMPLE SIZE:
It is more important to look at the absolute size of a sample than its relative size in relation
to the total population. Imagine 10% of a population as a possibly sensible sample. If the
population total is 100,000, then your sample size is 10,000 – yes this would probably be
a good sample size (but see the next problem on this list). However, if we apply a 10%
sample size to a population of 10, we have a sample of 1 unit or person – essentially a
Case Study (N=1). We can see that this unit or person could be quite unrepresentative of
the total population by itself. So relative sample size is not important. Absolute size is. The
bigger the sample size, the more the sample is likely to represent the population and the
lower is likely to be the sampling error. (Referred to as the Law of Large Numbers.)
5.4.2 STATISTICS AND THE CENTRAL LIMIT THEOREM:
The larger the absolute size of a sample, the more closely its distribution will be to the
“normal distribution” (What is this? If you have not done any work on statistics before, do
some quick web-searching or look at the index of the textbook to find out). If you wish
to conduct a statistical analysis on your data, the minimum size of sample for any one
category of data should be 30, as this is most likely to offer a reasonable chance of normal
distribution. If your sample frame is 30 or less, then it would be wise to include the whole
frame, rather than sampling.
When small sample sizes are used, a family of inferential statistics called “non-parametric
statistics” should be considered for use in the data analysis.
5.4.3 MARGIN OF ERROR:
The expected margin of error is affected by absolute size of sample within a population. Note
that a 5% margin of error (which is the same as saying 95% certainty) is the maximum
normally appropriate for rigorous research. If your population size was 50, you would have
to include at least 44 of them to achieve a 95% certainty that the sample would represent
characteristics of the population. A very high proportion of the population will be needed to
achieve 99% certainty. There is a diminishing need for higher samples at the high population
end of the table (the figures to achieve 95% certainty for a population of 1m are the same
as for a population of 10m).
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
74
5.4.4 TIME AND COST:
Bryman and Bell (2015) suggest the law of diminishing returns kicks in at around a sample
size of 1000 – i.e. that precision in the data increases up to a sample total of 1000, but
then begins to decrease, making it less worthwhile to interview or survey more than 1000
(p. 199). Of course, the population you are researching may be way below 1000 in total,
and it may in any case be very costly or time-consuming to use a large sample size. Practical
considerations are important in research studies. Just bear in mind that if you choose a
sample size which is small in absolute terms, then you must justify this action and take into
account the fall in generalisability and representativeness which may result.
5.4.5 NON-RESPONSE:
Inevitably your respondents are less likely to be as motivated as you, the researcher, about
your research, so some – and sometimes a majority – will not respond, i.e. refuse to take
part. On top of this, some of those who do respond may not produce “useable” data (e.g.
you may find that a high proportion of questions in a survey are unanswered, or that some
people or units in your sample frame have moved away, changed job, stopped functioning
in the role you expected etc.). All this is taken into consideration when a) choosing your
sample size and b) calculating the actual response rate. If the pertinent demographics or
characteristics of the non-respondents appears to be random, then non-response error is not
that much of a problem. However, if certain non-respondents represent a greater proportion
of a significant demographic or characteristic of the sample, the problem of sampling error
may arise.
Number of useable responses
Total sample – unsuitable or uncontactable units
5.4.6 VARIATION IN THE POPULATION:
If the population you are studying is highly varied, then the sample size will need to be
larger than if you are studying a population with less variation (e.g. people who have chosen
to join a membership organization).
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
7575
5.5 UNDERSTAND AND ASSESS REPRESENTATIVENESS OF
SAMPLES AND GENERALISABILITY FROM SAMPLES
Even if we use probability sampling techniques, we can only hope to produce generalisable
outcomes in relation to the population we were sampling. So if all questionnaire respondents
are chosen from one company or organization, the best to hope for is that our results can
be generalized to the whole workforce of that company or organization. We cannot assume
that these results will in fact describe other workforces, as very different conditions and
variables may apply in other organizations.
The Sample versus the Sampling Frame. This issue has to do with the accuracy or
“completeness” of the Target Population. Suppose you want to conduct research on all people
with telephones in a given city. Where would you get the “list” of these people? Whatever
“object” or device you use is the “sampling frame”. What are some potential problems that
lie between the composition of the actual population, the sample and the sample frame?
A sampling dilemma: You are a store owner and you want to know the average age and
income of your “customers”. How would you identify them? What is your sampling frame?
Do you really ever entirely know who all your customers are? Think about how customer
“churn” (turnover) impacts your estimate.
http://thecvagency.co.uk
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
76
In a similar way, we could conduct a large sample study by random sampling a country’s
population based on official census statistics, and if the study was large and rigorous, we
might propose conclusions, which apply to this country’s population (with a specified degree
of confidence in the statistics). However, we cannot then generalise these conclusions to
other countries without further research, nor can we apply these conclusions over time to
the same country, as major variables could have changed over time. Think back here to
what we discussed earlier about epistemology – what we can really know.
We find this kind of generalization being made all the time in the media. For practical time
and cost reasons, media production teams often take quota sampling research (or research
done by more dubious methods) and suggest its applicability to everyone watching or listening
to a programme. Look out for examples and try to find out what kind of sampling was
applied to their research. Remember the ethics discussion about not causing “harm” – how
does this relate to TV, radio or webcast research you come across?
If you are worried about the representativeness of your sample, in some cases it may be
possible to check this by using a test of statistical significant difference to compare the profile
of characteristics in your sample with that of another data list e.g. a census or company
database. Clearly if there is no statistically significant difference between your sample and
the full population data list, you have added more authority to the representativeness of
your sample.
If you are using a non-probability sampling technique then even the flimsy size rules
associated with probability sampling fall away. Your sample size for purposive or snowball
sampling will really depend on your research questions and objectives. In qualitative research,
the focus will not be on trying to estimate things about a population, but in trying to
understand or relate the data to theory or ideas. How many people do you need to talk to,
to understand their perception of something for example? It could be just one. Or it could
be several or many. The question is here, what are you trying to find out and what sample
size would give me confidence that my results had validity? We will go further into this
when we discuss different qualitative methods, but often a good lead can be taken from
research studies in peer-reviewed academic journals, where information has been given about
sample size in relation to research question. Find one that is close to your area of study
(which you would want to do anyway in your literature review) and check the sample size
studied in this type of enquiry.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Choosing samPles from PoPulations
77
5.6 SAMPLING SIMULATION EXERCISE
Go to the web link provided below to complete a sampling simulation and exercise which
demonstrates the principles of random sampling. You will be asked to select a “representative
sample” from an entire population of small circles. The circles are all of different diameters.
http://www.learner.org/courses/learningmath/data/swfs/1d_circles.swf
After you have completed the exercise, answer the following 4 questions:
• What was the average diameter of your (non-random) selection?
• What was the average diameter of the random selection?
• Which was closer to the actual average diameter of the population (all 60 circles)?
• If you increased your sample size from 5 to say 25, how do you think it might
affect the average diameter of your selection?
Explain how this simulation could apply to a real world business environment or example
(i.e. translate the “dots” into employees, or customers, or products of the assembly line
etc.) and describe how the simulation concepts would apply. Try to incorporate correct
terminology from the chapter into your explanation.
5.7 QUESTIONS FOR SELF REVIEW
1. Why are random numbers useful for sampling?
2. Why don’t academics consider convenience sampling more often?
3. How do you calculate a response rate?
4. What kind of minimum size would you need in a sample used for statistical inference?
5. What level of certainty is needed for statistical sampling in academic research?
6. What reasons would you give for not exceeding a sample size of 1000?
5.8 REFERENCES
Bryman, A. & Bell, E. 2015, Business Research Methods 4th edn. Oxford University Press,
Oxford, UK.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
Thompson, S.K, 2012, Sampling 3rd edn. John Wiley & Sons, Hoboken, NJ,
http://www.learner.org/courses/learningmath/data/swfs/1d_circles.swf
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
78
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
78
6 QUANTITATIVE RESEARCH
METHODS: COLLECTING
AND ANALYSING DATA
6.1 CHAPTER OVERVIEW
6.1.1 LEARNING OUTCOMES
By the end of this chapter successful students will be able to:
1. Anticipate how the research design is affected by data collection and analysis tools
2. Recognise different types of data for analysis
3. Code and enter data for computer analysis
4. Choose appropriate ways to present data through charts, tables and descriptive statistics
5. Select appropriate statistical tools for the research variables
http://s.bookboon.com/Subscrybe
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
79
6.2 ANTICIPATING HOW THE RESEARCH DESIGN IS AFFECTED
BY DATA COLLECTION AND ANALYSIS TOOLS
It is never too early to start to think about data analysis. A common problem with research
studies is that we focus mostly on our research questions and finding samples, discussing
methods etc. and don’t ask simple questions about what data we are looking for and how we
will then analyse that data. Asking these questions early on, can avoid much disappointment
later, when we realise that the data collected simply can’t be analysed in a straightforward way.
Suppose for example that you want to know the three most useful management textbooks
that a large group of 100 managers have found effective. The question might look like this:
Q1 What are the three management books which have been most useful to you so far in
your management career?
You might leave three lines of space so that the respondents can write in their answers.
Think about how this might be coded as a question response for analysis. Since most
managers will not choose the same three, you will have a wide range of different answers.
We cannot code each book separately with a sample size of 100 and 300 potential books
in the answer range. So can we make any useful data out of this question?
You might answer that you wouldn’t ask this kind of question anyway! However it is a form
of question which is quite common e.g. What five competencies are needed by successful
salespeople?, What are the three most important experiences which have helped you to
achieve your current senior role? What three benefits do you feel you have gained from
mentoring? Etc. etc.
It is possible to turn the question into a list of possible answers from which respondents have
to tick three which apply to them. This means you can give each possible answer a unique
code in advance and then count the frequency with which each code is used. However, if
you want your respondent to have a free answer choice, because perhaps you really don’t
know what you might find out, then we have to delay coding until we have received some
answers. If we take the first 50 responses and make notes on the characteristics, which define
the responses, it becomes possible to group the responses. Once grouped, a code can be
assigned to each group and you can then go back and code each answer according to this
pre-defined group. In our question, we might find that answers include classical management
textbooks from the twentieth century (coded 1), simple How to…guides (coded 2), books
by management “gurus” e.g. Tom Peters, Charles Handy etc. (coded 3), books which are
not about management directly but have given readers inspiration (coded 4) and books
which don’t fall into those categories (coded 5). We now have 5 unique codes and can go
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
80
through all the responses collecting numerical data for each code. Now we have a data set
for analysis.
A much more simple issue is the questionnaire which contains only yes/no answers. Think
about analysing this data. A set of data results is going to look pretty boring, and how much
is it going to tell you about your research question? In the next chapter we will investigate
questionnaires further, but for now, we need to think about the data which will result from
our questions, how useful it might be, and how we might analyse it.
6.3 RECOGNISING DIFFERENT LEVELS OF DATA FOR ANALYSIS
Think of “Numbers” as the official language of business. All competent business people
speak in numbers. Business professionals don’t just say ‘we did well yesterday’. We say
“productivity was 8,000 units, a .5% increase from yesterday”. We also don’t merely say
“margins are fine”. We say “margin is 3.76%”. It’s the difference between being qualitative
or quantitative. And statistical method is the way we “process” this data.
Another name for numbers is data and another purpose of business research is to take this
data and convert it into meaningful information and knowledge for use by the organization
in order to improve the quality of decision making. The business research process helps the
business professional collect meaningful data and convert it into knowledge, information
and power. This is what is meant by what we often hear as knowledge is power in the
organization. As a business professional you are paid by your organization for what you
know. Your knowledge is power and that’s what you bring to the organization.
Not all data and numbers are the same. In fact, data can be classified into four levels,
forming a hierarchy from the lowest level of data to the highest level of data. The statistics
that you can generate and use depends upon the level of the data you have to work with.
It’s not that low level data is useless, it’s just not as robust as high level data. And because
of its hierarchical nature, data at higher levels can be treated as if it’s at a lower level, but
lower level data cannot be treated as higher level data.
6.3.1 NOMINAL DATA
The first or lowest level of data is nominal. Nominal data uses numbers to represent a
category. Nominal level data does not imply any form or rank order of importance or
power. Numbers are merely used to classify things into categories. A survey asking people
where their car was made, using numbers to represent these countries makes sense from
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
81
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
81
a data processing perspective. A “1” is assigned for GB, “2” for USA, “3” for Germany.
“4” for Japan, etc. The “number” merely classifies the category and assigns it a number. A
social security number, a phone number, your student number, and an employee number
are examples of nominal data. The numbers on the strip assigned to football players is also
a nominal level number. Nominal level data can have 2 or more categories. When nominal
level data have just 2 categories (yes/no; on/off, etc.) it is referred to as a dichotomous
variable. Dichotomous variables are a special case of nominal level data.
6.3.2 ORDINAL DATA
The second level of data is called Ordinal data. This has all the properties of nominal data
but it also infers a rank or order of importance from lowest to highest. Ordinal data is often
the result of questionnaires where grouping data makes sense. For example, questionnaires
might ask how old are you? (1) 18 to 25, (2) 26 to 40, (3) 41 to 55, (4) 55 to 67 or (5)
68 or older. Ordinal level data “ranks things from low to high, but there is not an implied
equal interval between one number and the next.
Nominal and ordinal level data collectively are often referred to as “low level” (or
qualitative) data.
http://s.bookboon.com/EOT
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
82
6.3.3 INTERVAL DATA
The next highest level of data is called Interval level data. Interval data has all the properties
of ordinal level data but it is more powerful than ordinal level data because a scale value is
used. That is, interval level data implies an “equal interval” between one number and the
next – that’s the distance between one number and the next. This data is meaningful, but
implies an arbitrary, rather than an absolute zero. Data from the social sciences is often at
the interval level, because of this issue – it lacks an absolute zero. As a result, researchers
often treat interval level data as if it is ratio level. The interval data level was created in
order to allow “quasi-ratio” data to be treated as such for statistical purposes.
6.3.4 RATIO DATA:
The highest level of data is called ratio level data. It has all of the properties of interval
level data but it also has an absolute zero. An absolute zero means that “0” indicates the
absence of that property, making ratios meaningful. This is the most versatile data for analysis
purposes. For example, your age, specific salary, your weight, the number of employees in
your company is ratio level data. 100 employees is twice as many as 50. A person weighing
200 pounds weighs twice as much as a person weighing 100 pounds. Interval and Ratio
level data are collectively referred to as “high level” (or quantitative) data.
6.4 CODING AND ENTERING DATA FOR
COMPUTERIZED STATISTICAL ANALYSIS
6.4.1 DATA MATRIX
In order to analyse quantitative data, once we have identified the kinds of variable we are
collecting, we can then set out the data in a matrix. This can be done in Excel or another
spreadsheet first, or put directly into a statistical package such as IBM’s SPSS Predictive
Analytics. To make the transition from, say, questionnaire to data matrix, answers will
need coding. For example, nominal variables will be text names and will need to be given
a unique number to allow entry into a statistical package. Non-responses will also need a
unique recognisable number (which doesn’t appear elsewhere in the data). Dichotomous
responses such as Male/Female will also need a number e.g. Male= 1; Female= 2.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
83
6.4.2 CODING
Most sources recommend that you keep a “code book” or list of exactly how the codes you
devise for your data relate to the questionnaire or other research element. This is vital for
two reasons. The first is that codes are often worked out on scraps of paper quite quickly; if
the paper is lost and you have a break between entering your data and coming to make sense
of it, it is possible you will have a hard time remembering exactly what the results mean.
The second is that it is important not to lose sight of the question when analysing the
results of quantitative data. Unusual patterns in the data must be scrutinised and going
back to exact coding and possible different interpretations of the question wording, which
may have caused the response, will be vital. So keep a retrievable, clear and accurate record
of coding as the link between respondent and data.
6.4.3 USING SPSS FOR WINDOWS
Coding is a way of enlisting the help of computer analysis techniques – whether these
involve using a spreadsheet, such as Microsoft Excel, or a package like the commonly used
SPSS (Statistical Package for Social Sciences) for Windows package which is specifically
designed to analyse quantitative data from social sciences research. SPSS for Windows is
the most commonly used tool to produce all statistical tests and analysis outlined in the
sections below. Using the package is very straightforward, provided you have access to it
on a computer. Start the program, which should put you into the IBM SPSS Predictive
Analytics Data Editor, which has two components: Data View and Variable View. Screen
tabs allow you to switch between these two views. Data View is the screen through which
you enter your data (like a spreadsheet). You must enter your data so that each column
represents a variable, and each row represents a case. For example, if you have information
on the age, salary and qualifications of 100 employees, you enter the variable data for each
employee along a row, with column headings of age, salary, qualifications.
It is probably obvious, but in data view you will not enter any text. To describe your
variables, you go to Variable View. Text variable names can be a maximum of 8 characters
with no spaces. This means it is helpful to make a rough plan of how you will enter data
into SPSS – in which order you will show the variables and what variable names you will
use. There is a field called “variable label” in which you can put more detailed text if needed.
It is also possible to enter labels for Values (all except interval values), so for example you
may have a variable labelled Gender, which has values labelled Male and Female, though
you have coded Male as 1 and Female as 2 in the Data view. Value and variable labels will
be used by SPSS in the Output charts.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
84
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
84
When you perform an analysis with IBM’s SPSS Predictive Analytics (by clicking Analyse
and entering any relevant information about what you want done) it is held as Output in
an Output viewer screen (which only appears after an analysis has been done).
While not particularly difficult, statistical formulas and calculations can be lengthy (depending
on the size of the data set) and thus often prone to simple calculation errors. The beauty
of using IBM SPSS Predictive Analytics, or any other statistical software program, is that
the researcher can focus on understanding the statistics and interpreting them rather than
on calculating the statistics. The researcher need only focus on careful data planning and
data entry. Once entered into the software, the program then creates an electronic version
of the original data and codebook.
