Chapter 4: Research Methodology and Research Design

The preceding chapters reviewed the literature on how firms have been deriving competitive
advantages with changing times. World has now evolved into a digital economy and has
become one global market with the advent of Internet. This has changed the way businesses
and consumers transact. In this information-driven economy, literature establishes that
analytics tools are increasingly being adopted by online retailers and large enterprises to derive
competitive advantage. This chapter builds upon the research objectives stated in the earlier
chapters and describes the research methodology. The below figure 4.1 schematically presents
the structure of this chapter.

4.1 4.2 4.3 Research 4.4 Profile of 4.5 Profile of

Philosophical Methodological Design Survey Case
assumptions choice in this Companies Companies
and thesis
methodology

Various phases
and timelines
are presented

Sampling
Mixed method strategy chosen
The research is discussed Profiles of the Profiles of the
as the choice of
philosophies 85 survey 3 case
methology is
and paradigms companies are companies are
discussed and Reliability and
are disussed presented presented
justified Validity issues
are presented

Ethical issues in
research are
presented

Figure 4.1: Outline of the chapter- Research Methodology and Research Design

Section 4.1 of this chapter elaborates the philosophical underpinnings of the research. The
section focusses strongly on two opposing schools of thought namely- positivism and
interpretivism, illustrating some of the key epistemological and ontological considerations that
are important in a research.
The chapter goes on (section 4.2) to discuss methodological choice for this thesis. This
section starts by providing an overview of mixed method research and the rationale for
selecting mixed method for this study. Further, the research design section (4.3), explains the
approach adopted which was implemented using mixed methods by using the techniques of
case studies and a questionnaire-based survey. The phases and timelines of various activities
done during the research, the sampling strategy adopted, data collection methods and the
research validity and reliability are also described. This section also explains ethical issues in
research and how they were taken care of in this research.

4.1 Philosophical assumptions and methodology

4.1.1 Research Philosophies and Paradigms

The philosophy supporting any kind of research is closely knit with the concepts of ontology,
epistemology and axiology. The philosophies are closely related to the fundamentals of
paradigm with paradigmatic preferences of the researcher that helps the researcher arrive at
methodological choice.
This section starts with presenting the theories of ontology and epistemology school of
thoughts, with the nature of paradigms and takes it further to discuss paradigmatic preferences
to ontological, epistemological and other assumptions made by the author.
Finally, this section concludes by connecting these preferences to the body of SCM and
competitive advantage theories that has been developed over the years –which are the focus
areas of the literature review in Chapter 2. The paradigmatic preferences and philosophical
stand of the author is presented next that helps author answer the research questions set out in
Chapter 3.
Ontology and Epistemology

Ontology can be defined as “the branch of metaphysics concerned with the nature of being”,
as per Oxford English dictionary. According to Saunders et al. (2009, p. 110) “ontology is
concerned with the nature of reality”. Easterby-Smith et al. (2008, p. 60) take this definition
further and state “philosophical assumptions about the nature of reality”. As per Bryman and
Bell (2011, p. 19) “questions of ontology are concerned with the nature of social entities”.
Epistemology as defined by Oxford English dictionary states “the branch of philosophy
that deals with knowledge, especially about its methods, validity, and scope”. Collis and
Hussey (2009, p. 59) define epistemology from a research perspective that is concerned with
the core question: “what do we accept as valid knowledge?” As per Easterby-Smith et al. (2008,
p. 60) epistemology can be defined as “general sets of assumptions about the best ways of
inquiring into the nature of the world”.
As noted earlier, the ontological and epistemological assumptions and stances of researchers
have a major impact on the way in which research is carried out. In relation to SCM research
this is articulated effectively by New (2009, p. 89): Although both academic literature and
practitioner discourse are happy to bandy the term ‘supply chain’ around, the implicit imagery
behind the metaphor – materials flowing through the economy via some neat system of
imaginary pipes – is intrinsically problematic. The difficulties arise because of questions of
ontology (for example, what do we mean by ‘product’?) and epistemology (how can we
know?).
The following sections discuss paradigms in particular the positivist and interpretivist
positions, with specific reference to their ontological and epistemological assumptions.

Paradigms

The emphasis on the concept of a paradigm in research philosophy in recent decades is usually
attributed to the work of Thomas Kuhn in the 1960s. Kuhn (1962) discussed paradigms in the
context of “the progress of scientific discoveries in practice” and went on to define paradigms
as “universally recognised scientific achievements that for a time provide model problems and
solutions to a community of practitioners” (p. viii).
Bryman (1988) provides a useful definition of a paradigm: “a cluster of beliefs and dictates
which for scientists in a particular discipline influence what should be studied, how research
should be done and how results should be interpreted” (p. 4). Gummesson (2000, p. 18)
suggests that the concept can be used to represent “people’s value judgements, norms,
standards, frames of reference, perspectives, ideologies, myths, theories, and approved
procedures that govern their thinking and action”.
Perhaps of more interest and consequence in the context of the current research is the impact
of the “world-view” represented by the paradigmatic preference of a researcher on the conduct
of research in the social sciences generally and in business research specifically. The work of
Burrell and Morgan (1979) is central to this discussion. In their text Sociological Paradigms
and Organisational Analysis, they applied the notion of a paradigm to the social sciences and
described four distinct paradigms (see figure 4.2).

Sociology of Radical Change

Radical Humanism Radical structuralism

Subjective Objective
Interpretive Sociology Functional Sociology

Sociology of Regulation

Figure 4.2: Four Paradigms of Social Theory

The Burrell and Morgan Framework (1979, p.23)

The vertical axis in the above classification makes the assumptions about the nature of
society and the horizontal axis makes assumptions about the nature of social science. From the
perspective of research in business and management, the researchers are more interested in the
horizontal axis that helps them position themselves along the subjective objective continuum.
While the continuum is important to recognize, the debates and discussions between followers
to the extreme ends of the spectrum has its own concerns with the relative merits of objectivist
(positivist) and subjectivist (interpretivist) positions
In the view of Tashakkori and Teddlie (1998), it is more suitable for the researcher in a
study to consider and think of the adopted philosophy as a continuum rather than opposite
positions. Morgan and Smircich (1980, p. 492) take this ahead and come up with a six-stage
continuum between “pure” positivism and “pure” interpretivism. Although it is useful to
understand the characteristics of the purist positions to illustrate the different types of world-
views represented by each. These notions of ontology and epistemology are inherent in these
world-views.