6.4.4 WEIGHTING CASES
It is possible to weight cases when using stratified random sampling, when there is an
unequal response rate for different strata. This is simple to do and researchers do this from
time to time, but it does impose constraints on how statistical inferences can be drawn,
since cases in the lower response stratum are treated as if there were more of them than
PDF components for PHP developers
www.setasign.com
SETASIGN
This e-book
is made with
SetaPDF
http://s.bookboon.com/Setasign
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
85
there are (i.e. higher weighting in the dataset). Best avoided if possible unless you are really
confident in statistics.
6.5 CHOOSING APPROPRIATE WAYS TO PRESENT DATA
THROUGH CHARTS, TABLES AND DESCRIPTIVE STATISTICS
You may have a clear idea of what you are looking for in the data, but once the data is
entered into either a spreadsheet or an analysis package like SPSS, other possible ways of
analysing the data become apparent.
We usually begin by attempting to describe particular values, their range, their central
tendency, their dispersion around the mean. We can look at the data trends over time, and
look for proportions in the data. This is called univariate analysis because we are looking
usually at one variable at a time.
Once we have a clear picture of how the individual variables are behaving, we can start
looking for relationships between variables – bivariate analysis. A range of methods is shown
below for these two kinds of analysis.
6.5.1 FREQUENCY TABLES – UNIVARIATE
Tables show a list of categories (types of response) and the numbers of people responding to
each. Sometimes this is just shown as a number, sometimes a percentage of the total choosing
this response. When building a frequency table for interval level variables, categories will
usually be grouped (if not the table would probably be too long). Make sure your groups
of categories are exclusive e.g. for ages 21–30, 31–40 etc. not 20–30, 30–40 as this leads
to difficulties of coding for age 30.
6.5.2 BAR CHARTS, HISTOGRAMS AND PIE CHARTS – UNIVARIATE
These are generally used for nominal or ordinal variables, so bars will be separated along
the x axis. If using an interval variable, then a histogram would be used rather than a bar
chart. It looks very similar but the axis shows a continuous interval range and adjoining
“bars” are not separated. Note that pie charts should not show more than six segments –
more than this will be very difficult to read, so either use a bar chart, or group the data
before producing the pie chart.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
86
6.5.3 MEASURES OF CENTRAL TENDENCY – UNIVARIATE
This will be mean (average), median (midpoint value in ranked list) or mode (most frequently
occurring value) in a range of values. The measure is a single figure so is not representable in
a chart, however, a series of means, medians etc. can be charted or shown in a table. Mean
can only be calculated for interval level variables. Median can be calculated for interval or
ordinal variables. Mode can be calculated for any variable.
6.5.4 MEASURES OF DISPERSION – UNIVARIATE
This will be the range (difference between maximum and minimum value in a list of
interval variables), the inter-quartile range (data must be in rank order, then this will show
the difference within the middle 50% of values) or the standard deviation (data should be
normally distributed for this to be effective). The standard deviation is the average amount
of variation around the mean (calculated by taking the difference between each value and
the mean, totalling these differences and dividing the total by the number of values). A
higher standard deviation therefore means greater variation around the mean.
We might use a box plot to look at both central tendency and dispersion in a chart format
(SPSS can produce these from your data). A box plot shows where the median of the
data lies and how the data clusters around that median or middle value. 50% of the data
will lie in the “interquartile range” shown in a box plot as a rectangle with the median
line cutting vertically through it. In such an example, the median is off-centre to the left,
so we can see that this set of values is “skewed positively”, rather than showing a classic
normal distribution (see notes on sampling). The plot shows with an extended horizontal
line the extent of the lower quartile (i.e. the 25% of the data with the lowest values) and
the higher quartile – same but for the highest values. There are two more values from the
data set which sit outside the range of most of the data, called “outliers” – they are on the
right of the chart. This kind of chart is useful when in your research you want to give an
interquartile range (“half of the values are between x and y”) and to see whether a normal
distribution applies. This will also affect your later statistical analysis.
6.5.5 CHARTS, DIAGRAMS AND TABLES – THE DETAIL
It is probably quite obvious, but all diagrams etc. which are presented in a research report
will need to be checked for detail. When you are putting the last-minute touches to a report
before a deadline (at study and at work) it is easy to imagine that everyone will know what
this graph shows. This leads to a big problem if we leave it at that. You must check each
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
87
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
87
graph to ensure it has a clear title, the units of measurement involved are shown, any data
source is shown, the sample size is shown where relevant, the axes are labelled, the variables
read in a comparable way if more than one chart uses the same axes and variables e.g. left
to right or top to bottom and there is a key or legend which is readable (importing from
Excel often leads to very tiny illegible legends – they must be reformatted). The general rule
is that the title for a table is shown above the table itself, while a figure (i.e. chart or graph)
title is shown below the figure. It can also be helpful to introduce a chart in the text with
an idea for the reader of what it will show, then after the chart in the text, explain what
you think it did show. Of course, readers will want to make up their own minds, but it is
helpful to let them know what you think they should look for in the chart.
6.5.6 TRENDS OVER TIME
Usually shown in a line graph where time is on the horizontal axis. This is always a good
first step in analyzing data over time. Then if you wish to look at a trend over time for
a single variable, the most common method is the use of index numbers – such as the
FTSE100 index of share movements over time based in London (FTSE stands for Financial
EXPERIENCE THE POWER OF
FULL ENGAGEMENT…
RUN FASTER.
RUN LONGER..
RUN EASIER…
READ MORE & PRE-ORDER TODAY
WWW.GAITEYE.COM
Challenge the way we run
1349906_A6_4+0.indd 1 22-08-2014 12:56:57
http://s.bookboon.com/Gaiteye
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
88
Times Stock Exchange) or the Standard and Poor (S&P) 500 based in New York. The base
period is usually represented by the number 100 (or 1000 as in FTSE).
The FTSE100 is an index of the stock value of the 100 largest companies listed on the
London Stock Exchange, whereas the S&P 500 represents an index value of the 500
largest companies on the NYSE or NASDAQ. Rather than the mere “average price” of
the component stocks, these indices are “weighted” by the market value of each company,
providing a more meaningful metric.
In a simple index, each value is converted to an index number by dividing the data value
for the case by the data value for the base period and multiplying by 100. Why bother
converting each value to an index number? Generally because it makes comparison across
time or numbers much simpler – can be done at a glance.
Try to find an example (from the web or media) of a trend using index numbers.
Suppose we want to take the trend further and estimate where it will go after the actual
data we have to hand? Here we are into forecasting and we will be covering this in our
last but one chapter.
6.6 SELECTING APPROPRIATE STATISTICAL TOOLS
FOR THE RESEARCH VARIABLES
6.6.1 RELATIONSHIPS BETWEEN VARIABLES – BIVARIATE ANALYSIS
A relationship between variables means the variation in one variable coincides with variation
in another variable, it does not imply a causal or “cause and effect” relationship, i.e. it does
not necessarily follow that one will be an independent and one a dependent variable. Though
this can sometimes seem obvious – e.g. if the two variables include something like age or
gender which can influence the other variable but not be influenced by other variables.
(Presumably the amount you eat could be influenced by your age, but your age could not
be influenced by the amount you eat!)
6.6.2 CONTINGENCY TABLES OR CROSS-TABULATIONS – BIVARIATE
Set up as a frequency table including column percentages but showing both variables against
the chosen categories. If one variable is suspected of being the independent variable, this
is shown as a column variable not a row variable. Such tables are used to look for patterns
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
89
of association in the data. Frequency tables display low level data – just a frequency count
of how many times an event occurs.
6.7 FAMILIES OF STATISTICS
Research and statistics are aligned and complementary with each other in many respects.
Statistics are used to help describe and analyze the data collected through the research process.
Statistics can be useful in qualitative research as discussed in Chapter 9. Even qualitative
responses to survey items, or questions posed during exploratory research, can be thematically
analyzed and make use of core statistics such as frequency counts and percentages to assist
in the analysis and synthesis of such qualitative data.
Statistics can be classified into three “families”. Descriptive statistics, Correlational statistics
and Inferential statistics. Furthermore, inferential statistics can be classified as parametric or
non-parametric. These families of statistics align well with the Classifications of Research
(as presented in Chapter 3).
6.7.1 Descriptive statistics are often used as the basis of both Exploratory and Descriptive
research. Furthermore, descriptive research uses descriptive statistics – that is, they are useful
in answering research questions, focused around questions inquiring about who, what, when,
where, how much/how often.
6.7.2 Correlational statistics, and especially the correlation coefficient (described further in
this chapter) are used to provide a measure of association between two or more variables.
Correlation coefficients go beyond mere description of data, but begin to show the level of
association vis a vis the correlation coefficient. One must be careful to recognize that just
because there is a strong correlation between two (or more) items, this does not suggest or
imply a cause and effect relationship. The correlation coefficient merely indicates the degree
of strength of association between variables. That is, the degree to which one variable is
related to another variable. In a positive relationship, as the value of one variable increases,
so does the other. The reverse is true when a negative relationship exists. Nil correlation
indicates that there is a completely random relationship between data of two or more
variables. Correlation coefficients are measured quantitatively. A perfect positive relationship is
indicated by +1.00. A perfect negative relationship is indicated by -1.00, and 0.00 indicates
the lack of any type of association between variables at all.
6.7.3 Inferential statistics are used when conducting experimental research, and they are used
to help determine a cause and effect relationship (and can help answer the question “why”
things happen). Inferential statistics are used to test hypotheses, rather than answer research
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
90
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
90
questions. There are many types of inferential statistics, and the choice of a specific statistic
depends on the nature of the research design being used and the types of data collected.
Within Inferential statistics are two sub categories: Parametric and Non-parametric statistics
(Corder & Foreman 2000). The choice of using a parametric or non-parametric statistic
depends on a few things. Parametric statistics are best suited for conditions where the sample
size is large (generally over 30), the distribution of the data of the population is assumed to
be relatively normal, and the level of the data being analyzed is “high” (interval or ratio level
data). Non-parametric statistics are useful for the opposite reason. They are often referred to
as “small sample size” statistics (where the N is 30 or less). Also, non-parametric statistics
are useful when it can’t be assumed that the data distribution of the population is normal.
However, researchers rarely know the exact shape of the population data, so assumptions
need to be made. Finally, non-parametric statistics are very useful when the level of the
data being analyzed is “low” (nominal or ordinal). Non-parametric statistics are especially
useful in analyzing exploratory research data because the data here is often of a low level.
Many of the non-parametric statistical tests are included in IBM SPSS Predictive Analysis
software. Refer to the Corder & Foreman text included in the reference list for an excellent
source on the use and interpretation of non-parametric statistics.
How to retain your
top staff
F I N D O U T N OW FO R FR E E
Get your free trial
Because happy staff get more done
What your staff really want?
The top issues troubling them?
How to make staff assessments
work for you & them, painlessly?
DO YOU WANT TO KNOW:
http://s.bookboon.com/performancereviewpro
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
91
The table below shows some of the popular parametric significance tests, along with their
non-parametric equivalent.
Parametric test Non-parametric equivalent
One sample t-test The sign test; The Wilcoxon one sample test
Two sample t-test The Mann-Whitney U; Student’s t-test
One way ANOVA Friedman’s test
Pearson Product Moment
Correlation Coefficient (r)
The Spearman rho
6.8 MEASURES OF CORRELATION – THE
CORRELATION COEFFICIENT
A correlation coefficient is a statistic which is designed to quantify the degree of correlation
between two variables. Generally speaking, a correlation coefficient will be expressed as a
number which will either be positive or negative. The “sign” (positive or negative) indicates
the direction of the relationship, i.e. +1 is a perfect positive relationship (as one variable
increases, the other increases, and -1 is a perfect negative relationship (as one variable
increases the other decreases). The absolute value of the number indicates the strength of
the relationship. The value will be between 0 (indicating no relationship) and 1 (indicating
a perfect relationship). Calculating correlation coefficients is easy, BUT they are subject to
error due to the numerous (albeit simple) calculations which have to be made. Using a
statistics program such as SPSS makes the calculations easy as all the researcher has to do is
enter the raw data. The program does the error proof work (assuming error proof data entry)!
There are four different types of correlation coefficients which are described below. The choice
of which to use depends upon the level of data being used for the variables being correlated.
6.8.1 PEARSON’S R (ALSO CALLED THE PEARSON’S PRODUCT
MOMENT CORRELATION COEFFICIENT (PMCC)) – BIVARIATE
The Pearson r looks for a relationship between two interval level variables. Before calculating
a Pearson r, it is worth constructing a scatter diagram for the two variables, as it should only
be used when there is a broadly linear relationship, it will not hold for a curve relationship.
The scatter plot provides a visual image of the correlation.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
92
6.8.2 SPEARMAN’S RHO (ρ) – BIVARIATE
The Spearman rho is used when at least one of the two variables is ordinal level data,
and the other is ordinal or interval level data. This calculation produces the same kind of
outcome as Pearson’s r, i.e. a positive or negative relationship between 0 (no relationship)
and 1 (perfect relationship).
6.8.3 ETA – BIVARIATE
Eta is used to explore relationships between an interval level variable and a nominal level
variable. Eta can only show strength of relationship, not direction. It does not assume a
linear relationship.
6.8.4 PHI COEFFICIENT (Φ) AND CRAMÉR’S V – BIVARIATE
Phi is used for exploring a relationship between two dichotomous variables, Cramér’s V does
the same for two nominal level variables. Phi outcomes are like Pearson’s r and Spearman’s
rho and can vary between 0 and + or -1. Similar to the Eta, Cramér’s V can only show
strength of relationship, not direction (the coefficient is always positive).
6.9 REGRESSION ANALYSIS
Regression analysis is a coefficient of determination (it can also be called a regression
coefficient). It can be calculated by squaring the value of Pearson’s r and multiplying it
by 100. This produces a percentage, which describes the proportion of variation in one
dependent variable accounted for by the other independent variable. So if we explored the
relationship between age and weight in a sample, producing a Pearson’s r value of -0.35,
then the coefficient of determination would be 12.25%, which suggests that in our sample
12.25% of the variation in weight was accounted for by variation in age. A similar analysis,
where more than one independent variable is involved, is called multiple regression analysis.
(This will be discussed further in Chapter 11.)
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
93
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
93
6.10 STATISTICAL SIGNIFICANCE
A way of testing the level of confidence we can have that a probability sampling technique
has generated results, which can apply to the full population. Such a test can also estimate
the chances of no relationship in fact existing between two variables, when bivariate analysis
suggests that there is. We often use the word “significant” to mean the same as important
when we are writing text. Your understanding of the phrase “statistical significance” should
prevent you from now on from using “significant” in academic work, unless you are relating
this to a statistical test.
To calculate statistical significance, we set up a “null hypothesis” i.e. that two variables in
the sample are not related. Then decide the level of statistical significance we find acceptable,
i.e. the level of risk that we would reject the null hypothesis (i.e. say the variables are
related) when in fact they were not related. It is usual to say that the maximum level of
0.05 is acceptable (i.e. p<0.05). This suggests that in no more than 5 cases out of 100,
will we be wrong (i.e. suggest a relationship which is not there) – the same as saying we
have 95% certainty that the relationship is correct. We can choose a more stringent level
of certainty (e.g. p<0.01 where there is only a 1 in a 100 chance of our relationship not
existing when we say it does). This would, however, increase the risk of a “Type 2 error”,
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
http://www.deloitte.ca/careers
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
94
which means confirming the null hypothesis (that there is no relationship) when in fact
there is a relationship.
We should bear in mind that the likelihood of a statistically significant result will increase
with sample size – for the obvious reason that the bigger the sample in relation to the
population, the less likely that any analysis on the sample will differ from the population
by chance. So if we think there is likely to be low statistical significance, we should increase
sample size if possible, to make the analysis more sensitive to statistical significance. Very
small samples, below 30, are more likely to show an unacceptable p level, i.e. above 0.05
probability that the difference is caused by chance.
We use a chi-square test (x2) to produce our level of statistical significance (or probability
level). This test looks at each cell in a contingency table and calculates the expected value if
there was no relationship but the value was a product of chance, works out the difference
between each expected value and the given value and sums the differences. This produces a
single chi square value for the table, which is not important in itself, but is produced with
a statistical significance level (ρ). This is the number we are looking for, to check against
our desired level of certainty. The chi-squared test can only be used with nominal level
data and provides a useful method for determining statistical significance when dealing
with such low-level data.
As well as applying chi-square tests to contingency tables, tests of statistical significance
should be applied to all bivariate analysis outcomes (coefficients) such as Spearman’s rho
and Pearson’s r. This helps us to be sure that the correlation we expect from the sample,
really does exist in the population.
6.10.1 TESTING WHETHER GROUPS ARE DIFFERENT – MULTIVARIATE ANALYSIS
If we want to test whether the distribution of a variable in a sample is similar to or different
from the distribution of a population or census which is already known, then we can use a
Kolmogorov-Smirnov test (only if data is ranked). The test produces a D statistic, which is
used to calculate whether the sample distribution differs from the full population distribution
by chance only.
Where we have a quantifiable variable which can be split into two independent groups of
values using a descriptive variable, we can test the probability of the groups being significantly
different using an independent groups t-test. The lower the t statistic, the more likelihood
of any difference in the two different groups being caused by chance.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
95
Similarly, a paired t-test can be used to measure similar pairs of variables, e.g. a machine’s
speed of operation before and after maintenance. The paired t-test is used when the measures
being compared are from the same group (e.g. before/after).
Differences between three or more groups can also be tested to see if they are likely to be
occurring by chance or if there is really a “statistically significant difference” – this is done
using one-way analysis of variance, ANOVA, and produces an F statistic plus a significance
probability level. A high F statistic and a significance p level of below 0.05 should offer a
“statistically significant” result, i.e., not one occurring by chance. An ANOVA example might
involve members of three or more different groups of staff producing values for “degree of
learning” after a training course. The ANOVA test can establish whether different results
in degrees of learning after training which seem to be shown by the different staff groups
could occur by chance, or whether there is a “statistically significant” difference between
them. There are some data requirements for ANOVA, but broadly this can be used provided
there are at least 30 values in each group and each value is independent of others. Whereas
t-tests compare either two different groups, or the same group twice, ANOVA allows the
researcher to compare 2 or more groups at the same time. Both t-tests and ANOVA require
high level (interval or ratio) data.
This chapter has been very factual and is not easy to take in, unless you are already familiar
with statistical analysis and find it easy to follow. It is intended just to give some revision
pointers based on earlier reading or teaching you may have experienced.
You might like to consider the following question to think through this area. If I asked
the question: “Please rank order the following benefits of a colour laser printer: speed,
professional output, capacity for more than one ream of paper, faster speed on black and
white print, capability to back copies, other”. Then what kind of values will be produced
if 100 people respond to this question? (a) how would we code each response, including
the “other” response? (b) and what kind of technique could we use to analyse the response
data, and here we could assume that we know whether the respondent actually uses laser
printers or not.
6.11 QUESTIONS FOR SELF REVIEW
1. Why is it important to think through the data likely to be produced from your
research at an early stage?
2. Why do you need to know the difference between nominal, ordinal, interval and
ratio level variables?
3. What is bivariate analysis?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
96
QUANTITATIVE RESEARCH METHODS:
COLLECTING AND ANALYSING DATA
96
4. What is the minimum number of cases you need to make a sample useful for
statistical analysis?
5. What is the level of probability (p) needed to state in your research results you
have found a “statistically significant” difference?
6. What is the purpose of using index numbers and an example from the web or media?
6.12 REFERENCES
Corder, G.W. & Foreman, D.I. 2014, Nonparametric statistics: a step by step approach, 2nd
edn. Wiley and Sons, Hoboken NJ.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
Velleman, P.F. & Wilkinson, L. 1993, “Nominal, Ordinal, Interval and Ratio typologies are
misleading”. The American Statistician, vol. 47, no. 1, pp. 65–72.
TMP PRODUCTION NY026057B
PSTANKIE
gl/rv/rv/baf
ACCCTR0005
Bookboon Ad Creative
4
6 x 4
12/13/2013
Bring your talent and passion to a
global organization at the forefront of
business, technology and innovation.
Discover how great you can be.
Visit accenture.com/bookboon
©
2013 A
ccenture.
A
ll rights reserved.
http://s.bookboon.com/accentureCZintl
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
97
7 QUESTIONNAIRE DESIGN
AND TESTING
7.1 CHAPTER OVERVIEW
7.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. appreciate and overcome the difficulties associated with questionnaire design
2. recognize the applications when survey research may be useful
3. choose from a range of survey question formats
4. design, pilot and administer questionnaires as part of a research strategy
7.2 APPRECIATE AND OVERCOME THE DIFFICULTIES
ASSOCIATED WITH QUESTIONNAIRE DESIGN
Questionnaires can be a powerful source of information. But effective questionnaire design
is of utmost importance if you want to get the information you desire. To begin with, you
must decide what the purpose (objective) of the questionnaire is, and ultimately what are
the research questions you want to answer as a result of using a questionnaire. From there
you begin to design questions which accomplish your objective. You also need to consider
the length of the questionnaire. Respondents who begin to suffer from “survey fatigue” will
often either fail to finish the questionnaire or just mindlessly rush through the questions to
finish it quickly. As a researcher, you want to eliminate both of those possibilities.
What kind of difficulties and problems arise with questionnaires? Surely it is quite straightforward
to write them? It is said that a person’s wisdom can be judged by the questions they can
compose rather than the answers they know! In fact, designing questionnaires is particularly
difficult. What do we need to think about when designing effective questionnaires?
1. The format and design of the questionnaire – not too off-putting, not too long,
not too difficult to read, easy to know what you have to do to complete it
2. How much general information do you need to have about the respondent? If you
need biographical data such as age and gender etc., why is that? What extra value
will it add to your research question? Should you start with easy questions like
gender, or end with them?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
98
3. What proportion of open and closed questions should be in the questionnaire?
Closed questions start with a verb (Do you come here often?) and invite a simple
yes/no answer. Or they may be “forced-choice” i.e. only a limited number of
alternative answers are available to choose from. They are easy to code but limiting
in detail. Open questions give much richer information but are widely variable
across responses and therefore harder to code and analyse.
4. What kind of questions can we ask? Straight questions with a clear answer?
Questions about which people must reflect? Tickbox questions or written answer
questions? Numerical scale questions? These are rating scale questions which allow
the respondent to mark a numerical scale in response to a question; for example
“How important is it to have a clear organizational policy on harassment at work?”
Answers may range from 1 (not important at all) to 4 (very important). Likert
scales often have middle points which allow a neutral response. How do we lay
out the questions? For example, if they are scale responses – do we lay them out
horizontally or vertically?
5. How much space on the page do you give someone to write an answer to an
open question?
6. Should you include check questions, for example asking the same thing two different
ways to ensure you are getting consistent answers?
7. How much information do you give the respondent about why you are asking the
questions? Technical research detail? Just enough to know who you are and how
data will be used?
8. How do they get it back to you? If email – what does that do to anonymity? Should
you include stamped addressed envelopes, drop boxes?
9. Should you communicate with potential respondents before the survey itself? And
after delivery to encourage completion? How many times could you prompt for
a reply?
10. Should you use post, fax, email or online surveys?
11. What happens if the response rate is too small to be useful? 30% is considered a
minimum but this will depend on the volume of response.
12. What if some of the questions are misunderstood? How can I prevent this happening?
13. Should I use incentives for survey return? How does that affect results?
14. How many surveys should be used for a pilot survey to test the questions? Can we
use pilot responses in the results?
15. Where do I keep returned surveys?
16. How much do I have to spend on printing and/or designing and/or posting out
and chasing questionnaires? Can they go out with other mailings to save cost, or
will this lower the response rate?
These are all practical questions which are best answered in the context of your specific
research questions in relation to the actual population who will receive the questionnaire.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
9999
You probably already have views about the answers, but if not, these are all detailed practical
issues which you will have to decide before sending out your questionnaire.
You may like to search online for an article by Vidal Diaz de Rada on Questionnaire
Design (Diaz de Rada, V., 2005). This author discusses some of the formatting details for
questionnaires such as size, colour and cover page. All these factors play an important role
in your response rate by those asked to complete the questionnaire.
7.3 CHOOSING FROM A RANGE OF QUESTION FORMATS
Textbooks generally list the following as types of closed questions used in questionnaires:
1. List – select any answer of those provided (offer an “other” category if necessary)
2. Category – select one answer (also called multiple choice)
3. Ranking – put answers in order
4. Rating – score or give a value to answers
5. Quantity – respond with amount
6. Grid – complete a matrix to provide more than one answer
Unlock your potential
eLibrary solutions from bookboon is the key
Interested in how we can help you?
email ban@bookboon.com
eLibrary
http://bookboonglobal.com/en/qualities/corporate-education-elibrary-solution/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
100
We could add to the list above:
1. Personal factual attribute question – e.g., age, employment status, qualifications
2. Likert rating scale – strength of response (e.g., strongly agree through strongly
disagree) indicated against numeric scale of 1 to 5 with a neutral midpoint
3. Rating scale – strength of response indicated against verbal scale
4. Rating scale – strength of response indicated against bipolar or self-anchoring
numeric scale (opposite statements at either end of numeric scale)
5. Semantic differential scales – opposite adjectives at each end of numeric scale
6. Frequency scale – verbal scale or numeric between always and never
7. Fill in the blank (open response)
8. Dichotomous choice (e.g. Yes/No, True/False, Agree/Disagree)
9. Match pairs
You may be able to think of more? Remember closed questions are designed to check facts
or perceptions, confirming information and producing answers which either qualify the
respondent in some way, or give comparable data across your sample.
Open question formats include:
1. Open list – number of answers required, type of answer free
2. Open essay – often used as a final option to let respondent comment
3. Personal question about opinion – free answer
4. Personal question about behaviour – free answer
5. Vignette or scenario – question is set in an example context, answer usually open
See if you can devise a business research question for each of the formats above. It is only
by doing this, that we get an idea of any problems in wording. Wording of the questions
on surveys is one of the significant challenges in designing effective questionnaires. Preparing
meaningful and carefully thought out response sets to the questions is also an important
part of survey design.
In addition to having an objective for the questionnaire when you begin, you should also
start by listing what information you want to collect, and then design questions which elicit
that information; you also need to consider the type of information you want to collect
and the level of data your questions will provide. For example, you may want to know
the age (or income, etc.) of your respondent. You can ask that as an open-ended question
and request they fill in the blank with their actual age (or income), or you can design a
closed ended question with 5 different ranges of ages (or incomes) which the respondent
can choose from. While the question is the same, the answers yield very different levels of
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
101
data which will impact the kind of analysis you can perform on the data. Also, when asking
more personal questions, you may get a more honest answer if ranges are used. But range
data will be at the ordinal level of data, while asking for actual numbers provides ratio level
data. You have to decide what level of data you need. Don’t ask for actual numbers if all
you are going to do with it is categorize it later on anyway. When designing ranges, think
about the highest value likely to be in the population. Typically, 5–7 categories is standard,
so divide the highest number by 5–7 to establish the range which the categories should
include. Unless you absolutely know the highest value, leave the last range open (if asking
age, the last category should say “xx or older” rather than xx–yy). If too many respondents
fill in just one category you probably did not consider this range issue carefully enough! Be
sure to collect all the information you need, but do not collect information you will not
need or use. This is a common problem novice survey designers often face.
There is another kind of question in many questionnaires and that is a filter question
(sometimes a whole filter section). Filter questions are used when some parts of the
questionnaire are not relevant to all respondents. We may use a filter question such as “If
your answer is no, please move to question x”. Or we can clearly label sections “if x applies
to you, please omit this section”. The important thing will be clarity and avoiding filters if
at all possible, since they are often a cause of error and non-response. The exception will
be if we use filters in an online survey, here they can be automatic for the respondent and
remove all doubt as to which question to answer.
When designing questionnaires, it is worth referring to a book on this subject by A.N.
Oppenheim. The second (2000) edition of this classic work (originally written in 1992)
includes far more information on questionnaires than most of us will need, but discusses
specific issues very clearly. Additionally, Saunders (et. al., 2016) dedicate an entire chapter
(Chapter 26) to the topic of internet research methods (e-research).
7.4 HOW TO DESIGN, PILOT AND ADMINISTER QUESTIONNAIRES
Remember that questionnaires are not just the self-completion kind sent through the post
or email or found on the web. Question sets are also created for structured interviews and
semi-structured interviews either face-to-face, over the telephone or online through video-
conferening (we will look at these in detail in the chapter on interviewing). Well-designed
questions are the skeleton of any good research study. Even when we don’t ask them directly
of respondents, we often have to prepare them to collect data – for example when preparing
to conduct participant observation, it will help to have clear questions in mind and perhaps
some kind of pro-forma for us to complete during the experience.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
102102
Here are ten tips to avoid survey question pitfalls:
1. Keep checking back to your research focus and objectives – what do I want to
find out?
2. For every question, jot down the maximum number of ways in which you think it
can be answered – this immediately identifies problems with wording and scaling.
This is even better done by someone else for you as a double check, and serves as
a proof reading and a pilot testing of the survey
3. Closed questions are easier to use for data but including only closed questions will
provide you with limited data. Unless you believe you can anticipate everything a
respondent will say, use some open questions as well.
4. Check for vague terms in the question e.g. often, usually, sometimes, this year,
most, few – people will have different meanings for these terms so their answers
will not necessarily coincide with what you intend.
5. If you are asking a “why” question, think about the frame of reference of your
respondent – are they likely ever to have thought about this “why”? If not, can you
make it easier to answer? Perhaps offer alternative answers with an “other” category?
6. Check for “leading” questions which lead a respondent to agree or disagree with
something (“Push-polling”).
.
http://s.bookboon.com/AlcatelLucent
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
103
7. Check for “double-barreled” questions. Look for the word “and” in your questions!
Allow respondents to answer one thing at a time.
8. Avoid technical words and acronyms (e.g. management jargon such as incremental,
optimum, marginal, strategic, motivation, ROI [return on investment] etc.) unless
you are sure the respondents are familiar with them.
9. If you are using yes/no or true/false questions, make sure that no-one is likely to
want to give an answer which is not available (e.g. sometimes). Make sure the
question is truly a dichotomous choice.
10. Ask yourself if some of your questions are actually the kind of things only staff
with some considerable experience of the company will know. If so, do you have an
attribute question to qualify whether they have that experience? Use filter questions
(in an online survey) so that respondents only see those relevant to them.
More generally, here are some pointers to consider about the layout and testing of your
survey questionnaire. You should have tested your questionnaire against each of these points
before you send it to important respondents.
• Have you considered whether to use five intervals in a scale or four? Five intervals
will encourage a central tendency, i.e. respondents find it easier to give a mid-point
reply than an extreme reply. So given Very poor, Poor, Average, Good, Excellent,
there are likely to be a lot of Average answers. Four intervals are better as they force
respondents to commit themselves on the positive or negative side, but should also
include a Not Sure or Not Applicable, usually located at the end of the scale for
ease of coding.
• For similar reasons, if you are using a large number of rating questions, switch some
around (i.e. they should not all be expressed just positively or just negatively) so
that respondents have to think, rather than running quickly down a ticklist. They
will need switching back before coding.
• If you are asking for company information e.g. sales or customer profiles etc. – are
you sure all respondents will be able to answer this? Could you just find this out
from a senior manager or contact, rather than asking everyone in your sample?
• Get someone to look at your questionnaire. Could you make the layout and format
easier to read?.
• Check the questionnaire’s spelling and grammar – both are vital to ensure transmission
of accurate meaning and maintain researcher credibility.
• Have you clearly shown how to respond to each question (e.g. tick mark or circle
(difficult on email forms as this requires symbol, a X is easier, this is simplified if
using forms software as in Word or Google Docs); and how will the respondent
return the form?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
104
• Check whether you have authority to send this questionnaire (if it is to an
organization’s staff etc.).
• Pilot-test the questionnaire, and pre-pilot (an earlier draft stage). You will only get
one chance to get these questions answered by this sample!
• Maintain a record of all changes you have made to your questionnaire and why.
This will be helpful when you are writing up your research method and should
remind you of the learning you have achieved.
There are a number of articles on web-based questionnaires in recent issues of the International
Journal of Social Research Methodology (Fox, J., Murray, C. et al. 2003; Heerwegh, D.,
Vanhove, T. et al. 2005) and the Electronic Journal of Business Research Methods (McCalla,
R.A. 2002). Look through them and consider what you believe to be the most important
differences between web-based and postal questionnaires. We know that the web offers us
speed and often ease and convenience of use. We also know that some companies offer
simple online questionnaire building sites which may be free or involve just a small charge
(e.g. www.surveymonkey.com) and this can help give a quick questionnaire a professional
look and offer automatic response summaries. If you are a member of a higher education
institution or organization which has a professional online survey tool, such as Qualtrics
(www.qualtrics.com), then be sure to use this rather than a free version, as your formatting
and potential for data analysis will be much more professional.
Survey Monkey and Qualtrics provide a means to both design and deploy your surveys
through email, the web, and other online portals including mobile devices. Using such
online questionnaire and survey design resources can help make the survey research process
efficient. A note of caution is that resources such as these place the ability to conduct a
questionnaire or survey into the hands of virtually anyone. While these programs make
creating questions easy, they do not make the survey design process foolproof. Many a
bad question, or even entire survey has been deployed by “amateur” researchers using high
power research tools. And the downside of using these online tools is that they do have
some limitations regarding the types and layout of your questionnaire.
The choice of a web survey or physical print survey (post or e-mailed) will always depend on
the population you are targeting and the context of your research. Clearly some populations
will not have web access and/or may not like using the web interface. Others, for example
people seeking a job online, would be ideal candidates for web survey. There is some evidence
that if a group to be surveyed is offered a choice between online or physical questionnaires,
they will respond better to physical copies if these are administered together in one place
and time, but it’s better to use online questionnaires if not co-located.
http://www.surveymonkey.com
http://www.qualtrics.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
105105
7.5 QUESTIONS FOR SELF REVIEW
1. Why do most questionnaires for self-completion have a lot of closed questions?
2. What incentive would it take for you to answer a 20 item questionnaire, a 50 item
questionnaire and a 100 item questionnaire? Think about your response in relation
to how many items you might include.
3. What is a Likert scale? What different kinds of scale questions are there?
4. How would you design the category ranges of a question if you were interested in
knowing the ages of parents of nursery school children versus the age of patients
in a nursing home?
5. Why should a questionnaire always have a covering letter/email?
6. Should you use an odd or even number of scales for a survey question? Why?
AXA Global
Graduate Program
Find out more and apply
http://s.bookboon.com/AXA
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Questionnaire design and testing
106
7.6 REFERENCES
Couper, M.P. 2008, Designing effective web surveys. Cambridge University Press, New York.
Diaz de Rada, V. 2005, “Influence of questionnaire design on response to mail surveys.”
International Journal of Social Research Methodology, vol. 8, no. 1, pp. 61–78.
Fink, A. 2016, How to conduct surveys: a step by step guide, 6th edn. Sage Publications, London.
Fox, J. Murray, C. & Warm, A., 2003, “Conducting research using web-based questionnaires:
practical, methodological and ethical considerations”. International Journal of Social Research
Methodology, vol. 6, no. 2, pp. 167–180.
Heerwegh, D. Vanhove, T. Matthijs, K. & Loosveldt, G., 2005, “The effect of personalisation
on response rates and data quality in web surveys”, International Journal of Social Research
Methodology, vol. 8, no. 2, pp. 85–99.