Philosophies of research: positivism, interpretivism and pragmatism

Let us first understand the different research paradigms, before we decide on the
methodological choices and issues, as it is important to get an overview of philosophical
perspectives as they help shape the way researches conduct research and impact the data
collection process, analysis of data and interpretation of results.
A paradigm can be defined as a world-view core to the process of research in all areas
(Saunders et al. 2009). It is primarily a set of basic beliefs that defines the nature of the world
and the place that and individual holds in the world and it also guides individual’s action (Guba
and Lincoln 1994; Denzin and Lincoln 2000).
Research methods are grounded in philosophical traditions in a discipline that stem
from the prevailing paradigm, defined as a “basic set of beliefs that guides action” (Guba, 1990,
p. 17). The philosophical paradigm of the researcher heavily influences the methods chosen to
address research questions (Frankel et al., 2005). Kuhn’s (1962) influential book, “The
Structure of Scientific Revolutions”, ignited the paradigm wars of the late twentieth century in
the social and behavioural sciences.
One of the main issues of empirical scientific research is concerned with the appropriate
choice of research strategies as the key to further knowledge creation. As such, the use of
research methodologies is controversially debated among scholars, also affecting the research
in supply chain management (Creswell, 2013, Frankel et al., 2005, Mentzer and Kahn, 1995,
Meredith et al., 1989). As ontological as well as epistemological considerations play a vital
role in this discussion, the presentation of research strategies is outlined based on these
grounds.
According to Frankel et al. (2005) the individuals’ world view, the ontological position,
sometimes referred to as Weltanschauung (Checkland, 1993), determines the starting point of
every research endeavour. Two extreme ontological views can be differentiated from a
philosophy perspective: The objectivist view portrays the understanding that “social
phenomena and their meaning have an existence that is independent of social actors” (Bryman
and Bell, 2011, p. 19).
Following the antithetical constructionist position, the “social phenomena and their
meanings are continually being accomplished by social actors” (Bryman and Bell, 2011, p.
20). This implies that social actors are the drivers behind social phenomena. Social reality can
thus not be regarded as being definite as it is subject to constant change, triggered by
individual’s influence on the environment.
Following extant theory (Bryman and Bell, 2011) the ontological framework inevitably
influences the cohesive selection of research strategies, ultimately governing the author’s
decisions in regard to the choice of the epistemological paradigm. Thus, from an
epistemological perspective, the debate about research strategies is influenced by two
distinctive scientific paradigms, namely positivism and interpretivism (Bryman and Bell, 2011,
Burgess et al., 2006). Scholars who adapt the positivist view aim to imitate the natural sciences
(Bryman and Bell, 2011), driven by the point of view that reality can be understood (Bhaskar,
2008). Accordingly, as seen by Mentzer and Kahn (1995), positivist researchers pursue the
“goal to explain and predict reality, where reality is considered to be objective, tangible, and
fragmentable” (p. 232). The opposing interpretive view helps to “understand a phenomenon
but not to explain or predict” as it depicts a “collective of multiple socially constructed realities
[where] people [are] proactive and voluntaristic” (Mentzer and Kahn, 1995, p. 232). In a
logistics and SCM research context it has been found that the scholars’ research focus gradually
shifted from mainly following the positivist paradigm (Mentzer and Kahn, 1995) towards a
more balanced use of the interpretivist and positivist paradigms (Golicic et al., 2005). This
research study belongs primarily to the field of competitive advantage and supply chain
management, in which there has been two commonly used research paradigms or philosophies,
namely positivism and interpretivism (Hussey and Hussey, 1997; Collis and Hussey, 2009).
The below table 4.1 shows the alleged differences between the two paradigms.

Table 4.1: Positivism vs. Interpretivism

 Source: Weber (2004), Hussey and Hussey (1997) and Creswell (2009)

Positivism necessitates an ontology that the reality and researcher are separate and the
reality is made up of atomistic, discrete and observable events, whereas interpretivism requires
that researcher and reality are inseparable and reality is internal & socially constructed (Weber
2004). The interpretative paradigm is viewed as qualitative, inductive and subjectivist, while
the positivist paradigm is described as quantitative, deductive and objectivist (Burrell and
Morgan 1979; Hussey and Hussey 1997; Denzin and Lincoln 2000; Gummesson 2000).
Positivism and interpretivism are often known as the opposing ontological and epistemological
perspectives of research (Weber 2004).

Positivism

Positivism, goes back to its origins in the philosophy known as realism. This was developed
by theorists that included Auguste Comte (1798-1857), John Stuart Mill (1806-1873) and
Emile Durkheim (1859-1917). It is positioned at the objective end of the paradigmatic
continuum shown in figure 4.2 (above).
Positivism originates from the term “positive” that means progressive which indicates
positive and progressive development of the field of social sciences. Taking the ontological
stand, it is believed by positivists that there is one social reality this is external to the researcher.
On the other hand, epistemological stand maintains an objective and independent attitude as
they believe the only observable and measurable phenomena can be validly regarded as
knowledge (Collis and Hussey, 2009). Therefore, positivists, are inclined to place a great
emphasis on the use of quantitative methods to deductively test hypotheses in an objective
manner.
The primary characteristic that positivism portrays is to solve key practical problems
and determine through statistical analysis any existence of causal relationships (Kim 2003).
Per Hussey and Hussey (1997), real world related phenomenon or knowledge should be viewed
objectively independent, externally and singularly. Positivists assert that with the ontological
assumption, scientific investigation and deductive reasoning will both converge upon objective
truths (Plack, 2005). Blaikie (1993) states, on taking epistemology assumptions, positivists
assert that knowledge is an outcome of sensory experiences as result of experimental or
comparative analysis.
Positivism places a large emphasis on scientific objectivity (Dupuis, 1999). As per
Dupuis (1999), to obtain objective and accurate data, there is a need to separate subject from
object, and anything attached personally to the researcher such as selves, personal experiences
and emotions should be removed. The use of objective measures, quantitative data and methods
in the social world for theory testing is emphasized. Nevertheless, this method also has faced
criticism by researchers as it also comes with many potential pitfalls associated in its adoption
of this position related to human issues. The pitfalls in the positivist approach led to the
emergence of interpretivism, subjectivism etc. The new set of paradigms is referred by multiple
names, such as, interpretivism, phenomenology and subjectivism (Gummesson 2000).
Positivism has played a great influencer in the progress and development of research
traditions in the areas of natural and social sciences (Plack 2005). Positivism research belongs
to the category of social sciences that aims to gain insights from the social dimensions and
management behaviour using data driven decision making. In social science research, it is
commonly applied to generate more accurate, measurable and objective data (Burrell and
Morgan, 1979).

Interpretivism

To challenge traditional positivist paradigmatic position, the interpretivist paradigm was

developed in the social sciences. It is positioned at the subjective end of the paradigmatic
continuum as shown in figure 4.2 (above).