McCalla, R.A. 2002, “Getting results from online surveys: reflections on a personal journey”,
Electronic Journal of Business Research Methods, vol. 1, no. 1, pp. 55–62.
Oppenheim, A.N., 2000, Questionnaire design, interviewing and attitude measurement 2nd
edn. Bloomsbury Academics, London.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
107
8 USING SECONDARY DATA
8.1 CHAPTER OVERVIEW
8.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. Identify the value of secondary data to business research
2. Understand what to look for as secondary data and where to find it
3. Understand the disadvantages of using secondary data in business research and
how to overcome them
4. Distinguish between proprietary and public access sources of secondary data.
8.2 THE VALUE OF SECONDARY DATA TO BUSINESS RESEARCH
8.2.1 BROAD DATA GROUPINGS AVAILABLE
Survey secondary data will usually have been analysed for its original purpose and could be
a national periodic compulsory census, a regular e.g. annual survey or a one-off survey. You
should be aware that this is unlikely to be raw data, i.e. some filtering and data decisions
will have had to be made (e.g. coding of non-responses, grouping of data etc.).
8.2.2 CONTEXTUAL BACKGROUND
Much business research will require an awareness of industry, national or sector context
(for example if you are conducting primary research in a healthcare organization, it will be
useful to set the context for this by comparing national or international healthcare statistics,
or you may be reviewing your local area’s labour force and want to see how this relates to
your country’s or other country’s labour force statistics).
If you would like to try out a search for some international labour force statistics on the
web, then conduct the following experiment. Try Google first and note down what you
can find in 10 minutes. Then try Eurostat (the European Union statistics website) and
note down how long it takes to find labour force statistics for member countries and any
other issues which arise. Then try www.esds.ac.uk which should give you access to OECD
http://www.esds.ac.uk
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
108108
Labour Force statistics. You may also want to try the US based Conference Board (www.
conference-board.org/data) and select “international comparisons”. Note down how long this
takes and what kind of information seems to be available. This experiment should give you
a clear idea of how complex some secondary data can be and the types of data, particularly
statistical data, which is easily available to the researcher.
If you are likely to use such sources, think about:
1. How the different sources compared
2. How long did it take to use each source. Would it be easier and faster next time?
3. How do the labour force statistics compare? Were you finding similar statistics?
8.2.3 QUICK AND INEXPENSIVE DATA
Secondary data is often cost-free and it can be accessed over the web or from your local/
university library. Secondary data can be classified as proprietary or public access. Proprietary
data is that which has restricted access, whether it is protected from use by others without
permission, or whether it is accessed through a subscription to a service providing the
��������
�����
������
��������
���������
�����������
��������
�����
��
���������������������������
������
������
������������������������������ ����� �����
�
����� ������!���"�
������������
�
�
�����#$%����&'())%�*+������
�
���,���������
�-
.�
��������������������
������������
��
��������
��
����
�� ��������
��
����
���������
���
����������
����������
�����
��
���������
The Wake
the only emission we want to leave behind
http://www.conference-board.org/data
http://www.conference-board.org/data
http://www.mandieselturbo.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
109
data. When using proprietary secondary data it is important to be sure you have the right
permission to use the data. As students enrolled at a university, often you have access to
such proprietary data through your university library. For example, Lexis-Nexis, Hoovers
and Mergent-online are private sources to extraordinary amounts of business intelligence and
secondary data. However, for example if you just go to “mergent.com” you will be taken
to its website, but in order to access its databases, you will have to have a subscription and
login to the service. If your university (or public) library holds a subscription to Mergent,
you would access it through the library’s gateway to these databases.
Often, students are not aware of the sources of private domain business intelligence to
which they have subscription access through their place of study. Take time to visit your
university library (either online or in person) to explore the possibilities. Additionally, many
businesses often have subscriptions to such databases for its employees to access and use.
As knowledge-based professionals, it’s important for you to know where to find and how to
use such business intelligence and data to assist you in performing your job, and in helping
make your employer successful.
8.2.4 LONGITUDINAL AND CROSS-CULTURAL DATA SOURCE
Much national and international data is collected on a periodic basis over time, so allows
longitudinal research studies – not normally possible through primary research in view
of cost and time constraints. Similarly, cross-cultural studies can use large survey data,
when conducting this as primary research is particularly complex. For more information
on cross-national studies and some of the problems which can arise see Lynn’s article on
this in International Journal of Social Research Methodology (Lynn, P. 2003), available
fulltext online.
8.2.5 META-ANALYSIS MADE POSSIBLE
Meta-analysis (conducting research on other people’s research, therefore at one remove from
it) can also produce surprising fresh insights – partly on the basis that at this perspective
it may be easier to see “big picture” patterns. Meta-analysis also involves the process of
trying to integrate the quantitative and qualitative results of numerous research studies
employing various methodologies and designs; all of which are on a similar topic or thesis.
Meta-analysis (opposed to a literature review) seeks to draw more comprehensive conclusions
and often involves quantitative methods of data analysis and synthesis of these multiple,
but common-topic research studies. Meta-analysis operates under the Gestalt principle of
“the whole is greater than the sum of its parts”.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
110
8.3 WHAT TO LOOK FOR AS SECONDARY
DATA AND WHERE TO FIND IT
8.3.1 WHAT IS SECONDARY DATA?
Secondary data is data, which the researcher did not collect for themselves directly from
respondents or subjects. This means that secondary data was not collected with the researcher’s
purpose and objectives in mind. It may have been collected
• by other researchers, perhaps in the process of academic studies (could be available
in journal articles, or published doctoral theses or conference proceedings) or
• in the process of normal operations (e.g. an organization’s “grey” material –
information it publishes internally such as sales figures, information about product
launches, company minutes etc., or an individual’s personal diary or learning log)
• by institutions, whose job is to collect data (e.g. government or regional offices
of statistics and information, international bodies whose purpose is information
collection e.g. OECD or academic, media and professional bodies set up for the
purpose of collecting information and data directly and from these government or
international bodies).
For many business research studies, especially qualitative ones, it will be difficult to find
exactly the kind of data needed, since it is unlikely you would be doing the research were
it not for the fact that it hadn’t been done before! So most studies will need to design
collection methods for primary data. However, there is a vast amount of secondary data
out there, much of it surprisingly accessible over the web, which may save us considerable
time or give us a useful benchmark or context in which to set up our research design or a
way of triangulating our results.
Where your research relates to a national or international level of operation, it is likely that
national and international statistics will form part or all of your study, since these studies
take time and money to achieve. Much of this kind of data. e.g. census data will be available
free over the web or free from a variety of Government offices. Secondary data may be
documentary, survey or multiple source, i.e. a mix of documents and surveys.
Access to secondary data implies two things – first of all you know how to find it and
second you have permission to use it.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
111111
8.3.2 KEY BUSINESS INFORMATION SOURCES:
These sources and links are subject to regular change, be prepared to try several to find
useful data for your research study.
EU: Eurostat – source of data on the European Union
http://ec.europa.eu/eurostat/web/main
UK: UK Data Service. This is funded in the UK by the ESRC with contributions from the
University of Essex, the University of Manchester and JISC. The main website is https://
www.ukdataservice.ac.uk, but there is a special introduction to the service for students at
https://www.ukdataservice.ac.uk/use-data/student-resources
UK: www.statistics.gov.uk (census and other data surveys)
UK: CIPD Recruitment Survey/Training & Development Survey etc. www.cipd.co.uk
UK: Workplace Employee Relations Survey (WERS) www.data-archive.ac.uk This is a digital
data archive for Social Sciences and Humanities
UK: FT info http://news.ft.com (company information)
NA & other countries: Hoover’s Online www.hoovers.com (company information)
NA: www.Fortune.com (US & Global business data sources); www.mergent.com (Mergent
Online) NA: dol.gov; (US Dept. of Labor; incl. international statistics); www.bea.gov/
(Bureau of Economic Analysis); www.Census.gov/ (US Population & Economic Census data)
http://ec.europa.eu/eurostat/web/main
https://www.ukdataservice.ac.uk
https://www.ukdataservice.ac.uk
https://www.ukdataservice.ac.uk/use-data/student-resources
http://www.statistics.gov.uk
http://www.cipd.co.uk
http://www.data-archive.ac.uk
http://news.ft.com
http://www.hoovers.com
http://www.Fortune.com
http://www.mergent.com
www.bea.gov/
http://www.Census.gov/
http://s.bookboon.com/elearningforkids
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
112
UN: www.un.org
World-wide: OECD www.oecd.org; www.Lexis-Nexis.com
The US Census Bureau also contains a wealth of population and economic census data.
The population census is performed every 10 years, and the Economic Census is conducted
every 5 years.
www.census.gov (select “Population” or Economy” underneath “Topics”)
You can probably add sources to this list which you may have found in earlier studies.
The list is growing all the time and the web enables us to find new data sources around
increasingly specific topic areas.
8.4 THE DISADVANTAGES OF USING SECONDARY
DATA IN BUSINESS RESEARCH
8.4.1 DIFFERENCE OF PURPOSE
Because the original researchers had a different purpose and constraints from your current
project, there may be inconsistencies or elements of the research which are not compatible
with your own. This could include currency, terminology, samples, market changes, boundary
changes, new discoveries or technology since the research was carried out etc. Additionally,
secondary data may not be in the form in which you would like to use it. As a result,
secondary data may need to be transformed and “recoded” from its original format. For
example, you may find specific secondary data on the “subjects” of your study; however,
you need to know “annual income” but it may be found as “monthly income”, in which
case you will need to multiply by 12 in order to use it in the form you require. Or in
doing international business research, you may have company financial data reported in the
currency of each country but will need to convert it into a common currency appropriate
for the audience to which your research is directed.
8.4.2 AGGREGATION AND PRESENTATION OF DATA
Other researchers, working for other research purposes, will often aggregate data in a way
which is not useful for your own research, for example showing regional rather than city
data, or street rather than household data, or not disaggregating by gender. The presentation
of data will depend on the purpose of the original research too – especially if the research is
done for a media purpose, where headline stories sell media. There may be some apparent
distortion in the data because of this.
http://www.un.org
http://www.oecd.org
http://www.Lexis-Nexis.com
http://www.census.gov
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
113
8.4.3 DATA QUALITY
From this book, you will be getting an idea of the attention to detail and careful planning
and thought, which goes into good quality research. This is one thing which is difficult
to check when using secondary data. Instead, the best we can usually do is to ensure the
credibility and professionalism of the source institution, rather than the data. Be particularly
careful when using secondary data from internet sources, where organizations are not known
to you, as anyone can put up false data on the web without challenge. This also leads us to
be particularly cautious about detailing and documenting our reference sources.
A further step in assessing data quality will be to critically evaluate the research methods used
to collect the secondary data. It is often reasonable to contact the data source to establish
their methods, if you are considering using their publicly available data. Government sources
usually publish detailed technical background alongside the data to enable you to interpret
the data appropriately.
8.4.4 MEASUREMENT VALIDITY
Think back to our discussions on epistemology – we cannot expect secondary data to be
some kind of “truth”. The data will reflect the purposes, and pre-conceptions, of the original
researchers. It would also be useful to think back here to your reading on the taxonomy of
facts (Bannister, F. 2005).
8.4.5 DATA COVERAGE AND CLEANSING
Does the secondary data cover the exact population in which you are interested? Are there
any unwanted exclusions or inclusions, which may affect the way you use this data? Because
secondary data was not originally designed to be used by you, it often needs to be “cleansed”
for secondary usage. Cleansing involves eliminating extraneous data which may have been
included in the original data which is of no use to you, as well as performing the necessary
“data transformations” needed to get the original data in the form which you need it to be
in (as discussed earlier).
8.4.6 DATA USE
Unless you are simply going to quote from the secondary data as background information,
you are likely to want to download or enter secondary data to run statistical analysis of
some kind. Some data will be in a format ready for this. Some may not.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
114114
8.5 BIG DATA
This is a buzzword in the 21st century as massive increases in data collected from business
transactions – indeed any transaction – over the web can be stored and analysed. This data
might come from a wide array of sources such as financial transactions, accounts, customer
behaviour, buying patterns, medical records, surveillance, as well as social media sources like
Facebook and Twitter. This data does not come in simple formats, but in semi-structured
or unstructured formats such as Web clickstreams, logs, machine data, location data and of
course text. The sheer volume of such data makes it hard to see how to derive value, but this
is what data analytics is all about: mining large and complex data sets to find information
at a level which is helpful for better decision making in business.
However, don’t be put off. Fundamentally, big data is about applying statistical understanding
to wide and messy data sets. It uses the same tools we have already discussed but generally
employs software to spot patterns and trends, and produces output in user-friendly formats
such as infographics and data visualisations. Business organisations such as eBay, Walmart,
Amazon and Facebook make a virtue from the huge volumes of data their business generates,
running systems which aim to identify trends (descriptive statistics) and relationships
(correlational statistics) within the data for use by the businesses.
http://s.bookboon.com/academictransfer
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS using seCondary data
115
If you wish to access some of the outputs from such organisations, take care to find the
assumptions on which data is collected and processed. For example, Twitter can make data
available, but rarely will it actually allow anyone outside the business to see the core data
(called FIREHOSE) on which its sample data is based. Linked data sets tend to be error
prone and errors at this scale can be seriously prejudicial to the results. The other problem
here is that data requires context to make it meaningful. Just crunching vast quantities of
data together may be less than helpful without clearly tracking source contexts. You may
wish to look at the UK big data research website, which is funded by the Economic and
Social Research Council together with three UK universities and includes current debate on
Big Data, including access to published research papers using data analytics and machine
learning: http://www.blgdataresearch.org/
8.6 QUESTIONS FOR SELF REVIEW
1. Why bother with primary research when you can use secondary data?
2. What are the potential problems of using national survey data?
3. What is a meta-analysis? How does it differ from Big Data?
4. Where would you be able to find a range of OECD survey data reports online?
8.7 REFERENCES
Bannister, F. 2005 “Through a glass darkly: fact and filtration in the interpretation of
evidence”, The Electronic Journal of Business Research Methods, vol. 3, no. 1, pp. 11–24.
Lynn, P. 2003, “Developing quality standards for cross-national survey research: five
approaches”, International Journal of Social Research Methodology, vol. 6, no. 4, pp. 323–336.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
http://www.blgdataresearch.org/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
116
9 QUALITATIVE RESEARCH
METHODS: COLLECTING AND
ANALYSING QUALITATIVE DATA
9.1 CHAPTER OVERVIEW
9.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. Understand and deal with key issues in qualitative data analysis
2. Identify a range of qualitative research methods applicable to different research topics
3. Understand how qualitative data can be prepared for analysis
4. Identify computer based methods for qualitative data analysis
9.2 KEY ISSUES IN QUALITATIVE DATA ANALYSIS
Clearly qualitative research is a different kettle of fish from a quantitative study – we explored
the differences in earlier chapters. At first sight, it may seem that qualitative research is more
difficult to pin down, less precise. In fact, qualitative methods are usually governed by clear
rules and offer a way of exploring issues, which cannot easily be expressed by numbers.
Qualitative and unstructured research often provides “rich” information which is difficult
to otherwise uncover by purely quantitative methods.
An article by Rowlands (2005) offers a detailed justification of a qualitative approach to
research on SMEs and training practice. This is a useful read to discover the steps taken
in justifying a qualitative method. Qualitative methods are increasingly accepted in social
science and business research as this branch of enquiry differentiates itself from a scientific
positivist paradigm. Human organizations and human behaviour are difficult to hold still
and isolate, since they change constantly and can offer different dimensions of themselves
to different audiences. Think about the function of Public Relations and the different faces
of an organization which may be shown to shareholders, customers, staff, suppliers for
example. So it rarely makes sense to look only at numerical measured evidence when trying
to understand what is going on in an organization or other group of people. This is not to
rule out quantitative study – naturally there are financial data and other quantitative data
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
117
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
117
which can be established and monitored around business organizations and which will be of
vital importance in their study and their day-to-day management. But there is also clearly
a place for perception studies, looking at what people think or feel is going on at work, as
this will also affect day-to-day and strategic long-term practice in organizations.
Some key differences between quantitative and qualitative method are shown below
(Bryman & Bell, 2015) Some of these distinctions are arguable – for example “structured”
vs “unstructured”, macro vs micro. Also, we should bear in mind that mixed quantitative
and qualitative methods can usefully be used, where elements of both approaches can be
used both to triangulate results and to develop richer pictures still of the phenomenon
under investigation.
Quantitative Qualitative
Numbers Words
Point of view of researcher Points of view of participants
Researcher distant Researcher close
Theory testing Theory emergent
http://thecvagency.co.uk
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
118
Quantitative Qualitative
Static Process
Structured Unstructured
Generalization Contextual understanding
Hard reliable data Rich deep data
Macro Micro
Behaviour Meaning
Artificial settings Natural settings
It is helpful to reflect on the influence of the researcher in qualitative research. As we have
already discussed, there is a “researcher influence” to some extent in all research and analysis,
however qualitative methods are more likely to suggest subjectivity. For this reason, it is
essential to reflect on ways in which your qualitative data and analysis could be affected by
your standpoint and contextual understanding, as well as your expectations of the research,
and to make this explicit within your research report. It will also be necessary to be very
clear and explicit about the method of research and analysis adopted, just as we must be
in quantitative research.
9.3 THE RANGE OF QUALITATIVE RESEARCH METHODS
APPLICABLE TO RESEARCH TOPICS
9.3.1 PRINCIPAL QUALITATIVE METHODS
Action research
Appreciative Inquiry
Case study
Ethnographic research/Participant observation
Focus groups
Interviews – structured, semi-structured, unstructured
Life history research
Participant diaries
Structured observation
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
119
Some quick web searching will give you plenty of information on these different qualitative
methods. Action research for example will involve the researcher as an active participant in
the situation under study. As an actor in the organization in which they may be employed,
an action researcher seeks to explore and understand the world of which they are a part,
and action research can help all involved in that business situation to understand better
what is happening through a time of radical change – for example business restructuring,
redundancy etc. German Psychologist Kurt Lewin is known to be the “father of action
research”. He noted that action research involved a 3-step process of planning, taking action,
and then reflecting on the results, feeding back into its next cycle. It’s a cycle of continual
mapping, feasibility, implementation and improvement. The outcomes of such research are
classified as “actionable knowledge”, of real value to business organisations.