The origin of interpretivism dates to the intellectual traditions of hermeneutics and

phenomenology (Kim 2003). As per Bryman and Burgess (1999), interpretivism can be defined
as a strategy of social research that helps interpret social phenomena in terms of meanings.
Thereby, the whole emphasis is on the social actors’ own language, experiences and
perceptions and not on the measurement and prediction of phenomena, but (Lee, 1991).
However, in the most recent edition (Collis and Hussey, 2009) they use the word
interpretivism to “suggest a broader philosophical perspective” (p. 57). For similar reasons, the
word interpretivism is used in this context throughout this thesis.
In ontological terms, interpretivists believe that social reality is subjective because it is socially
constructed. For this reason, the phrase “social constructionism” is often used to describe this
paradigmatic position – the world is an emergent social process. Thus, there are multiple
realities as each person has his or her own sense of reality (Collis and Hussey, 2009, p. 59).
Knowledge in an interpretivist’s epistemology is a resultant outcome from everyday
experiences, concepts and meanings (Blaikie 1993). Interpretivists aim to understand what and
how about any phenomenon. Research states that in social science research, important and
context specific factors, and personal experiences and emotions of researcher, that are often
not considered and easily ignored, must be taken into consideration, in keeping with inter-
subjectively created meanings of the social world (Lee 1991; Dupuis 1999; Kim 2003). In
effect, interpretivists, have placed a strong emphasis on the usage of qualitative methods to
reason and build theory in a subjective way. Here, the whole emphasis is on the development
and overall understanding of phenomena, rather than deriving possible causal relationships
between various measures.
Therefore, to explore the social world and obtain qualitative data, researchers have preferred
qualitative methods and theory building/generation over theory verification. There has always
been opposing thoughts on the positivist approach and the interpretive approach, as the
positivists claim that approaches of natural science are only regarded as scientific ones, on the
contrary interpretivists claim that the research of organizations, institutions and people
demands methods that are overall different likes of natural science (Lee 1991).

A New Paradigm war on Positionality

At the either end of the paradigm continuum, there is either “Pure” positivism and “pure”
interpretivism that are regarded as extreme positions. There has been a widespread debate over
the years of giving relative merits for each position and the extent of impact of each
positionality on the research process (Herr and Anderson, 2005).
Robson (2002) refers to this debate as “so-called ‘paradigm wars’ endemic” (p. 43). Mangan
et al. (2004) note that “the various paradigmatic positions are now often discussed in terms of
an antithesis between two schools of philosophy” (p. 566). Similar points are made by several
authors, notably Gummesson (2000) and Hussey and Hussey (1997). The forthright views of
Byrne (1998) in relation to the positivist position in social sciences research is revealing:
“Positivism is dead. By now it has gone off and is beginning to smell.” (Byrne, 1998, p.37).
However, the distinguishing characteristics of each position is worth considering in
consideration with ontological, epistemological and other possible and relevant issues. These
characteristics are shown in table 4.2 below, which is compiled from various sources (Easterby-
Smith et al. (2008), Saunders et al. (2009), Bryman and Bell (2011), Robson (2002), Denzin
and Lincoln (2000), Collis and Hussey (2009).
 Table 4.2: Characteristics of Positivism and Interpretivism

Sources: (Easterby-Smith et al. (2008), Saunders et al. (2009), Bryman and Bell (2011),
Robson (2002), Denzin and Lincoln (2000), Collis and Hussey (2009))

When discussing ontological assumptions, the position of positivist is objectivist in that

reality is observed separately from the researcher. On the contrary, the interpretivist position is
viewed subjectively by the researcher. It is positioned at the left-hand end of the matrix of
Burrell and Morgan (1979) as shown in figure 4.2. In terms of epistemological assumptions,
the position of a positivist is characterized by independence. Thereby, the researcher is
independent of what is being researched. On the contrary, the interpretivist position assumption
identifies the inter-relationship and interaction between researcher and the researched. To
elaborate further, the researcher (observer) is a part of what is being researched (observed).
Quantitative researchers most often work from the positivist paradigm or the post
positivist paradigm. Research conducted from positivism is expected to be objective, free of
values, hypothesis driven, and measurable. Positivists use deductive reasoning and seek to find
causes that precede, or occur at the same time as, effects. The post positivist paradigm has
replaced positivism (Schwandt, 1997) or follows positivism as “the (current) predominant
philosophy for (quantitative) research in the human sciences” (Teddlie and Tashakkori, 2009,
p.69). Research consistent with post positivism is influenced by researchers ‘values and their
chosen theory or conceptual framework’. According to the post positivist paradigm, facts
cannot necessarily prove a theory and determine a cause. Reality is socially constructed, and
internal and external validity are both important.
 Qualitative researchers work mostly from the constructivist (or interpretivist) paradigm,
which supports the notion that there are many realities that are constructed as the research
engages with participants. Realities are constructed by participants and researchers who seek
to understand participants ‘points of view’. Observations of reality are influenced by
researcher’s values. Multiple realities exist, and our understanding of these realities is
constructed individually and socially. Constructivists believe that determining a connection
between cause and effect is impossible; therefore, description of reality is important.
Qualitative researchers engage in inductive reasoning as they work from units of data toward
a theory, or as they work from the specific or particular to the general. Statements about reality
are limited to the time and context of the study, so generalizability is limited to transferability
of results from one context to another (Teddlie and Tashakkori, 2009)
Philosophical differences between positivist/post positivist and constructivist
paradigms contributed to tension, or “paradigm wars” (Teddlie and Tashakkori, 1998, p.3),
between qualitative and quantitative researchers. “Qualitative researchers stress the socially
constructed nature of reality, the intimate relationship between the researcher and… emphasize
the value-laden nature of inquiry… [Qualitative researchers note that] quantitative studies
emphasize the measurement and analysis of causal relationships between variables, not
process…within a value-free framework.” (Denzin and Lincoln, 2005, p.10)
Mixed methods researchers accept the idea of compatibility between qualitative and
quantitative methods taking the pragmatist point of view (Howe, 1998). The researchers instead
of requiring to choose between qualitative or quantitative methods focus on determining how
both qualitative and quantitative methods will help answer the research questions. Both
inductive and deductive reasoning logic are used, and if required hypotheses may be proposed
as well.
For pragmatists, internal validity and credibility play an important role. When it comes
to generalization of findings, pragmatists place a lot of emphasis on external validity and
credibility, in combination taking into consideration that hypotheses belong to time and context
(Teddlie and Tashakkori, 2009)

Towards a Multi-Paradigmatic Position?