Case study research will involve more than one way of deriving data about the case or
organization/ unit under study. This may include collecting and analysing documents,
talking to people, survey data, participant observation, consumer research and any other
data collection techniques which offer qualitative information about the case. A case study
investigates a single subject at a more detailed level; that is the unit of analysis is just one
subject, whether it is a person, an organization or institution. Think of a case study as one
in which the “N” is just one (“N=1”). The intention of the case study is Gestalt in nature.
A paper by Walsham (2006) discusses the nature of interpretative case studies and methods
for carrying them out in information systems, business and other areas of social sciences.
Another useful source reference is Yin: Case Study Research – Design and Methods which
is now in its fifth edition (published by SAGE publications).
A more recent development in qualitative research methodologies is that of Appreciative
Inquiry (Cooperrider, D.L. & Srivastva, S., 1987). As a problem-solving technique,
Appreciative Inquiry (AI) focuses on what is ‘right” about a situation or condition rather
than the traditional approach to look at what is wrong. The AI process involves a four step
methodology: 1) Discover or focus in on what is working well in a system, 2) Dream or
envision other process that would work well 3) Design by focusing in on one of the dreams
that would work best in the situation and 4) Deploy or implement the proposed improvement.
Ethnographic research comes from the study of anthropology, where “tribes” are lived in and
observed for purposes of research. This kind of research will raise ethical issues, especially
about the impact of the research on the life and behaviour of the group studied. The
presence of a researcher in any group is likely to affect how people behave to one another.
In less deep and sustained involvement, participant observation may offer similar researcher
impact on results.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
120
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
120
Action research, case studies and ethnographic research tends to be more naturalistic in its
process, in that there is no direct intervention or artificiality as found when quantitative, and
even other forms of qualitative research such as focus groups, interviews etc. are conducted.
9.3.2 APPROACHES TO QUALITATIVE ANALYSIS
Analysis method Outline definitions
Analytic induction
Systematic analysis of text or other qualitative data to build categories
and sub- categories within the data – attempting to construct a rich
picture from data.
Cognitive mapping
Determining how individuals construct mental models, often using visual
or spatial means (diagrams, pictures) to produce a map of relationships
between concepts as understood by the subject.
Data display
and analysis
Analyzing data by first reducing qualitative data to a simple set of ideas,
then displaying those ideas in order to draw and check conclusions from
the data.
http://s.bookboon.com/Subscrybe
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
121
Analysis method Outline definitions
Discourse analysis
Analysing language used by subjects to determine how the world of a
subject is constructed through the use of language.
Feminist research
Seeing reality through the perspective of gender. Feminist research tries
to revise how organisations are understood, challenging male-oriented
values and paradigms.
Grounded theory
Inductive approach to building theory from data, usually from interviews
and observations. Various versions exist, but the principal method is
to code phenomena within the data (e.g. an interview transcript) and
to repeatedly revisit the data in order to refine and “saturate” the
categories derived, before meaningful theory can be built.
Historiography
Involves the study of historical method, for example when revisiting an
event, this would mean analyzing the epistemological position taken
when versions of the event are written and challenging the authenticity
of such positions.
Narrative analysis
Collecting and analyzing qualitative data without fragmenting it – this
method preserves the narrative/story of the data to maintain a sense of
time, place and sequence in the data.
Phenomenography
Seeks to describe the way different people view a phenomenon, usually
based on interview data taking an interpretivist approach The aim is to
develop an intense picture of the phenomenon by defining the outcome
space (different people’s views of the same thing) and investigating
relationships within that space.
Phenomenology
Approaches social phenomena from the perspective that they are
socially constructed. Concerned with discerning meaning within the
phenomenon and trying to put aside researcher’s prior understanding
of the phenomenon.
Template analysis
Creating a hierarchy of categories from the data and using this as a
template to explore qualitative data examples
These approaches may overlap in some cases: for example grounded theory is a very detailed
iterative method of, usually, interview transcript analysis and thus involves some similar
activities to phenomenography, where such transcripts may also be interrogated by the
researcher in a very detailed and iterative way. The aim in both cases is to dig into the text
to look for categories or themes in the data which may be built into rich ideas or theories.
Both are examples of inductive research i.e. theory building rather than theory testing.
While these different methods of qualitative analysis are very distinct – any research methods
textbook or website will give full descriptions – they all involve a rigorous attempt to look
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
122
at qualitative data (descriptions, discussions, activities, ideas presented verbally, audio-visually
or in text) which offer a range of research interpretations. It may be the case that different
researchers using the same analysis method on the same data could find different ideas and
theories. They will be interpretations and these will be subject to debate and challenge.
Qualitative analysis must therefore be as rigorous and transparent in method as possible,
to allow readers of such research to understand how conclusions and findings are achieved.
They may not be exactly reproducible, as would be expected in experimental science, but
that does not invalidate such results. The outcomes of qualitative research, like those of
quantitative research, may be disputed; which is why it is vital to detail the methods used
for collecting and analysing the data, and to explain as clearly as possible the researcher’s
own paradigm or philosophy about research, so that readers may understand where the
ideas come from and how they may be filtered by the researcher.
9.3.3 WHAT ARE THE KEY OPERATIONS REQUIRED
IN QUALITATIVE DATA ANALYSIS?
• Where data is derived from interviews – individual or group, structured or semi-
structured or unstructured – there will be a need to transcribe the recording of
that interview before analysis. This brings its own problems of time, cost, method
and detail.
• Development of themes, categories or ideas (from the literature (which may then
be used to offer a hypothesis for testing in the data-deductive approach) or from
the data itself (inductive approach). This is often referred to as thematic analysis.
• Unitising, coding or finding units of meaning within the data, which relate to or
add to or amend the categories.
• Constant comparative method leading to saturation of categories – this terminology
comes from grounded theory but the activity is not confined to this approach; a
constant iterative process of checking how the data meanings fit the categories
or themes.
• Understanding the variation and role of language as an intermediary in the
communication of ideas.
• In many cases, the production of researcher summaries, log books, contextual notes
to help provide further explanatory detail to transcripts or observations.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
123
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
123
9.4 HOW QUALITATIVE DATA CAN BE PREPARED FOR ANALYSIS
In most cases, this will involve some kind of transcription. Although simply taking notes of
observations, and in some cases in interviews, may be sufficient, a transcript is important
for conversations in which the researcher wishes to play some part, so that they are not
required to both write notes and conduct an interview or group discussion. However, recent
advances in computer software provide programs which automatically transcribe voice to
text with considerable accuracy. This has eliminated many of the administrative inefficiencies
in conducting interviews and focus groups. (You will need to be sure to observe legal and
ethical implications and obtain permission and/or authorized consent when recording and
transcribing input from your subjects). Nuance’s Dragon 13 is one such program benefitting
from its 13 editions of revisions and improvements. There are other similar free programs
available online and for download.
This will mean gaining agreement for recording, finding a suitable instrument for recording
and transcription and undertaking the transcription itself. Suitable recording equipment
will not be too intimidating for the interviewee(s), will be reliable (!), will have a reliable
power supply, will have a microphone which can pick up every speaker clearly including
the researcher, and will produce adequate sound quality for transcription. Remember that
interviewees may begin by speaking clearly and loudly as they are aware of the recording,
http://s.bookboon.com/EOT
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
124
but voice tone and pitch may soften later, so set the recording level high. Always test the
recorder at the start of an interview.
Increasingly, mobile phones and digital voice recorders are being used for this purpose as
they are easily available and can record longer sessions than many audio tapes. You may
wish to use both a conventional audio tape recorder and a digital recorder to make quite
sure a useful recording is made. Digital recordings are useful as interviews can be played
back to the researcher through MP3 players, phones or computers, removing the need for
sitting by a tape player. Do not be tempted to voice record without gaining full agreement
from the interviewees, clearly this data cannot be used ethically if collected without consent.
Beware voice-activated equipment (which switches off when there is nothing being said)
as this can lose definition owing to the transition from off to on when a voice is heard.
It can be possible to pay someone to transcribe interview data for you, which can be
helpful if there is a great deal of interview data. However, this does deny the researcher the
opportunity of getting to know the interview in great detail during transcription; sometimes
it is preferable to do this personally – or it may be the only alternative available.
When transcribing, maintain a context sheet to record non-verbal interventions or
interruptions, as this data may affect how the transcript is understood. Transcripts should
ideally be double-spaced to allow for coding and other notes to be made on the document.
Decide rules for referring to individuals in the interview (actual names are not usually typed
up for reasons of confidentiality). You will also need to decide how to type up repetitions
of words and phrases, as this is common in speech patterns but usually adds little to the
data. Bear in mind that a good typist can take 4–8 hours to type up one hour of interview,
this is a very time-consuming process.
Respondent validation: it may be helpful to send transcripts for checking by the interviewees.
This helps to build credibility in the transcripts, but is not always acceptable to the interviewee.
At minimum, you the researcher must check every transcript against the recording, as it is
easy to make mistakes in transcription (sometimes a mind-numbing process) yet such errors
may lead to considerable effects on analysis.
9.5 COMPUTER BASED METHODS FOR
QUALITATIVE DATA ANALYSIS
Computer aided qualitative data analysis software (CAQDAS) is increasingly available,
though like voice recognition software, this will not necessarily reduce analysis time by a
great deal and will not be that straightforward. It is particularly helpful when there is a
very large amount of data for analysis.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
125
For example, QSR International produces a software program NVIVO, which allows researchers
to use the computer to organize and analyze the findings from purely qualitative and mixed
methods research. NVIVO helps researchers to input results from unstructured interviews,
and open-ended survey responses, focus groups, social media and other web content.
CAQDAS can produce quantitative data from qualitative methods, for example by producing
frequency data on particular events, words etc. However, its main use is the qualitative
analysis of such data as interview transcripts or narratives. They do this by organising the
data, providing instant access to all data once entered, searching and retrieving particular
words or phrases. As with other CAQDAS programs, NVIVO classifies, sorts and arranges
prose, and provides output which can be integrated into other software platforms such as
MSWord, Excel, or IBM SPSS Predictive Analytics. The program allows the researcher to
analyze large amounts of prose-based information and identify trends and themes which
are extracted. NVIVO helps those doing qualitative research by organizing and analyzing
prose information for the researcher to gain insight on, interpret, construct conclusions,
and take action.
If NVIVO is not available as a package on your computer, then visit the website of its supplier
www.qsrinternational.com and download a demonstration of the software to investigate what
it will do and how it feels. You can also opt for a 14 day free trial to try it out yourself.
9.6 QUESTIONS FOR SELF REVIEW
1. What are the key differences between qualitative and quantitative research methods?
2. What are the main activities involved in qualitative analysis?
3. What is action research – can you provide an example of how this might be used
in business research?
4. What is the case for and against someone else transcribing your interview data?
5. To what extent is NVIVO likely to produce different results from your qualitative
data than analysing manually?
9.7 REFERENCES
Bryman, A. & Bell, E. 2015, Business research methods, 4th edn. Oxford University Press,
Oxford UK.
http://www.qsrinternational.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
126
QUALITATIVE RESEARCH METHODS: COLLECTING
AND ANALYSING QUALITATIVE DATA
126
Cooperrider, D.L. & Srivastva, S. 1987 “Appreciative inquiry in organizational life”. In
Woodman, R.W. & Pasmore, W.A. (eds), Research in Organizational Change and Development,
Vol. 1, pp. 129–169. JAI Press, Stamford, CT.
Rowlands, B. 2005, “Grounded in practice: using interpretive research to build theory”, The
Electronic Journal of Business Research Methodology, vol. 3, no. 1, pp. 81–92.
Saunders, M. Lewis, P. & Thornhill, A 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
Walsham, G. 2006, “Doing Interpretative Research”, European Journal of Information Systems,
vol. 15, no. 3. pp. 320–330. (available online)
Yin, R.K. 2014, Case study research: design and methods. 5th edn. Sage Publications, Thousand
Oaks CA
PDF components for PHP developers
www.setasign.com
SETASIGN
This e-book
is made with
SetaPDF
http://s.bookboon.com/Setasign
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
127
10 PRACTICAL ISSUES IN
CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT
OBSERVATION
10.1 CHAPTER OVERVIEW
10.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. Distinguish practical considerations relating to participant observation
2. Anticipate and handle practical issues relating to interviews
3. Distinguish and prepare for practical issues relating to focus groups
4. Conduct interviews and focus groups to collect qualitative data
10.2 PRACTICAL CONSIDERATIONS RELATING
TO PARTICIPANT OBSERVATION
10.2.1 ETHNOGRAPHY OR PARTICIPANT OBSERVATION?
Both ethnography and participant observation involve submersion of the researcher into
the context under study. As mentioned in the previous chapter, ethnographic research has a
more social anthropological feel and may focus more on business “tribes” and organizational
settings such as departments and functions, or different national sites of operation. The
focus will be on the community described and its symbols, culture, interactions, rituals,
language etc. Participant observation will be used to allow the researcher direct experience
of a specific situation or event, perhaps working in a factory or office setting during a
transition period. However, in some texts, the two terms will be used interchangeably, so
when discussing practical issues, we can classify them broadly together.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
128
10.2.2 ACCESS
Both approaches involve intense involvement of the researcher in the field, in order to feel
like an “insider” and try to understand and explain what that feels like. This is usually
difficult to do in the short term, so a time commitment to the research will be the first
hurdle. In some cases, where a researcher is also employed in the organization being studied,
this should not prove too problematic (but does raise other issues of covert research to be
covered next). Where the researcher has no other role in the organization being studied,
there will probably be protracted negotiations to allow this kind of long-term access.
Think about how you might gain access to an organization for this kind of study? Letters?
Emails? Contacts? How do you convince them of your credibility and trustworthiness?
10.2.3 COVERT OR OVERT RESEARCH?
If research is undertaken covertly, without authority, then problems of access and of reactivity
disappear. However, a number of others appear instead. For example, the sheer practical
difficulty of taking detailed research notes when you are meant to be working on the job!
Also being unable to use other research methods during this period such as interviews.
There is anxiety about possible discovery of the researcher role and activities, anxiety which
is well-founded, since if the covert research is discovered, there is a strong chance the study
will have to be abandoned before completion.
Most of all, however there is a problem of ethics, since participants in the research will not
have the opportunity for informed consent and their privacy is violated. This can damage
the research and researcher if it is discovered, but can also damage the reputation of research
in general amongst those whose trust was betrayed.
Is there a happy medium? For example, is it possible to have senior management authority
but not to divulge your intentions to colleagues? What kind of difficulties might this cause?
Or could the broad purpose of “research” be discussed openly, but the specific focus and
question be kept secret?
Whichever conclusion you reach, your research report will have to show clear details of the
overt or covert nature of your research, and there would need to be very good reasons for
a covert approach.
Think about a scenario in which you would be tempted to try covert participant observation.
For example, suppose your place of work was threatened by relocation to a new venue and
you wanted to study the effects of this move on the team’s performance. As part of the team,
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
129
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
129
you are now in a double role – team-worker and researcher. Think about the challenges and
constraints this imposes on you. You want to use covert observation because you feel that if
you tell the team you are watching their reactions and conversations, they will either reject
you as a team-worker or will change their behaviour because they are being watched. Could
they get to know about your role as researcher somehow? That might seriously affect your
chances of continued employment in the team, since, even with your manager’s agreement
to the research, your colleagues may feel they cannot trust you again.
10.2.4 RELATIONSHIP-BUILDING
Whatever approach is taken to participant observation, the researcher will need to develop
skills of relationship-building,
• to allay colleagues’ suspicions about being a representative or spy from top management,
• to maintain a degree of objectivity while in the organization (rather than helping
to affect the very relationships being studied e.g. a particular view of management
or other functions or companies) and finally
EXPERIENCE THE POWER OF
FULL ENGAGEMENT…
RUN FASTER.
RUN LONGER..
RUN EASIER…
READ MORE & PRE-ORDER TODAY
WWW.GAITEYE.COM
Challenge the way we run
1349906_A6_4+0.indd 1 22-08-2014 12:56:57
http://s.bookboon.com/Gaiteye
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
130
• to develop key informants who can be used to provide broad organizational
background and check out stories you are told, or help you to find appropriate
people to get to know.
For example, a recent student conducted a study on flexible working in her own organization.
As a representative of HR, her role actually precluded the hearing of much gossip and
informal talk, as HR could be seen as “the enemy” in relation to contracts negotiation.
However, her gender, together with strong interpersonal skills, enabled her to get to know
about unofficial flexible working through informal networks built on trust over time.
This research ultimately exposed serious double standards in the way flexible working was
represented by some managers in the company.
10.2.5 ROLES FOR PARTICIPANT OBSERVERS
Bryman and Bell (2015) discuss research by Gold in 1958 setting out four roles for
participant observers: complete participant (covert observer), participant-as-observer (complete
participant but overt researcher as well), observer-as-participant (primary role is researcher
but can participate in work) and complete observer (no participation in work and little
communication with those observed). Further views of the different possible roles are offered
by Gans and Bryman & Bell. The sense of exchange is usually helpful, since research data
can be gained in exchange for consultancy advice, survey work or straight labour. Perhaps
the biggest temptation is to “go native” i.e. to become fully absorbed into the perspective
of the participant role, and thus to lose the objectivity of the researcher role.
If you are interested in ethnography or life history research, you may wish to consult an
article on these methods by Gordon and Lahelma (2003).
10.3 PRACTICAL ISSUES RELATING TO INTERVIEWS
10.3.1 STRUCTURED, SEMI-STRUCTURED, UNSTRUCTURED
If an interview is fully structured in format, does this mean it is quantitative research?
To some extent yes, in that clear questions are asked in a consistent way, similar to the
administration of a questionnaire by telephone. However, the mere fact that the interviewer
and interviewee are face-to-face brings another dimension to the research method. When
we can see our interviewees, we introduce the concept of non-verbal communication – not
just from them (which helps us understand more about them) but also from us – which
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
131
can steer or emphasise certain areas, mislead or explain further items which would otherwise
be misunderstood or left blank and so on.
Semi-structured interviews will be based on a question guide, the contents of which will
always be asked of respondents. However, since this is not fully structured, the interviewee
is allowed to go where they want with the questions and to divert to other things which
interest them. Since the focus of a qualitative interview is the interviewee, not the interviewer,
this is fine.
Unstructured or in-depth interviews can go right off the point – and that may be the point,
i.e. to discover much more about the interviewee by what they say and think, than how
they answer specific questions. These are conversations with a purpose, wide-ranging and
thus likely to deliver rich but inconsistent data.
The interviewer’s role is to manage the process (e.g. the time if a particular duration has
been promised, the key questions are asked and the conversation stays broadly around the
research question). The interviewee can be subject to very few constraints.
Many of the issues we raised around the design of questionnaires in an earlier session apply
here to questions used in interviews (for example no leading questions or double questions).
Where some are prepared in advance, it is advisable to give a copy to the interviewee in
advance if possible, so that rather than “whatever comes into their head at the time”, the
interviewer will have the benefit of a reflective response. Some structured questions in an
interview can help to provide consistency where multiple cases are studied or where more
than one interviewer is used.
10.3.2 THE ISSUE OF TIME
When setting up an interview, time booked will take on great importance to the organization
and the individual, who will be trying to fit this interview in around other duties of the
day. However, it is normally the case that, once the interview has started, the interviewer
will find difficulty in stopping the interview, as the interviewee enjoys the experience and
begins to use it for personal reflection or simply the enjoyment of discussing a work issue
with an adult in a way they cannot do with colleagues.
From the perspective of planning, especially if carrying out several interviews in one visit,
the interview period should be realistic but not too long, whereas a considerable margin of
time should be allowed between interviews in case of expected over-run.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
132
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
132
As with any work interview, including a selection interview, it is vital that the visitor to
the organization (the researcher) turns up on time and in time to begin at the agreed start
time, after getting to the right place and setting up and testing the recording equipment.
So arriving at least 10 minutes early is usually helpful.
10.3.3 THE INTERVIEW GUIDE
Preparing key questions in advance is very important if you aim to both achieve your
research outcomes and be consistent and professional in your approach to interviewing.
However, being over-dependent on the pre-prepared interview questions can be dangerous.
A professional interviewer is genuinely interested in the interviewee’s perspective and so
will flex the questions to follow new directions suggested by the interviewee. Flexibility will
make each interview more enjoyable to conduct, rather than feeling slavishly controlled by
the pre-set guide. Finally, there is a common tendency for an interviewee to anticipate later
questions, often without any prompts from the interviewer. It will be important to allow
them to go there, rather than saying “I wonder if we could leave that point as it comes
up later.” Inevitably this will cause the interviewee and interviewer to forget what was just
said, so you probably won’t get it later.
How to retain your
top staff
F I N D O U T N OW FO R FR E E
Get your free trial
Because happy staff get more done
What your staff really want?
The top issues troubling them?
How to make staff assessments
work for you & them, painlessly?
DO YOU WANT TO KNOW:
http://s.bookboon.com/performancereviewpro
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
133
However, if later questions are covered early on, don’t worry about running out of questions.
Confident interviewers, by demonstrating empathy and genuine regard for the interviewee,
can always facilitate further discussion by simple prompts such as “can you tell me more
about…?” “that’s an interesting point, I hadn’t thought of that, so what exactly do you
mean by…”, “I’m not sure I have fully understood, can you explain that a little further…
or give me an example?”. (Such questions assume you have not run out of time, and the
initial questions are all answered.)
10.3.4 INTERVIEW BEHAVIOUR
Research cited in Bryman and Bell (2015) by Kvale suggests that an interviewer should be:
• Knowledgeable
• Structuring
• Clear
• Gentle
• Sensitive
• Open
• Steering
• Critical
• Remembering
• Interpreting
They also add the adjectives: balanced and ethically sensitive to the list. To this list, we can
add interview competencies suggested by Saunders, Lewis and Thornhill (2015):
• Opening the interview
• Using appropriate language
• Questioning
• Listening
• Testing and summarising for understanding
• Recognising and dealing with difficult participants
• Recording responses
• Closing the interview
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
134
10.3.5 AFTER THE INTERVIEW
Urgent action is needed to make notes about what happened. These are contextual notes,
which will later shed much light on the event. You might note down personal impressions
of how it went, where it happened, specific comments on the outcomes, the setting in
which it took place, the state of mind of the interviewee from your current perspective etc.
It will also be necessary to arrange to transcribe the interview from the recording as quickly
as possible. Within a day or so, it is easy to remember what an interviewee was trying to
express, even if the recording is not good. Later on, this will become very difficult.
You may wish to search for an article on interviewing (Carter 2004) which offers some
useful ideas about the practical challenges of the interview process.
10.4 PRACTICAL ISSUES RELATING TO FOCUS GROUPS
A focus group method is a focused group interview. There will be several participants, the
researcher as facilitator and a method of recording what is said, preferably video recording,
as audio can be difficult to follow when several people are speaking. Video recording will
also give much richer contextual evidence about how people interact. However, resistance
to video recording is much greater than audio recording.
Another key difference with a focus group is that there is usually a specific topic on which
discussion is to be held, rather than a whole series of topics. The point of interviewing in
this way is to explore the joint construction of meaning around a specific topic and to see
how group dynamics and interaction work in this process.
Focus groups are not easy to run, although they get easier with practice. Focus groups can
be creative places, but can also be full of challenge and conflict – this needs a light touch
of management from the researcher, only to ensure good standards of communication and
respect are encouraged, not to stop conflict since this can be a productive source of creativity
and meaning development. Issues can surface in a much freer way in a focus group than
in an individual interview, and can be considered a more naturalistic context for testing
and developing ideas.
In order to decide how many focus groups to hold and who should attend, some of the
sampling issues discussed earlier should be discussed, such as stratified or snowball sampling.
For example, are there variables, which must be represented in focus group membership
(e.g. different departments, levels of work, length of experience etc.) – this may increase the
number of groups held. Broadly however, focus groups can continue to be held until the
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
135
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
135
ideas and themes raised become familiar and can be anticipated by the researcher (saturation
of categories). To some extent, feasibility and cost/time issues will also dictate number of
groups held.
Each individual focus group should consist of participants of a rather homogenous grouping.
Then, collectively, all focus groups pertaining to a common theme can be integrated to
identify differences and similarities among the groups. For example, an organization may
wish to determine the “organizational climate” among its employees. Several individual focus
groups would be necessary. One may consist of “rank and file”/hourly employees, one group
would consist of supervisors, one of middle management, and one of upper level executives.
Mixing employees of different ranks in a single focus group is possible but can inhibit or
contaminate the information collected. Similarly, a marketing department may wish to
conduct a series of focus groups on customer preferences. Stratifying their customers against
key demographics and creating focus group sessions for each of these demographics would
be necessary. Homogeneity among individuals in any one focus group is particularly helpful.
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
© Deloitte & Touche LLP and affiliated entities.
360°
thinking.
Discover the truth at www.deloitte.ca/careers
http://www.deloitte.ca/careers
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
136
Size of groups will depend on practical factors, including size of available rooms, but ideally
six to eight will be the easiest number to manage. While attendance may be easy to control
within an organization, inter-organizational studies will prove harder to schedule, with no-
shows a common feature if people have to travel to attend the focus group.
The facilitator role varies greatly in focus groups, partly depending on the understanding of
the process by participants. Too much control from the facilitator will make it difficult for
a free-flowing discussion to construct meaning and reveal new insights. Too little control
from the facilitator may lead to lack of time discipline and the ignoring of some of the
key issues. Some greater control is usually helpful at the outset, in setting ground rules for
the session and explaining that the facilitator does have the right to intervene for time or
agenda reasons, or perhaps to request an explanation. Once this is set up, the group can
be encouraged to warm up on its own, and will soon get going, provided they trust the
researcher. Where steering is needed, the facilitator can then intervene as needed.
Analysis of the focus group is done once the focus group is completed and the recordings
are transcribed, and evaluated. Hyden and Bulow (2003) have produced a useful account
of focus group methodology which may help you to review this approach.
10.5 QUESTIONS FOR SELF REVIEW
1. What are the (small) differences between participant observation and ethnography?
2. How could you adapt a semi-structured interview process to be conducted by email?
3. What do you think would be your biggest challenge in conducting research
interviews? What could you do about this?
4. Why do you think focus groups are so widely used to test new products and new
policy ideas?
10.6 REFERENCES
Bryman, A. & Bell, E. 2015, Business Research Methods 4th edn. Oxford, Oxford
University Press.
Carter, J. 2004, “Research note: reflections on interviewing across the ethnic divide”.
International Journal of Social Research Methodology, vol. 7, no. 4 pp. 345–353.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS
137
PRACTICAL ISSUES IN CONDUCTING INTERVIEWS,
FOCUS GROUPS, PARTICIPANT OBSERVATION
137
Gordon, T. & Lahelma, E. 2003, “From ethnography to life history: tracing transitions
of school students”. International Journal of Social Research Methodology, vol. 6, no. 3,
pp. 245–254.
Hyden, L.C. & Bulow, P.H. 2003, “Who’s talking: drawing conclusions from focus groups –
some methodological considerations”, International Journal of Social Research Methodology,
vol. 6, no. 4, pp. 305–321.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
TMP PRODUCTION NY026057B
PSTANKIE
gl/rv/rv/baf
ACCCTR0005
Bookboon Ad Creative
4
6 x 4
12/13/2013
Bring your talent and passion to a
global organization at the forefront of
business, technology and innovation.
Discover how great you can be.
Visit accenture.com/bookboon
©
2013 A
ccenture.
A
ll rights reserved.
http://s.bookboon.com/accentureCZintl
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
138
11 FORECASTING TRENDS
11.1 CHAPTER OVERVIEW
11.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. Understand why forecasting is not widely covered in the business research
methods literature
2. Identify existing methodologies for forecasting
3. Understand a range of qualitative and quantitative forecasting tools
4. Understand various measures commonly used to evaluate forecasts
5. Understand the value of forecasting methods in business practice
11.2 WHY FORECASTING IS NOT WIDELY COVERED IN THE
BUSINESS RESEARCH METHODS LITERATURE
Surprisingly few research methods textbooks contain sections on forecasting trends. Why
surprising? Because, in business, this is a key activity. We can see that the main business
of a researcher is to look backwards and try to see what was happening in a particular
situation involving particular variables and people. If research is rigorous, then it may be
possible to apply lessons from the past to a current situation. But it is not seen as the
job of an academic researcher to try to predict the future. Attempts to do this are found
only in concluding paragraphs of research articles, and they will often be suggesting more
research in what appear to be developing trends. The Makridakis et al book mentioned in
the reference list remains one of the seminal works on forecasting.
This is not coy. It is simply because predicting and forecasting trends is a very risky business,
and rigorous research aims to avoid very high risk strategies. Yet people do, of course, predict
trends. Management gurus and writers frequently aim to identify what is about to happen
in business. If we see them as credible people, we may be persuaded by their predictions.
But reality often proves them wrong. Some predictions in business will be about the next
wonder product. New Product Development is always highly risky but is engaged in order to
develop profit streams, deal with product life cycles and develop businesses. The risk is clear
if we think about the current split between companies developing two different technologies
for enhanced DVD performance. In early 2007, the market was impossible to call – would
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
139
we be buying High Performance or Blu Ray? Some companies are producing equipment
compatible with both systems to avoid what happened when the last major split in this
market caused the death of Betamax video systems in favour of VHS. Now in 2014 these
physical video playback media are being replaced with internet-delivered video through a
variety of sources. Could this all have been predicted? If so, how? Read on.
Surely we can do better when forecasting something we know about – such as next year’s
company sales figures? The evidence shows us that in fact we are poor at predicting even
figures with which we are intimately involved and which directly affect our company’s future.
To start with, we confuse personal views and impacts with professional forecasts. If we ask
a sales person to predict the level of her/his sales next year, that prediction will be partly
based on market knowledge, partly on protecting their own position in the company, partly
on an estimate of the outcomes of getting it wrong. People, unlike machines, are complex
and unpredictable. For example, if sales people predict a high increase in sales, targets are
likely to be set high, making it hard for them to achieve targets. On the other hand if they
set them to show any kind of downturn, the sales people themselves will probably get the
blame. So predictions tend to be cautious when personal targets and responsibility could
be at stake. Equally if we have a new business idea and want to borrow money from the
bank, it is likely that we will over-estimate potential sales, and the time at which our cash
flow will turn positive.
Even when we are simply trying to predict sales forecasts, there will be many different
people and departments of a business involved in this prediction. Information from inside
and outside the company is relevant, and the quality of both may vary. Since many people
are involved and different variables studied, any errors or inconsistencies or communication
failures will make this a very imprecise activity indeed.
To quote a larger example cited by Makridakis et al (p. 491), look at the Eurotunnel project
involving major engineering work between UK and France to build a rail link under the
sea. In 1986, passenger estimates were 16.5 million for the first year of operation. In 1993
this forecast was reduced to 13 million. In 1994 it was reduced to 6 million. The first full
year (1995) produced 3 million passengers. The actual cost of building was also more than
twice the initial estimate and the intended data of opening was missed by almost two years.
Such big projects as this, and, for example, bids to host the Olympic games, are frequently
subject to major revisions as more actual data becomes available.
Just because the idea or technology is possible, it isn’t necessarily feasible or implementable;
and just because it is feasible, doesn’t mean people will want to do it. Can you think of
examples of possible technologies, which either are or are not feasible technologies, which
are not wanted?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
140140
Forecasting is nonetheless an important business activity underpinning the accurate
determination and acquisition of resources (human, capital, buildings, money, energy,
materials etc.) and the scheduling of their use.
Clearly this chapter is more focused on research in business, than on academic research
about and for business. But as we discuss the issues in forecasting, the debates we have been
having will begin to recur; for example, the extent to which the researcher’s assumptions
affect the outcomes of that research, the opportunities and risks associated with quantitative
data and its analysis, and an attempt to understand how qualitative research can contribute
to this field. All these ideas apply to forecasting trends.
11.3 EXISTING METHODOLOGIES FOR FORECASTING
Forecasting methodologies can be divided into:
1. quantitative techniques, which generally use historical time series data as the basis
for projection, and regression analysis to determine the relative importance and
relationships of variables,
Unlock your potential
eLibrary solutions from bookboon is the key
Interested in how we can help you?
email ban@bookboon.com
eLibrary
http://bookboonglobal.com/en/qualities/corporate-education-elibrary-solution/
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
141
2. qualitative techniques, using scenarios which are known to explore the unknown, and
3. creative techniques, which aim to suggest possible alternatives where there is no
factual basis of information.
11.3.1 QUANTITATIVE TECHNIQUES
These assume some historical, numerical data is available and that the patterns found in
the data may continue (assumption of continuity).
Managers frequently use numerical data in an intuitive way, using their judgement and
experience (or the toss of a coin!) to predict how trends will move. This is so widespread as
to be a norm in much business practice – why is this, when formal quantitative techniques
using statistics both exist and can tell us more about the possible trends? It seems that we
are seduced by numbers and react emotionally to them – usually seeing them as important
because they are numbers, not from any intrinsic importance. (This explains why spurious
research studies can get great media coverage by producing shocking statistics – you may
wish to look at the book Bad Science by Dr. Ben Goldacre for some interesting examples).
Time series forecasting
This is about finding patterns in historical data and extrapolating them into the future.
This approach does not attempt to understand why the data behave as they do, because the
data is seen as too complex to understand or difficult to break down and use, or because
we don’t need to know what affects the data pattern, only the outcome of the data values.
Such techniques are used to plan and schedule outcomes in business. The assumption of
time series forecasting is that past behaviour/performance of a system will predict future
behaviour/performance of the same system. In times of significant change, this assumption
is rarely true.
Explanatory models
Here we do look at why the data behave as they do, and attempt to identify the key variables
affecting the data values. It is unlikely that the variables we investigate will account for all
the change in the data value, so an element of error is introduced to represent what we
cannot explain. Such models are used for policy formulation. Identifying the key variables
is a challenge in explanatory forecasting models.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
142
11.3.2 QUALITATIVE TECHNIQUES
These may be used alone or in conjunction with quantitative techniques and involve
the contribution of experts. Such experts may be professional forecasters and planners,
or consultants with a deep knowledge of an industrial sector, or facilitators who know
how to harness the knowledge of in-company talent to produce forecasts for the medium
and long term. Qualitative approaches may be used for strategy formulation and product
development. A Dephi study is one such qualitative technique utilizing the judgement of a
panel of experts and synthesized into a final report and prediction by the panel facilitator.
11.3.3 CREATIVE TECHNIQUES
These are used broadly when neither of the other sets of techniques can help because there
is a lack of historical data. For example, how do we extrapolate trends for new technologies,
which have only just arrived? How do we predict macro level changes such as climate change,
when vast computing power is needed for the number of potential variables, and much of
what is known is estimation not proven knowledge?
The answer is to use the power of the human brain to make connections between the
forecasting problem and other knowledge. The use of analogy, for example, i.e. finding a
storyline, which may be made to fit the problem in order to explore possible outcomes or
add to possible predictions, dates back at last to Aristotle. Analogies may be taken from
a different discipline (e.g. biology related to engineering) or fiction (well known plotlines
which can be applied to a situation to develop possible outcomes), or simply factual stories
of other products, or business decisions. Makridakis in Chapter 9 describes three helpful
characteristics, which seem to apply to long term predictions:
• Accurate over the long term but impossible to identify when (e.g. Roger Bacon
predicting submarines in 1260).