On the relative merits of two “pure” positions, according to Bryman and Bell (2011), “The
point being made here is that quantitative and qualitative research represent different research
strategies and that each carries with it striking differences in terms of the role of theory,
epistemological issues and ontological concerns. However, the distinction is not a hard-and-
fast one: studies that have the broad characteristics of one may have a characteristic of the
other” (p. 26). Similarly, as per Guba and Lincoln (1994, p.105) “both qualitative and
quantitative methods may be used appropriately with any research paradigm” and as per
Easterby-Smith et al. (2008) “research problems (in management) often require eclectic
designs which draw on more than one tradition” (p. 56).
Conceivably Bryman (1988) points out that in practice there is a greater reconciliation between
researchers of two schools and this is particularly true in the applied areas. SCM and
Competitive Advantage areas could be regarded as such an “applied areas” and there is enough
evidence that this “rapprochement” is found favourable by many scholars in this field.
As per research, positivism belongs to the branch of philosophy known as realism1.
Robson (2002) came up with an approach to “real-world” known as critical realism. Per him
“realism permits a new integration of what are usually referred to as subjectivist and objectivist
approaches” (p. 35). Robson argues that the real-world research needs to retain the advantages
of positivism by ensuring that a “scientific attitude”. Therefore, in this context a “scientific
attitude” requires that research execution involves systematic, sceptic and ethical approaches.
Pragmatism

As per Creswell (2009), pragmatism does not bind itself to either of the above-mentioned
philosophies and views of reality. Pragmatists are of the view that pragmatism is neither false
under interpretivist nor positivist philosophies. Per pragmatists, it is absolutely possible for a
researcher to work with both philosophies (Creswell 2009; Saunders et al. 2009). Further,
pragmatists claim that mixed methods, that includes both qualitative and quantitative methods,
are possible and highly suitable within one research to provide more comprehensive evidence
and strength than doing research with one method (Blaikie, 2009). The compatibility thesis has
also been argued from the viewpoint of paradigm relativism, which has been called “pragmatic
relativism” or “pragmatism” (Howe, 1988; Tashakkori and Teddlie, 1998). Many mixed
methods scholars argue for some version of pragmatism as the philosophical foundation for
mixed methods research. Pragmatists adopt “whatever philosophical and/or methodological
approach works for the particular research problem under study” (Tashakkori and Teddlie,
1998, p. 5). The basic belief of the pragmatic paradigm is that the research problem dictates

1
Many labels exist, including scientific realism, critical realism, fallibilistic realism, subtle realism and
transcendental realism.
the research method. Hence, pragmatism rejects the incompatibility thesis and claims that
research paradigms may either remain separate or be combined in new research approaches.
As stated by several authors, the researcher’s philosophical stance is totally dependent
upon the nature and objectives of the research. As per Saunders et al., (200), it is important to
arrive at clearly defined research objectives before getting into research methodology, as the
background of research objectives will determine the precise research methodology to be
followed. Therefore, to address the research objectives and questions as stated in Chapter 3
the researcher needed to collect both qualitative data (that involves more depth and insightful
exploration with limited number of interviewees) and quantitative data (that can be collected
from broader samples to generalize). This goes to show that the researcher follows pragmatism
and therefore the philosophy of pragmatism fortifies the researcher’s choice of mixed methods
in this research. As explained earlier in Chapter 1 (introduction) and Chapter 2 (literature
review), there has been little research on comparison between traditional and Internet based
businesses in the way they use information and analytics for competitive advantage.
This research aims to understand this phenomenon and achieve this by answering the following
questions.
Research Question 1: What is the extent of use of analytics as a source of competitive
advantage by online intermediaries from theory?
Research Question 2: What is the extent of use of analytics as a source of competitive
advantage by traditional businesses in the following areas?
Research Question 2.1: What is the extent of use of descriptive analytics by means of data
visualization techniques by traditional businesses in the following areas?
 Interactive visual analysis of customer purchasing
 Interactive visual analysis of product sales and profitability by various channels
Research Question 2.2: Do traditional businesses use predictive analytics in the following
areas of competitive advantage?
 Customer segmentation for targeted marketing and discounts using Clustering
Techniques;
 Dynamic Pricing;
 Association Rules recommendations, cross-selling and up-selling;
 Use Forecasting Models to predict superstar products and profitable customers;
 Compute Customer Lifetime Value (CLV),
The above-mentioned research questions demand the use of research methods and tools that
are likely to yield useful data, both qualitative and quantitative, to achieve the set research
objectives. This demands employing a suitable research design that comprises of several
methodological considerations including research philosophies, research method and research
design.
4.2 The Methodological Choice in this Thesis

4.2.1 Mixed Method research as a Methodology

The Definition

This study uses the definition provided by Creswell and Plano (2007); that mixed methods
research focuses on the collection, analysis and mixing of both quantitative and qualitative data
in a single study or a series of studies (Molina-Azorin, 2009). Mixed method research dates as
far back as 1959, when Campbell and Fisk used multimethod to study the validity of
psychological traits and encouraged others to use it to examine multiple approaches to data
collection. It was born out of the belief that all methods have limitations and the view that
biases from any single method could neutralize or cancel the biases of other methods which
further herald triangulation of data sources as a means of convergence, connection or embedded
to the integration of quantitative and qualitative data. This has led writers worldwide to develop
procedures for mixed strategies that take numerous forms (Creswell, 2004). Mixed method
research recognizes that both quantitative and qualitative research methods are useful and
important and does not seek to replace either but rather draw from the strengths of each, to
minimize the weaknesses of both in single research studies and across mixed studies. Other
proponents of mixed methods research include works of Brewer and Hunter (1989); Newman
and Benz (1998) and Reichardt and Rallis (1994), in (Johnson and Onwuegbuzie, 2004).
A mixed-method research design was adopted in this research to investigate if
traditional businesses use analytics- descriptive and predictive to gain competitive advantage,
as their counterparts in web-based businesses and online intermediaries do.
Mixed methods research is further defined as the type of research in which a researcher,
or a team of researchers, integrates qualitative and quantitative research approaches within a
single study or a set of closely related studies (Creswell and Plano Clark, 2007; Johnson et al.,
2007). The roots of mixed methods research design can be traced to the introduction of the
multitrait-multimethod (MTMM) matrix, which was developed to assure that observed
variance is attributable to the trait under study, rather than the method (Campbell and Fiske,
1959). Subsequently, the term “triangulation” was introduced to describe investigations that
combine multiple data sources and multiple methods (Denzin, 1978). Jick (1979) further
explicates the concept of triangulation by describing “within-method triangulation” (i.e.
multiple quantitative or multiple qualitative methods) and “across-methods triangulation” (i.e.
combining qualitative and quantitative methods).
 Qualitative data tends to be open-ended without pre- determined responses while
quantitative data usually includes closed-ended responses such as found on questionnaires or
psychological instruments. Qualitative research is related with the using of words to provide a
descriptive meaning and understanding of experiences and situations of individuals which are
part of the phenomena being investigated (Denzin and Lincoln, 2000). Quantitative research
focuses on numerical data which can provide an overview of relationships, patterns and trends
(Zikmund, 2003). The next section discusses in detail the advantage of using a mixed-method
research design which is that it enhances the examination of multiple arrays of data that can
impart added rigor, breadth and depth to the research (Saule, 2002) and why mixed methods
research, as a methodology, is appropriate for the current study.
Ultimately, the possibility of mixing methods can be very large because of the many
potential classification dimensions, as it opens an exciting and almost unlimited potential for
future research (Johnson and Onwuegbuzie, 2004). Mixed method research strategies or
designs can be developed from either a mixed-model (mixing qualitative and quantitative
approaches within or across the stages of the research process) or mixed-method (including a
quantitative phase and a qualitative phase in an overall research study). Johnson and
Onwuegbuzie, (2004) provided a detailed list of across and within mixed-method designs and
the different kinds of mixed-methods designs based on Morse (1991). However, researchers
should consider two primary decisions; firstly, whether to operate within one dominant
paradigm or whether to conduct the phases concurrently, sequentially (Johnson and
Onwuegbuzie, 2004) or transformatively (Creswell, 2004).
Secondly, in a mixed methods design, the findings must be mixed or integrated at some point.
Nonetheless, one can easily create more user specific and more complex designs. For example,
quantitative-qualitative-quantitative stages or both a mixed-model and mixed methods
approach. Furthermore, a design may emerge during a study in new ways, depending on the
conditions and information that is obtained. The most important point is for the researcher to
create designs that effectively answer their research questions. This view contrasts with the
approach where one completely follows either a quantitative or qualitative paradigm (Johnson
and Onwuegbuzie, 2004).
Mixed methods research (Johnson et al., 2007: 123) is: “the type of research in which
a researcher or team of researchers combines elements of qualitative and quantitative research
approaches (e.g., use of qualitative and quantitative viewpoints, data collection, analysis,
inference techniques) for the broad purposes of breadth and depth of understanding and
corroboration”. This research methodology is argued to be intellectual and practical, as it is
likely to take the advantage of overcoming the weaknesses in singular methods (Johnson and
Onwuegbuzie, 2004), and to provide the most informative, complete, balanced, and useful
research results (Johnson et al., 2007).

Figure 4.3: Mixed Methods Research

(Source: Johnson and Onwuegbuzie (2004)

As explained by Tashakkori and Teddlie (1998), there has been growing interest in
research methodologies such as mixed-method and mixed-model studies in many fields.
Nevertheless, there has been a lack of conceptual clarity and not enough examples of the mixed-
methods being used (Tashakkori & Teddlie, 1998). As described by Tashakkori and Teddlie
(1998), a few authors (for example Creswell, 1994; Miller & Crabtree, 1994; Morse, 1991)
have attempted to develop taxonomies of mixed-methods. Tashakkori and Teddlie (1998)
describe mixed-methods as a combination of qualitative approaches (the
constructivists/phenomenological) and quantitative approaches (the positivists and post-
positivists) in the methodology of a study (such as in the data collection stage). Tashakkori and
Teddlie (1998) mentioned mixed-model studies as the combination of these two approaches
across all phases of the research process (such as conceptualisation, data collection, data
analysis, and inference). However, they stated that, although the growth of the mixed-method
or mixed-model has been retarded by the vestiges of the paradigm wars, researchers are now
free to use the methods in their research questions. In fact, the research questions are best
answered with mixed-method or mixed-model research designs, rather than with a sole reliance
on either the quantitative or the qualitative approaches.
A multi-methodological approach to study systems development (Ovaska, 2009);
analysing the use of information systems in logistic industry (Pokharel, 2009); and applied
multi-case research in a mixed-method research project (Jansen, 2009) were conducted in the
field of Information Systems.

4.2.2 Mixed Method used in this study

“By combining multiple observers, theories, methods, and empirical materials, researchers can
hope to overcome the weakness or intrinsic biases and the problems that come from single-
method, single-observer, and single-theory studies. Often the purpose of triangulation in
specific contexts is to obtain confirmation of findings through convergence of different
perspectives. The point at which the perspectives converge is seen to represent reality." (Jakob,
Alexander, 2001).

Figure 4.4: Mixed Method used in this Research

(Source: Author)

This study adopts the mixed method approach- triangulation design to answer the research
questions as shown in the figure 4.3 above. The most common and well-known approach to
mixing methods is the Triangulation Design (Creswell, Plano Clark, et al., 2003). The purpose
of this design is “to obtain different but complementary data on the same topic” (Morse, 1991,
p. 122) to best understand the research problem. The intent in using this design is to bring
together the differing strengths and non-overlapping weaknesses of quantitative methods (large
sample size, trends, generalization) with those of qualitative methods (small N, details, in
depth) (Patton, 1990). This design and its purpose of converging different methods has been
discussed extensively in the literature (e.g., Jick, 1979; Brewer & Hunter, 1989; Greene et al.,
1989; Morse, 1991). This design is used when a researcher wants to directly compare
quantitative statistical results with qualitative findings or to validate or expand quantitative
results with qualitative data.
Triangulation Design Procedures. The Triangulation Design is a one-phase design in
which researchers implement the quantitative and qualitative methods during the same
timeframe and with equal weight (see Figure 4.5). The single-phase timing of this design is the
reason it has also been referred to as the “concurrent triangulation design” (Creswell, Plano
Clark, et al.,2003). It generally involves the concurrent, but separate, collection and analysis of
quantitative and qualitative data so that the researcher may best understand the research
problem. The researcher attempts to merge the two data sets, typically by bringing the separate
results together in the interpretation or by transforming data to facilitate integrating the two
data types during the analysis. Jenkins’ (2001) single-phase study of rural adolescent
perceptions of alcohol and other drug resistance is an example of a Triangulation Design. She
collected and analyzed quantitative and qualitative data and merged the two data sets into one
overall interpretation, in which she related the quantitative results to the qualitative findings.

Variants of the Triangulation Design

The four variants are the convergence model, the data transformation model, the validating
quantitative data model, and the multilevel model. The first two models differ in terms of how
the researcher attempts to merge the two data types (either during interpretation or during
analysis), the third model is used to enhance findings from a survey, and the fourth is used to
investigate different levels of analysis.
The fourth variant is the Triangulation Design is what Tashakkori and Teddlie (1998)
referred to as “multilevel research” (p. 48). In a multilevel model, different methods
(quantitative and qualitative) are used to address different levels within a system. The findings
from each level are merged together into one overall interpretation. For example, Elliott and
Williams (2002) studied an employee counselling service using qualitative data at the client
level, qualitative data at the counsellor level, qualitative data with the directors, and
quantitative data for the organizational level.
 Figure 4.5: Four Variants of Triangulation Design
Source: Creswell (2006)

Strengths of the Triangulation Design.