• Disbelief from most people, even those directly affected, about the potential of
new inventions (famous example of the chairman of IBM predicting a maximum
demand for computers of 100 in the early 1950s, similar reluctance to predict the
spread and use of mobile phones).
• Over-prediction of the benefits and scale of new technology once it has started to
spread (this is the “paperless office” type prediction).
Scenario-building is another creative technique in future forecasting, in fact for many global
businesses, there are departments specialising in this area of research.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
143143
Scenarios are built on some historical information, plus subjective interpretations, hunches
and specific assumptions. Their purpose is not necessarily to provide accurate predictions,
but to challenge linear models of prediction, since actual change is not usually linear, but
most predictive methods produce linear outcomes. Big business must invest in this type of
activity to protect its territory and find opportunities before the competition.
11.3.4 FORECASTING STAGES
1. Define the problem and the need
2. Collect information – quantitative and qualitative data
3. Exploratory analysis – look for patterns in the data, possible trends, seasonality,
cyclical patterns, relationships in the variables
4. Select forecasting techniques e.g. exponential smoothing, regression and more
advanced statistical models or opting for qualitative or creative techniques
5. Use the model and evaluate the forecasts produced.
.
http://s.bookboon.com/AlcatelLucent
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
144
11.4 BASIC FORECASTING TOOLS
Here are the popular ones:
• Time series – much of the secondary data discussed in our earlier chapter is produced
in this format, so could be used for prediction. Clearly cross-sectional data is all
from one time period so cannot be used for time series.
• Graphical summaries – line graphs of a variable against time (horizontal axis),
shows trends in historical data, special events, cyclical or seasonal patterns (latter
from monthly data). A seasonal plot will use a line graph over the period of a year,
with different annual data plotted together to show similarities and differences.
A scatter diagram will be useful to show cross-sectional data in how one variable
relates to another, this may be of use for explanatory modelling. Where a linear
trend can be seen in a graph, a “straight line” forecast can be made (though will
not necessarily be accurate!).
• Numerical summaries – univariate statistics e.g. mean, median, mode, standard
deviation and bivariate statistics e.g. co-variance and correlation have been discussed
in our chapter on quantitative techniques. All help to get to know the data in
preparation for forecasting. All statistics can be shown over time. Time series data
can compute autocorrelation, which can be shown clearly in a graphical way e.g.
correlogram.
• Averaging –a simple forecast method uses an average of monthly data over a time
period of some years to be the predicted forecast figure for that month in the
next year.
• Prediction intervals – used to give an estimate of the range within with the actual
value will fall, if the forecast value and Mean Squared Error has been computed.
The formula uses a standard z-value, which is associated with a particular probability
level – i.e. in the example z=1.645 is associated with a 90% probability level.
Fn+1 ± z..MSE
• Least squares estimates – a way of estimating values for which the mean squared
error (MSE) is at a minimum. It is an estimation of goodness of fit of a relationship
between variables.
• Accuracy of linear regression, when using an explanatory model and a particular
variable has impact on the forecast. In other words, this is about working out the
relationship between one dependent variable (to be forecast) and an independent
variable which could explain how the dependent variable changes. If there is more
than one independent variable, multiple regression is used. Forecasting is done by
understanding the relationship between the dependent and independent variables,
such that we can use new values for the independent variable(s) and predict
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
145
corresponding values for the dependent variable. The dependent variable is often
referred to as the “outcome variable” (the thing being predicted) and independent
variables are often referred to as “predictor variables” as they are the variables used to
make the predictions. The accuracy of multiple regressions improves as the number
of non co-linear variables are introduced into the regression equation.
• Transformations and adjustments – include mathematical transformation of the data
values (e.g. square root or logarithm) of each value to smooth the variation and
make forecasting simpler, calendar adjustments to take account of different lengths
of months in some data given per day, adjustments can also be made for numbers
of trading days in a month or for inflation or population change. A simple moving
average (e.g. averaging the value before during and after the period and using this as
the new data value) will provide a simple and understandable smoothing technique
to allow patterns in the data to be more visible.
11.5 REGRESSION AND DISCRIMINANT ANALYSIS
Regression Analysis is a quantitative statistical technique often associated with forecasting.
It is also used as a predictive technique. Regression attempts to forecast or “predict” the
value of an outcome variable (a dependent variable) based upon one (“Simple Regression”)
or more (“Multiple Regression”) independent variables in the regression equation. The basic
equation for regression analysis is:
Y= a + (b1 * x1) + (b2 * x2) + (bn * xn)
Where: Y = the thing being predicted (the dependent variable), a = the alpha weight (constant)
of the equation, b1= the beta weight for each independent variable x1= the actual value for
each independent variable (note: Y = the predicted value and Y' = [Y prime] the actual
value). The simple correlation between Y and Y' is called the multiple correlation coefficient.
The power of multiple regression increases when there is a weak, or little, correlation between
all of the independent variables included in the equation. Multi-collinearity is the term used
to describe the amount of correlation among the independent variables. As a researcher,
you want the multi-collinearity of the independent variables to be low. The independent
variables should be individually related to the dependent variable, but relatively unrelated
to each other. Under such conditions, the accuracy of the prediction equation increases.
The purpose of multiple regression analysis is to create a “prediction equation” based upon
the relationship between high level (interval or ratio) or dichotomous independent variables.
When using dichotomous independent variables, they are referred to as “dummy variables”,
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
146146
and coded numerically as a 0 (its absence) or a 1 (its presence) of the variable (as illustrated
in the gender variable in the example below).
Example: A Marketing Manager believes that the “sales volume” of her workforce is a function
of 1) Experience 2) Education and 3) Gender. The manager is doing salary planning and
forecasting, is interested in more closely linking pay with performance in her organization.
The manager makes some enquiries and finds that the overall average sales volume per month
per salesperson is $25,000. She also discovers that there is a $1,500 increase in sales for
each year of experience, and a $500 per month increase in sales for each year of “formal”
education. She also discovers that women tend to outsell men by $450.00 per month.
The manager wants to predict future job performance of 2 candidates for a salesperson
position in the company. What is the expected or predicted monthly sales performance for
each of the two candidates?
Candidate 1: has 8 years work experience, a 4 year college degree, and is a female.
Candidate 2: has 5 years work experience, a 2 year associate degree, and is a male.
AXA Global
Graduate Program
Find out more and apply
http://s.bookboon.com/AXA
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
147
(Note: education was measured in years of formal education – for Candidate 1: 12 years
through high school, + 4 years of college, and for Candidate 2: 12 years of high school +
2 years of college)
Y= a + (b1 * x1) + (b2 * x2) + (bn * xn)
Candidate 1:
Y= 25,000 + (1500 * 8) + ( 500 * 3) + (450 * 1)
Y= 25,000 + 12,000 + 1,500 + 450
Y= $38,950
Candidate 2:
Y= 25,000 + (1500 * 5) + (500 * 4) + (450 * 0)
Y= 25,000 + 7,500 + 2000 + 0
Y= $34,500
This simplified example reveals how regression analysis works. Today, all the data would
be collected and entered in a statistical software program (such as IBM SPSS), and all the
necessary calculations would be performed. This tedious and lengthy arithmetic process
would be instantaneous. As the researcher, your job is to select and define the independent
variables, quantify them, and enter them into the statistical software and interpret its outcome.
If the company illustrated in the example above had 300 employees on the salesforce, the
researcher would need to collect the data on all 3 of the independent variables (experience,
education and gender), for all 300 people in the salesforce in order to create the “prediction
equation” for future performance of the salesforce. Other independent variables could have
been included. Notice that “age” is not an independent variable included in this example.
Should it have been? If “age” can be substantiated as a significant variable, it could have
been included in formulating the regression equation.
Once the regression equation is formulated, the researcher would then insert the values for
the independent variables of a specific case, and using the prediction equation, generate
a “predicted” outcome for the specific case. As shown above, Candidate 1 would have a
predicted sales performance of $38,950 / month, and Candidate 2 would have a predicted
sales performance of $34,500 / month. The researcher’s role is to identify and measure the
independent variables, which are related to the outcome (dependent) variable.
Variations of multiple regression (e.g., stepwise regression, hierarchical regression, etc.) allow
a researcher to include many independent variables in the data set. The statistical program
will then determine which of the independent variables have the least multi-collinearity
among each other, but the greatest relationship to the dependent variable, and then generate
the most accurate prediction equation possible given the data. For additional details about
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
148
regression the book Understanding Regression Analysis: An introductory Guide, 2nd edition
(Schroeder, Sjoquist & Stephan, 2017) provides an excellent primer on the subject.
Another prediction methodology, similar to multiple regression analysis is multiple discriminant
analysis. The major difference between regression analysis and discriminant analysis is that
when using discriminant analysis, the outcome variable (the dependent variable) is nominal or
ordinal level. In multiple regression analysis, the dependent variable is interval or ratio level.
11.6 MEASURES COMMONLY USED TO EVALUATE
FORECASTS & PREDICTIONS
11.6.1 STATISTICAL MEASURES
Comparing forecast and actual figures per time period will give a data series which can be
averaged to give mean error of the forecast. However, positive and negative errors will tend
to cancel each other, so mean error is likely to be quite small. It should, however, tell us
of systematic forecasting error.
Mean Absolute Error is computed the same way but taking all differences between actual
and forecast as positive. Mean Squared Error squares each difference and produces a similar
clearer picture of the error in forecasting, than the Mean Error.
A more useable error can be calculated through Percentage Error (PE), where each value of
the difference between actual and forecast is divided by the actual value, giving a percentage
error value. From these PEs, a Mean Percentage Error can be calculated, which is a useful
meaningful estimate of error provided there is a meaningful origin to the scale used and
the time series does not contain zeros.
The multiple correlation coefficient (R) is used to determine the correlation between the actual
value (Y') of the dependent variable and the predicted value (Y) of the dependent variable
in a multiple regression analysis. The Pearson Product Moment Correlation Coefficient (r)
is most frequently used statistic to calculate this “coefficient of multiple correlation” between
the actual and predicted values.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
149149
11.6.2 OUT-OF-SAMPLE ACCURACY MEASUREMENT
This method simply divides the data set in two and uses part of it to estimate parameters
and set up the forecast model. This is then tested on the second part of the data for accuracy
(referred to as the in-sample or training set. The test data can then be used to determine
how well the model will actually forecast a new set of data.
11.6.3 COMPARING FORECAST METHODS
The naïve method of making forecasts uses the more recent data as the prediction, (or
doing the same but with seasonally adjusted data) and computes the Mean Absolute Error
and Mean Absolute Percentage Error of these naïve predictions when compared with actual
data. The naïve method is used to compare the results of more sophisticated forecasting
techniques to actual results.
��������
�����
������
��������
���������
�����������
��������
�����
��
���������������������������
������
������
������������������������������ ����� �����
�
����� ������!���"�
������������
�
�
�����#$%����&'())%�*+������
�
���,���������
�-
.�
��������������������
������������
��
��������
��
����
�� ��������
��
����
���������
���
����������
����������
�����
��
���������
The Wake
the only emission we want to leave behind
http://www.mandieselturbo.com
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
150
11.6.4 THEIL’S U-STATISTIC
The Makridakis text (1998) suggested offers a good description of U-statistic. This is
essentially a coefficient. Note that the value of the U-statistic suggests the accuracy of a
forecast as follows:
U=1 then the naïve forecast is as good as the forecasting technique being evaluated
U<1 then the forecasting technique is better than the naïve method
U>1 then the naïve method is better than the forecasting technique.
11.6.5 AUTOCORRELATION FUNCTION (ACF) OF FORECAST ERROR
ACF is used to identify any pattern in errors after a forecasting model has been used. You
can calculate the autocorrelation function to see if there is a pattern of error which could
be avoided. Again, the Makridakis text (1998) provides a good explanation of detail.
11.7 EXPLORING THE VALUE OF FORECASTING
METHODS IN BUSINESS PRACTICE
Statistical methods of forecasting are not nearly as widely used in business as we might expect.
Moving average and exponential smoothing, plus simple linear and multiple regression
analysis, are the most widely known methods of quantitative technique for forecasting.
While time series methods are generally found to be more accurate in prediction than
explanatory models e.g. using regression, it is the latter which are seen by managers to be
the most effective technique.
Makridakis, et al conclude that:
• Simple methods for forecasting are at least as good as complex statistical methods
• Some methods are better for short time horizons than others.
• Different methods vary in accuracy depending on the method of evaluating accuracy
• Averaging of forecasts using more than one method results in more accurate predictions.
• Short term predictions can take advantage of inertia in business phenomena and
use this with seasonality and cyclical patterns to make useful forecasts
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS foreCasting trends
151
• Medium term predications are likely to be affected greatly by economic and
environmental changes, so may vary in effectiveness depending on assumptions
about the direction and speed of these changes.
• Long term predictions will decrease the effectiveness of statistical modelling at the
business level and the use of creative technique may be the way forward here.
• Key advice for improvements in forecasting includes the keeping of accurate
records – without these we have only intuition.
11.8 QUESTIONS FOR SELF REVIEW
1. What are the three main approaches for forecasting in business?
2. What is the Delphi method?
3. Why is it useful to smooth data values?
4. What different naïve methods of forecasting can you suggest?
5. Describe two ways in which forecasting accuracy can be evaluated.
6. What is the difference between simple and multiple regression analysis?
11.9 REFERENCES
Schroeder, L.D., Sjoquist, D.L., & Stephan, P.E. 2017, Understanding regression analysis: An
introductory guide, 2nd edn. Sage Publications, Thousand Oaks, CA.
Makridakis, S. Wheelwright, S.C. & Hyndman, R.J. 1998, Forecasting: methods and
applications, 3rd edn. John Wiley and Sons, Hoboken, NJ.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
152152
12 REPORTING RESEARCH RESULTS
12.1 CHAPTER OVERVIEW
12.1.1 LEARNING OUTCOMES
By the end of this topic successful students will be able to:
1. Identify a personal approach to writing a research report
2. Understand the differences between writing a report for a business audience and
for academic purposes
3. Produce a clearly structured written report
4. Produce an oral presentation of key findings
5. Recognize different methods of presenting and disseminating research.
http://s.bookboon.com/elearningforkids
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
153
12.2 YOUR PERSONAL APPROACH TO
WRITING A RESEARCH REPORT
Which parts of a research study appeal to you most?
1. Exploring and defining a research problem.
2. Reading and reviewing the literature.
3. Designing the research methods.
4. Conducting the research.
5. Analysing data.
6. Writing the research report.
Did anyone answer “6”? Probably not!
For most of us it is other parts of the research process, which appeal most, yet without
stage 6, no-one else will ever reap the benefit of our work. It is a fundamental principle of
research that we must publicise and disseminate what we find in some way, and that way
usually involves writing reports.
One of the big issues with writing reports is that we leave it until near the end, believing it
to be a simple part of the work, which can be sorted at the last minute, before a deadline.
Of course, we are too intelligent really to believe this, but this is how we behave. One good
answer to this problem is to plan. Not a rough idea of stages, which gets lost during the
research and quickly becomes meaningless, but a proper Gantt chart of activities, showing:
1. how long we expect each activity to take,
2. which, if any, are dependent on the completion of other activities,
3. what resources are needed for each activity,
4. any help needed from others and
5. by when each activity will be completed.
Our best students, who complete detailed Gantt charts, are the ones who give us a new
updated copy showing actions achieved at every supervisor meeting, who include those
supervisor meetings as milestones in the chart, and who plan to start writing way ahead
of deadlines. As supervisors, this works well, as it allows us to decide how comfortable we
are with their level of writing, and enables us to make vital improvement suggestions at an
early stage if they are necessary.
In the book “How to get a PhD” by Phillips and Pugh (2015) they suggest that writing
is the only time when we really think (pp. 98–102). Do you agree with this statement?
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
154
We are not sure we do, but we do know that writing and thinking are inextricably bound
together, so if no other thoughts are happening, start writing.
12.2.1 HOW YOU WRITE
Do you have particular rituals and routines to get yourself started on writing? Most of us
do. There is no one right way, we are all different. Some people just write, others need to
collect everything they need together first. Others start with a coffee or sit in a particular
place to write. What about you?
12.2.2 WHEN YOU WRITE
Do you need to set chunks of time aside? It is rare to be able to write in a sustained way
in short pieces of time such as half an hour. Most of us need at least an hour or two to
think ourselves into the piece of writing and make some progress. It can then take another
half hour of immersion each time you start after a break. It is also important to make sure
you are physically and mentally fresh to write. Whether you write best at night or in the
day, there will need to be some energy and sustained focus, which usually only comes when
you are in good form. The alternative is to wait until pure adrenalin forces you to write at
the last minute, when there is no option. Not a great idea, especially for a piece of work
like a research report, where, if you are fresh, new and often valuable ideas will come to
you as you write.
12.2.3 TOOLS TO HELP YOU WRITE
Simply understanding everything you can do with your word-processing software is a good
start. That particularly includes using heading styles in Microsoft Word™, since this saves
time if you have to produce a table of contents for your work. If you haven’t used this
feature, check it now before you have to write your report.
Other tools include those which can help you get your ideas together, such as mind-mapping
software (e.g. Freemind). If you haven’t come across these pieces of software, don’t worry,
they are not essential. They are a great aid if you regularly have to produce written work
of some length and like the creative approach of mind-mapping. However, it is also very
easy to produce a mind map on paper!