'Triangulation' comes with several important opportunities for researchers. Firstly, it allows
researchers to be more self-assured of their results. This can help play more positive role such
as, it can excite formation of innovative methods and newer ways of defining problems that
strikes a balance with predictable data-collection methods. This may also help uncover the
unknown dimensions to a phenomenon. By its comprehensiveness, this can also serve as the
critical test for opposing theories. 'Triangulation' helps reduce the deficiencies of single-source
research. Two methods (qualitative and quantitative) complement and verify one another,
which reduces the influence of any inherent bias.
 This design has several strengths and advantages, including the following:
 The design makes intuitive sense. Researchers new to mixed methods often choose this
design. It was the design first discussed in the literature (Jick, 1979), and it has become
a framework for thinking about mixed methods research.
 It is an efficient design, in which both types of data are collected during one phase of
the research at roughly the same time.
 Each type of data can be collected and analysed separately and independently, using
the techniques traditionally associated with each data type. This lends itself to team
research, in which the team can include individuals with both quantitative and
qualitative expertise.

An Assessment of Challenges of 'Triangulation'

The 'triangulation' strategy comes with its own drawbacks. It starts with the premise that if the
research is not clearly focused theoretically or practically, it will not deliver a satisfactory
outcome. There are arguments that each method should be represented separately in a
significant way.
To address the challenges in combining the two paradigms in the same study, Morse
(1991) suggests likely two ways in which both quantitative and qualitative methods can be
triangulated. Firstly, a qualitative method used as initial investigation in a quantitative study;
in which, qualitative methods are regarded as an add-on method. Secondly, quantitative
methods go before as preliminary investigation in a qualitative study in the sense that
quantitative methods are regarded as supplementary methods. In principle, wherever a
researcher uses both qualitative and quantitative methods in the same research project, it is
assumed in advance that the researcher has clarity of the main ontological and epistemological
position of the phenomenon that is being investigated (Denzin and Lincoln ,1994). As
mentioned earlier in this thesis, both methods in qualitative and quantitative come with their
own strengths and weaknesses, therefore research recommends focusing on the within-method
and between-method type of 'triangulation'. Hence, a combination of the two approaches have
been encouraged to gain from two paradigms and thereby curtailing the drawbacks of each.
In summary, ‘triangulation' is possible and a good way to reap the benefits of both
qualitative and quantitative methods. This can cut across the qualitative and quantitative divide.
The use of 'triangulation', however, will depend on the researcher's philosophical position. It is
not aimed merely at validation but at deepening and widening one's understanding. It tends to
support interdisciplinary research rather than restricted within social sciences. In fine,
'triangulation' can, indeed, increase credibility of scientific knowledge by improving both
internal consistency and generalizability through combining both quantitative and qualitative
methods in the same study. However, effective 'triangulation' depends on coordination and
collaboration; particularly those who are actively involved in collecting data and response.

4.2.3 Rationale for selecting Mixed Method for this research

There are two major types of research methods that are quantitative and qualitative. Qualitative
research is in alignment with the positivist paradigm, whereas qualitative research is in close
alignment with the naturalistic paradigm (see table 4.3). In the context of quantitative research,
there exists a more formal, objective and deductive approach to solving a problem. Whereas,
in qualitative research the approach to problem solving is more informal, subjective and
inductive. The characteristics of both approaches are presented in the table below:
 Table 4.3: Comparison of positivist and naturalistic paradigms (Quantitative and
Quantitative approaches)

Source: Keele (2011)

Although quantitative research is viewed as the more rigorous of the two (qualitative and
quantitative) in the past, qualitative research has ended up gaining more credibility in the world
of science in the recent times. Both the methods are considered suitable for research such that
each method contribute significantly to the scientific body of knowledge. Choice of the right
method depends basically on the research questions(s) inquired. These questions stem from the
research objective and purpose/problem statement.
Factor- isolating questions ask, “What is this?” These questions name and describe factors or
variables of interest to the researcher. “What is the extent of use of analytics in your
organization to serve your customers better?” would be included in this category of questions.
The most appropriate research design to answer these questions would be descriptive.
Descriptive studies are designed to gain more information about characteristics of a topic of
interest. Descriptive level research is most appropriate when very little research is available on
the top. A case study design further helps in-depth understanding of a phenomenon through
detailed information.
The advantage to descriptive level research is that the researcher can collect a large amount
of data, however, even though there is breadth of data, it tends to lack depth for the sample. On
the other hand, case study research provides depth and richness of data, but lacks breadth since
it is limited to one person or event. Hence, this study employs a triangulation of both the
methods to answer the research questions effectively.

4.3 Research Design

A research design characterizes a plan and the procedures for conducting an investigation based
on the nature of the research problem and issues that are being addressed in combination with
the researcher’s personal experience (Creswell 2009).

Figure 4.6: The Research ‘Onion’

Source: Saunders et al., (2009)

The research methodology helps the readers understand the process followed by the
researcher in choosing the appropriate method to conduct the research. It echoes the overall
process that constitutes the research philosophy, research approach, research strategy, data
collection methods and data analysis and ensure that they are consistent (Saunders et al.
2009). The “Onion” approach proposed by Saunders et al. (2009), depicted in the above
figure 4.6, was used as a guiding principle behind the research design.

4.3.1 The Research Approach Adopted for this study

In the context of Mixed Methods research approach in SCM, Gammelgaard and Larson, (2001)
conclude that the method of mixing a survey and case study permitted them to apply
quantitative measures as well as probe more deeply into the context to develop a richer
understanding of the phenomenon.

Figure 4.7: Research approach adopted for this study

Source: Author

In the social sciences, the notion of triangulation can be traced back to Campbell and
Fiske’s (1959) use of multiple quantitative methods for assessing convergent and divergent
validity. Denzin (1978), however, distinguished such within-method triangulation, or the
application of multiple data collection and/or analysis techniques within an overarching
research method (e.g., the survey), from between-method triangulation, representing a more
genuine synthesis between dissimilar methods (e.g., case study and survey methods). The
emphasis of the present paper falls within the latter, arguably more rewarding but also complex
and challenging form of method triangulation (Denzin, 1978; Jick, 1979). The rationale for
between-method triangulation is that it offers some advantages in dealing with validity threats
stemming from the biases inherent in any single method. By combining relatively diverging
methods, such that the relative strengths of one counter-balance the weaknesses of the other(s),
researchers may enhance the credibility of their results whilst reducing the risk of observations
reflecting some unique method artefact (Denzin, 1978).
Triangulation between case study and survey methods arguably provides a relatively
potent means of assessing the degree of convergence as well as elaborating on divergences
between results obtained (Brewer and Hunter, 1989; Jick, 1979; Sieber, 1973). For example,
surveys may improve our understanding of the incidence of a particular phenomenon and/or
the form and strength of conceptual relationships observed in case studies. On the other hand,
case study methods may add to a more holistic and richer contextual understanding of survey
results and help to explain apparent anomalies or puzzles emerging from the latter.

4.3.2 Sampling Strategy and Data Collection

The Sampling Strategy

Mixed Methods sampling requires an understanding and acknowledgement of sampling
strategies that occur in QUAN and QUAL research. Probability sampling techniques are used
most often in QUAN research to obtain a sample that most accurately represents the entire
population. Purposive sampling techniques are used mainly in QUAL research to select
participants or other units of study who can provide or yield data that will address the research
questions. Although convenience sampling is sometimes used in QUAL and QUAN research,
it includes samples that are the most available to the researcher; these may not be representative
of the population being studied and may yield biased data. Because techniques for mixed
methods include choosing participants for a study using both probability and purposive
sampling, a comparison of purposive sampling techniques is presented in the table 4.4 below:
 Table 4.4: Comparison between purposive and probability sampling techniques

Source: Teddlie and Tashakkori (2008, p.179)

Mixed methods sampling includes characteristics of both purposive and probability

sampling. Combining sampling techniques for QUAL and QUAN methods are usually made
before the study begins. In the absence of an established classification or typology for mixed
methods sampling strategies, Teddlie and Tashakkori (2009) discussed strategies for sampling
and mixed method designs from the perspectives of probability and purposive sampling.
One of the basic mixed method sampling technique is stratified purposive sampling.
This involves identifying subgroups in a population and then selecting cases (participants) from
each subgroup and compare and contrast across the subgroups. Purposive random sampling
involves selecting a random sample of a small number of units (participants) from a larger
population (Kemper, Stringfield and Teddlie, 2003). Random selection of this sample reflects
probability sampling, and the smaller number of participants selected reflects purposive
sampling.
 Figure 4.8: Characteristics of Mixed Method Sampling
Source: Teddlie (2005)

Zone A consists of totally qualitative (QUAL) research with purposive sampling,

whereas Zone E consists of totally quantitative (QUAN) research with probability sampling.
Zone B represents primarily QUAL research, with some QUAN components. Zone D
represents primarily QUAN research with some QUAL components. Zone C represents totally
integrated mixed methods (MM) research and sampling. The arrow represents the purposive-
mixed-probability sampling continuum. Movement toward the middle of continuum indicates
a greater integration of research methods and sampling. Movement away from the centre (and
toward either end) indicates that research methods and sampling (QUAN and QUAL) are more
separated or distinct (Teddlie, 2005)

Figure 4.9: A quick snapshot of difference between quantitative vs.

quantitative sampling

In sequential mixed method sampling, researchers select units of analysis (e.g.,

participants) by using probability and purposive sampling strategies, one after another. That is,
probability sampling for the QUAN phase is followed by purposive sampling for the QUAL
phase is followed by purposive sampling for the QUAL phase or vice versa. Whereas in parallel
mixed methods as in the case of this research, probability and purposive sampling strategies
are used concurrently or with a slight time lapse between each phase. A probability sampling
is used to produce data for the QUAN phase and purposive sampling produces data for the
QUAL phase. These two sampling procedures are used to generate separates sets of data.
Parallel mixed methods sampling can also occur when the participants are selected using both
probability and purposive sampling (Teddlie and Tashakkori, 2009).
The sampling technique used in this study is illustrated in the figure 4.10 below.
As defined in Chapter 2 (Literature Review), the traditional businesses in India are those
that sell their products through a multi-tier distribution network to the end-consumer (Mulky,
2013). In this study, the population of interest were the traditional businesses in India that
manufactured and sold consumer products through the conventional distribution network. As
the quantitative part of this study involved descriptive research design, the sample traditional
businesses in the consumer goods segment, were randomly selected from the industry segments
as mentioned below. Therefore, stratified random sampling was used for this descriptive study
part of this research (see figure 4.10) that helped sample companies from the below mentioned
industry segments (strata).
1. Apparel
2. Bags and Luggage
3. Beauty and Health
4. Consumer Electronics
5. Food Products and
6. Footwear
On the other hand, a criterion sampling method was adopted for the qualitative case studies.
Three case companies were selected based on the criterion:
a) The companies must be producing and selling consumer products like apparel, food
products, footwear, bags, consumer electronics, cosmetics etc.
b) The products are already sold or are saleable through online retail channels.
 QUAN QUAL

Figure 4.10: Sampling technique used for this research

Data Collection
As this research uses mixed methods, the data collection involved survey questionnaire as
instrument for quantitative study and face-to-face interviews, examination of existing business
applications, process documents and existing data as an instrument for qualitative study.
The first step towards development of questionnaire for both the methods (QUAN and QUAL)
involved identification of two themes as mentioned in section 3.1 of chapter 3. The broad
themes were then further broken down into sub and sub-sub themes that formed the basis of
questionnaire development for both quantitative and qualitative study of this research.
Questionnaire Preparation
After identification of the themes, sub-themes and sub-sub themes, a draft closed ended
structured questionnaire for survey with categorical responses using nominal and ordinal
scales, was prepared to answer the research questions. The next step was pre-testing the survey
instrument to certify its validity and to identify changes needed prior to administering the final
version of the questionnaire.

Pilot Study
Pre-testing is an essential part of the questionnaire development process (e.g., Cavana et al,
2001; Malhotra et al., 2004). It helps the researcher to identify and remove any potential
problems prior to administering the survey (Burns & Bush, 2003; Cavana et al., 2001; Neuman,
2006). As there are usually several limitations to each of the pre-test methods, the use of
different combinations of approaches has been recommended (e.g., Malhotra, 2004; Churchill,
1995).
The questionnaire used in this study was pilot tested. The purpose of the pilot study was
two-fold: a) To examine and verify the appropriateness of the questionnaire; and b) to ascertain
the readability and appropriateness of the survey questions.
In this research, an expert panel and interviews were used to pre-test the questionnaire
to get over the shortcoming of using one method. The questionnaire was first assessed and
reviewed by the researcher’s supervisors to reach an agreement. Next, the instrument was pre-
tested through interviewing senior leadership team members of three potential respondent
companies and in addition, a small field test with 25 companies was also conducted. The
interviews with the senior leaders of potential respondent companies lasted for about 1.5 to 2
hours each. Post this step, some minor changes were made which included, making some of
the questions optional as not all companies will be willing to share the information. Since there
were no adverse comments received from the pilot study participants, the resultant
questionnaire was finalized for the main study.
In addition to the survey questionnaire, semi-structured open ended interview questions
around the identified themes and sub-themes were prepared to help conduct case studies at
three selected case companies. The questionnaire used in this study is presented in Appendix-
A.

The Data Collection Process

The following figure 4.11, illustrates the data-collection process employed in this study.
 Figure 4.11: An overview of data collection process

The Profile of Survey Companies

The closed-ended, structured questionnaire that had categorical responses was sent to an initial
sample size of 200 companies. The questionnaire was administered through email, google
forms and personal meetings with managers of the selected sample companies. Of the 200
companies, 85 responses were useful for progressing with this research and the rest 115 were
invalid and could not be used. The following figure 4.12 depicts the number of useful
responses received by industry segments.