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
155155
Finally, specific citation software programs can be a great help if you intend to research on
a regular basis and need to build a good retrievable archive of references. Packages include
Refworks, Endnote and Zotero – all the packages mentioned can be found on the web. A
citation software package is simply a tailored database, which prompts you to record all
the relevant details of a reference as you enter it, has space for notes on your reading of it,
and can automatically work alongside Word to insert text references and an automatically
generated bibliography. All you have to do is choose the format. Student versions of these
packages are available. If you do not want to go this far, think about how you will keep your
references in a retrievable format to save time when writing up. Chapter 2 presented additional
details on how to keep track of references and citations using a variety of style manuals.
12.3 THE DIFFERENCES BETWEEN WRITING A REPORT FOR A
BUSINESS AUDIENCE AND FOR ACADEMIC PURPOSES
12.3.1 BUSINESS REPORTS
Here there is a need for clarity, brevity and simplicity. Be sure to include an executive
summary which focuses on the problem and suggested action. Sometimes there will be a
http://s.bookboon.com/academictransfer
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
156
corporate house style to adopt. Bullet points are acceptable, but all recommendations should
have a case or rationale made for them. Include charts, data tables or any visual graphic
which can help the reader grasp quickly any statistical data.
12.3.2 ACADEMIC REPORTS
There is still a need for a clear, succinct style but using appropriate terminology, for example
on research methodology, which will not be everyday language. Avoid using description
wherever possible and instead take a critical analytical approach (discussed in the chapter
on literature reviews). Pay special attention to academic referencing and avoid plagiarism.
Read and use peer-reviewed academic journal articles to set the tone of academic writing.
Find a good academic journal article and consider trying to rate the suitability of the
writing, the sections used and the persuasiveness of the article. Remember that in articles
there are very strict and short word limits, which means we rarely see a detailed research
method description, except where this is the point of the article, and we also rarely see a full
literature review, which would be expected in work written for formal academic assignments
or research reports for funding bodies, masters’ theses and doctoral dissertations.
12.3.3 ELEMENTS OF AN ACADEMIC RESEARCH REPORT:
The organization of your research report is of utmost importance. But do not overlook
a well-designed cover page (and table of content for lengthier works). And while it may
seem simple, having a succinct descriptive and interesting title is worth spending some
time preparing. You want to capture your audience’s attention from the start, and this is
where it begins.
The typical organizational format for a research paper includes the following parts:
Abstract, – written last as this must include a flavour of results, don’t repeat phrases from
the main text. If we don’t get the reader’s interest in the short abstract, they are unlikely
to read the rest of the report.
Introduction, must immediately grab the reader’s attention, often by a dramatic statement
of the problem or situation to be researched.
Background, usually starts with a broad picture and gradually refines it to the narrow focus
of the research (a filter).
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
157
Literature review, see the earlier chapter on this subject
Research objective(s), hypothesis and method justification, – most of this book has been
about this section, but it must not appear as a stand-alone section. Every section including
this one should follow logically from the previous one and lead naturally to the next. So, for
example, the literature review section should end with a direction for the primary research,
which is then picked up in the research method section.
Findings, try to offer the findings of your research in as pure a form as possible. This doesn’t
mean giving raw data, it means finding a way to present that data so the characteristics of
the data are clear to the reader, without interpreting the data, so that the reader is dependent
on your view and cannot see the data for themselves. Visual methods such as charts and
tables can summarise and present data effectively, but not pages and pages of them which
soon cause overload.
Discussion and analysis, this is the real test of your ability to synthesise what you found
in the literature review and in your primary research and to pull out from that synthesis
what seem to you to be the most important points. It is not a place to put any description.
Writing should be clear but intense – all sentences must add value.
Conclusions, not just a summary of what you found and have already said in the analysis,
the conclusions section should step back a little and take an objective view of the outcomes –
theoretical and practical – from the whole project – there should be no new references at
this stage, but a clearly persuasive account of what has been achieved
Recommendations. – may be detailed and practical or may simply urge further research in
an area which has been uncovered by your research. Where practical suggestions are made,
they must be feasible, not “blue sky” ideas. Preferably there should be suggestions about
how they could be taken forward – sometimes with a tabular implementation plan.
Appendices. In an academic piece of work, the appendices are not there to gain extra marks.
They are there for two possible reasons: a) to add information to the main text where
word length or focus did not allow their inclusion or b) to maintain a complete record of
relevant information, particularly for your future use of this document. Keep appendices
to a minimum.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
158158
12.3.4 STYLE AND GRAMMAR
This is important whether you are writing in your first or a second language. In both cases
it will be wise to ask someone you trust to sub-edit your text. None of us is our own best
editor, as many errors can easily slip through. If you are submitting an initial draft section
to a supervisor, then errors are not so critical, but they must not be at a level which obscures
the meaning!
If you are concerned about points of pronunciation and grammatical style, the best place
to check is an English language national newspaper style sheet. These are available online
at the paper’s website e.g. The Times, or The Telegraph in UK. These are often better than
out of date grammar textbooks, as they incorporate current changes of accepted style, but
do not lead change, reflecting acceptable style in the world of the reader.
12.3.5 BULLET POINTS
This is a key issue for academic work in the 21st century, as students increasingly find bullet
points acceptable, and modern business favours the use of bullet points to encapsulate an
argument quickly and clearly (in Microsoft Powerpoint™ style). There is nothing wrong with
http://thecvagency.co.uk
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
159
using bullet points in business reports, they can often cut wordy paragraphs and get straight
to the point. However, in academic work it is usual to avoid them if possible, using them
only when giving a list of examples which require no further explanation, or summarising
the points which are then explained in more detail below. Why? Because an academic reader,
specifically a marker of academic work, cannot tell from a bullet point whether you have
understood something or merely copied out a list.
12.3.6 USE OF FIRST PERSON
Whether you use the word “I thought or I did…” in your academic writing will vary
according to the purpose of the section of writing. However, the general rule is not to use
the first person except in two specific cases: first in a reflective section, where it is entirely
legitimate to speak in the first person about your learning and experience, and second, in
narrative accounts or certain types of qualitative data analysis, where this is a usual convention.
In all other cases, it is best to write objectively from the standpoint of a third person,
provided you don’t have to tie yourself up in knots stylistically to achieve this!
12.3.7 A FEW MORE WORDS OF WISDOM
A common issue in academic writing is the use of verb tenses, as much of your writing
may be taking place as things happen, results come in etc., thus encouraging you to use
the present tense. However, as a general rule, it is better to use a consistent past tense as
you are writing up a report of something, which has happened. Again, certain types of
qualitative writing will demand a current tense, and of course quotations and transcripts
should reflect exactly what was said, however it is usual to spend some time converting text
to a past tense so that it reads consistently.
Lengthy sentences and paragraphs can get in the way of meaning. Try to ensure that sentences
introduce only one idea, and paragraphs group around one idea, rather than letting them
include many, which makes it harder for the reader to understand.
Subheadings can also help to break up long areas of text on a page and should be used
where sensible. Most importantly, your academic writing is for a particular purpose: to
persuade the reader of your ideas, which requires an engaging, clear and organised style.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
160
12.3.8 LOGICAL STRUCTURE OF RESEARCH REPORTS
A final point: the logic of your written report. For any audience, logical argument and flow
from one section to another is vital. In an academic research report, it can be helpful to
draft an audit of how specific findings in your research relate to particular literature and
particular ideas, which then feature in your conclusions. In this way, all conclusions should
be traceable back to the findings they came from and a logical flow established.
If you are not regularly used to writing such research reports or dissertations, then consider
logic this way: in a really good piece of fiction writing, the reader is led along by always
wanting to know what happens next. How can you apply this to your research report? The
introduction should cause the reader to understand why you looked at the literature, what
problem you wanted to solve or question you wanted to answer. When we read the literature
review, we find out what that told you, but are left understanding that the literature didn’t
fully answer all your questions, or perhaps raised new ones. We find this out in the conclusion
of the literature review and are left wanting to know how you are going to answer those
remaining questions. So we read on to the research method, in which you tell us why you
chose this particular way of finding answers to your research questions, and then, in the
findings, what you actually found as the answers.
But that isn’t enough. We are often left at the end of the findings section thinking –“how
did that relate to what we heard about in the literature?” So we want to read on to the
discussion to find out. By the end of the discussion, we know what you found and how
it stacked up with the literature, but we are tempted to say “so what?” You answer this in
the conclusion and recommendations by explaining what that means for the big questions
you raised in your introduction, and what else remains to be done if there are unanswered
questions which your research triggered.
All this means that each section concludes with a “cliff-hanger” – an unresolved question
or problem which makes the reader want to read more in the next section. Putting in
conclusions like this to each section, helps the reader to see the logic of your work.
12.3.9 LOGIC AS A “U” SHAPE.
You may also think about a “U” shape pattern to this structure, where the introduction and
context begins at high level with “big picture” issues – maybe about the economy or the
state of an industry sector. As you proceed through your research report, you drill down
into more detail, so that by the time we read the findings, we are reading very detailed
information in a particular context of your research, which you have found at the bottom
of the hole you have dug to find out more about your question.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
161161
Then you start to take us back to the surface as you relate this detailed set of findings to the
published literature, climbing back up eventually to conclusions at “big picture” level. Now
we can see the whole problem again, but now we have your original primary research to add
to our knowledge about that problem, and guide us where to go next in further research.
12.4 PRODUCING AN ORAL PRESENTATION OF KEY FINDINGS
In business, it will be usual practice to give an oral presentation of a report, possibly using
the report itself as a “leave-behind” for readers to follow-up their undoubted interest in
your subject! If using Microsoft Powerpoint™ software to present the gist of your ideas, then
it is simple to produce clear and professional-looking slides for projection which set out
the background, your objectives, your understanding of the context, your method(s) and
your results, together with next steps/recommendations for action. Remember that when
presenting orally, we must speak directly to the audience and encourage their involvement.
At the least this will involve a pause at some point for questions, but, for preference, time
will be designed in to get some audience participation at an earlier stage. Unless you are
very familiar with the audience, it is good practice to ask something early on, which tells
you a little about their experience of the topic, so that you can involve them in your talk.
http://s.bookboon.com/Subscrybe
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
162
Any presentation will be enhanced by visual aids rather than endless bullet-style slides.
Writing the bullets can be helpful for us to remember what we want to get across, but
the actual presentation may keep the bullets only for us, and for summary use, and focus
on simple and dramatic visuals (photos for preference) which relate to your research, the
problem or the outcome. Presenting to an in-company audience means not only a house-
style (often branded slides) but also using your researcher’s objectivity to add depth to a
focused corporate message. This is quite different from an academic presentation, which
will use your objectivity to show your academic credibility and focus on the extent of your
knowledge of published sources as well as the research you have achieved yourself.
Whether in the academic written report, the business report or a presentation, well-selected
quotations from your research data, which reveal and give a flavour for your findings, are of
high value. Not too many, just a few to show your connection with the “real world” which
your research was conducted and how it relates to your findings and recommendations.
Whatever you do, you should not use your PowerPoint slides as a script to be read to your
audience. Having just concluded your research project, you should now be the master of
its content. Use your presentation to capture your audience’s attention and focus in on key
points you want to get across to them.
To avoid the concern that slidesets are used constantly by everyone, you could try a different
form of presentation such as Prezi, which offers a more visual canvas for your ideas. Or
consider a PechaKucha style of presentation. This uses a slideset but on automatic timing,
you are allowed 20 slides lasting 20 seconds each. This is an excellent way of ensuring you
don’t spend too long on one idea, but it does need plenty of practice (in real time). Make
sure you don’t start to apologise if the automatic timing runs faster than you do and you
run out of time – professional presenting requires rehearsal to get it right first time.
We have mentioned social media sparsely in this book, but here is where it can come into
its own. Professional use of social networks can be a great way to make an impact with your
research and create a following. Social media platforms regularly change and mutate, but
at the time of writing, it would be wise not to over-use Facebook to try to make serious
professional connections – that is not its purpose. However, LinkedIn currently offers a
more professional network for summarizing your research and promoting your key ideas.
Blogging and micro-blogging (Twitter) can also produce great connections for research.
You may want to use a filtering technique, such as Lists in Twitter to make sure you can
regularly check professional contacts and posts concerning research without having to wade
through lots of personal chat. Bear in mind that social media may not simply be a way of
disseminating your research, but could be part of your methodology, particularly if using
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS rePorting researCh results
163
snowball sampling or convenience sampling to distribute questions. Make sure the limitations
of this technique are clearly stated.
Whatever platforms you choose, make your research report work for you. You began with a
big idea and have had to work hard to make it into good research based on valid evidence.
Now apply it, use it and promote it to those interested – perhaps your participants/
respondents, perhaps an organization who might value your results. See whether your
academic supervisor is prepared to co-author a research conference paper on your work,
using their connections to support your scholarship. Or if you produced this research for
your organization, see who in senior management might value your results. Your business
research is only valuable if people get to hear about it.
12.5 QUESTIONS FOR SELF REVIEW
1. Why do we have to write research reports and present our results?
2. What are the key differences between writing a business report and writing an
academic report?
3. How should a research report and presentation be organized?
4. What should be included in a research method section?
5. How long should a PowerPoint™ presentation be when delivering the results of
your research study to an academic audience? Why?
12.6 REFERENCES
Phillips, E.M. & Pugh, D.S. 2015, How to get a PhD: a Handbook for students and their
supervisors, 6th edn. Open University Press, McGraw edn. Berkshire, UK.
Saunders, M. Lewis, P. & Thornhill, A. 2016, Research methods for business students. 7th edn.
Pearson Education Limited, Harlow, England.
AN INTRODUCTION TO BUSINESS
RESEARCH METHODS Comments from Peer reviewer
164
COMMENTS FROM PEER REVIEWER
This book makes a great introduction to university students of all business disciplines who
undertake research projects. It covers all the important issues and concepts which are essential
to developing competent knowledge and skills for research. Its writing is concise but at the
same time comprehensive, making this text a highly accessible resource for students who
find research methods a mystery.
– Kiefer Lee, Principal Lecturer in Marketing,
Sheffield Business School, Sheffield Hallam University.
Preface
1 �Research problems and questions and how they relate to debates in Research Methods
1.1 Chapter Overview
1.2 Introduction
1.3 The nature of business research
1.4 �What kind of business problems might need a research study?
1.5 �What are the key issues in research methods we need to understand?
1.6 Questions for self review
1.7 References
2 �Putting the problem into context: identifying and critically reviewing relevant literature
2.1 Chapter Overview
2.2 How does literature relate to research?
2.3 What kind of literature should we search for?
2.4 Effective literature searching
2.5 Critical analysis of literature
2.6 Using Harvard referencing style
2.7 Questions for self review
2.8 References
3 �Choosing research approaches and strategies
3.1 Chapter overview
3.2 �Different perspectives of knowledge and research which underpin research design
3.3 Identify differing research paradigms for business
3.4 �Key differences between qualitative and quantitative research methods and how and why they may be mixed
3.5 �Criteria of validity and reliability in the context of business research
3.6 Your choice of research strategy or design
3.7 Classification of research
3.8 The Business Research Process
3.9 The Academic business research process
3.10 Questions for self review
3.11 References
4 Ethics in business research
4.1 Chapter Overview
4.2 �Understand how ethical issues arise in business research at every stage
4.3 �Ethical criteria used in Higher Education business research studies
4.4 �Strategies to ensure ethical issues in business research are addressed appropriately
4.5 Plagiarism
4.6 Questions for self review
4.7 References
5 �Choosing samples from populations
5.1 Chapter Overview
5.2 �Understand how and why sampling relates to business research
5.3 �Identify and use a range of probability and non-probability sampling techniques
5.4 Selecting the size of your sample
5.5 �Understand and assess representativeness of samples and generalisability from samples
5.6 Sampling simulation exercise
5.7 Questions for self review
5.8 References
6 �Quantitative research methods: collecting and analysing data
6.1 Chapter Overview
6.2 �Anticipating how the research design is affected by data collection and analysis tools
6.3 Recognising different levels of data for analysis
6.4 �Coding and entering data for computerized statistical analysis
6.5 �Choosing appropriate ways to present data through charts, tables and descriptive statistics
6.6 �Selecting appropriate statistical tools for the research variables
6.7 Families of Statistics
6.8 �Measures of Correlation – the correlation coefficient
6.9 Regression analysis
6.10 Statistical significance
6.11 Questions for self review
6.12 References
7 �Questionnaire design and testing
7.1 Chapter overview
7.2 �Appreciate and overcome the difficulties associated with questionnaire design
7.3 Choosing from a range of question formats
7.4 How to design, pilot and administer questionnaires
7.5 Questions for self review
7.6 References
8 Using secondary data
8.1 Chapter Overview
8.2 The value of secondary data to business research
8.3 �What to look for as secondary data and where to find it
8.4 �The disadvantages of using secondary data in business research
8.5 Big Data
8.6 Questions for self review
8.7 References
9 �Qualitative research methods: collecting and analysing qualitative data
9.1 Chapter overview
9.2 Key issues in qualitative data analysis
9.3 �The range of qualitative research methods applicable to research topics
9.4 How qualitative data can be prepared for analysis
9.5 �computer based methods for qualitative data analysis
9.6 Questions for self review
9.7 References
10 �Practical issues in conducting interviews, focus groups, participant observation
10.1 Chapter overview
10.2 �Practical considerations relating to participant observation
10.3 Practical issues relating to interviews
10.4 Practical issues relating to focus groups
10.5 Questions for self review
10.6 References
11 Forecasting trends
11.1 Chapter overview
11.2 �Why forecasting is not widely covered in the business research methods literature
11.3 Existing methodologies for forecasting
11.4 Basic forecasting tools
11.5 Regression and discriminant analysis
11.6 �Measures commonly used to evaluate forecasts & predictions
11.7 �Exploring the value of forecasting methods in business practice
11.8 Questions for self review
11.9 References
12 Reporting research results
12.1 Chapter overview
12.2 �Your personal approach to writing a research report
12.3 �The differences between writing a report for a business audience and for academic purposes
12.4 Producing an oral presentation of key findings
12.5 Questions for self review
12.6 References
Comments from peer reviewer