Figure 4.12: No. of useful responses received by industry segment

The table 4.5 below presents an industry wise summary of the number of companies to which
questionnaire was administered versus number of responses received.
 Table 4.5: Industry-wise summary of number of companies (questionnaire sent to)
Industry Segment Approximate Sales No. of No. of No. of Useful
Turnover of the Questionnaire Invalid Useful Responses
Company (in INR Sent to Responses Responses as % of
Crores) Received Received total
Apparel a) less than 50 crores 2 1 1 50%
Apparel b) 50- 100 crores 6 4 2 33%
Apparel c) 101- 500 crores 7 5 2 29%
Apparel d) 501 -1000 crores 5 3 2 40%
e) greater than 1000
Apparel 5 3 2
crores 40%

Bags and Luggage a) less than 50 crores 3 3 0 0%

Bags and Luggage b) 50- 100 crores 6 6 0 0%
Bags and Luggage c) 101- 500 crores 7 2 5 71%
Bags and Luggage d) 501 -1000 crores 7 4 3 43%
e) greater than 1000
Bags and Luggage 6 3 3
crores 50%
Beauty and Health a) less than 50 crores 2 1 1 50%
Beauty and Health b) 50- 100 crores 2 2 0 0%
Beauty and Health c) 101- 500 crores 10 5 5 50%
Beauty and Health d) 501 -1000 crores 9 4 5 56%
e) greater than 1000
Beauty and Health 5 3 2
crores 40%
Consumer Electronics a) less than 50 crores 0 0 0 -
Consumer Electronics b) 50- 100 crores 3 3 0 0%
Consumer Electronics c) 101- 500 crores 10 6 4 40%
Consumer Electronics d) 501 -1000 crores 15 5 10 67%
e) greater than 1000
Consumer Electronics 13 3 10
crores 77%
Food Products a) less than 50 crores 0 0 0 -
Food Products b) 50- 100 crores 8 6 2 25%
Food Products c) 101- 500 crores 8 2 6 75%
Food Products d) 501 -1000 crores 6 4 2 33%
e) greater than 1000
Food Products 8 4 4
crores 50%
Footwear a) less than 50 crores 0 0 0 -
Footwear b) 50- 100 crores 12 12 0 0%
Footwear c) 101- 500 crores 11 9 2 18%
 Industry Segment Approximate Sales No. of No. of No. of Useful
Turnover of the Questionnaire Invalid Useful Responses
Company (in INR Sent to Responses Responses as % of
Crores) Received Received total
Footwear d) 501 -1000 crores 14 6 8 57%
e) greater than 1000
Footwear 10 6 4
crores 40%
Grand Total 200 115 85 43%

The data collection process was conducted over a period of 9-10 months and the
percentage of useful responses received were 43%, which makes it a total of 85 consumer
products companies all over India.
In parallel, semi- structured, open ended- questionnaire around the research themes
aided the three-qualitative case studies process. The profiles of the both survey respondents
and qualitative case companies are presented in section 4.5 and section 4.6 respectively.
4.3.3 Reliability and Validity of Research

As noted earlier in the section 4.2.2, in social sciences, the notion of triangulation can be traced
back to Campbell and Fiske’s (1959) use of multiple quantitative methods for assessing
convergent and divergent validity. Denzin (1978), however, distinguished such within-method
triangulation, or the application of multiple data collection and/or analysis techniques within
an overarching research method (e.g., the survey), from between-method triangulation,
representing a more genuine synthesis between dissimilar methods (e.g., case study and survey
methods).
The emphasis of the present paper falls within the latter, arguably more rewarding but also
complex and challenging form of method triangulation (Denzin, 1978; Jick, 1979). The
rationale for between-method triangulation is that it offers some advantages in dealing with
validity threats stemming from the biases inherent in any single method.
As mentioned by Tashakkori and Teddlie (1998), internal validity can be
conceptualised as to the degree to which we can trust the conclusions and inferences of the
researcher regarding the causal association between variables or events. The mixed-methods
results when combining the qualitative and quantitative methods further contribute to the
validity and reliability of key issues. The correspondence of the two methods ‘case studies’ and
‘descriptive survey’ design for the two key themes- “a) Customer Knowledge measure b)
Customer Analytics Measure has been used to investigate the reality if traditional consumer
products companies’ have been using both variables. This is demonstrated using the mixed-
methods.
Creswell (2003) mentioned a mixed-methods design format, where the investigator
brings approaches that are included in both quantitative and qualitative formats related to the
purpose and the research questions. This assisted to deploy a solid methodology to collect data
from multiple sources to answer the two research questions by following the research purpose.
The mixed-methods, though challenging and involving a great amount of work, followed the
convergent design strategy (Creswell, 2003). There had been no ‘mixing’ of data from each
methodology in the initial stages, as each data item from each method was analysed separately,
presented in different chapters and later the data analysis from both methods is triangulated;
and at the interpretation stage, there was a common analysis to arrive at the results. But each
of the methodologies deployed was executed in parallel and converged at analysis and
interpretation stage. For instance, the three case studies assisted to understand the in-depth
qualitative insights into how traditional businesses are using customer data for gaining the
competitive edge and survey results of 85 traditional companies helps gain descriptive insights
into specific questions around the two key themes identified for this research.
Therefore, triangulation addresses the validity and reliability issues in research as
follows:
• Advantages of each (QUAL AND QUAN) complement the other resulting in a
stronger research design, and more valid and reliable findings.
• Inadequacies of individual methods are minimized.
• Threats to Internal Validity are realized and addressed.
• Triangulation between case study and survey methods arguably provides a
relatively potent means of assessing the degree of convergence as well as
elaborating on divergences between results obtained (Brewer and Hunter,
1989; Jick, 1979; Sieber, 1973).
• Surveys may improve our understanding of the incidence of a phenomenon
and/or the form and strength of conceptual relationships observed in case
studies.
• On the other hand, case study methods may add to a more holistic and richer
contextual understanding of survey results and help to explain apparent
anomalies or puzzles emerging from the latter.
• Quantitative design strives to control for bias so that facts, instances,
phenomena can be understood in an objective way.
 • Qualitative approach strives to understand the perspective of participants or a
situation by looking at firsthand experience to provide meaningful data.
To ensure reliability and validity, the following procedures were adopted by the
researcher in keeping with mixed methods:
 The physical presence for over 6 months representing approximately 200 hours spent
in three different case companies in the consumer products business.
 There were structured and semi-structured face to face interviews conducted with
senior leadership team with designations, CEO (Chief Executive Officer), CFO (Chief
Finance Officer), Vice President, Director, Senior Manager and Managers, who have
a wider domain (job responsibility) and administrative knowledge, implied high
reliability of responses received.
 The customer analytics solutions and models created by the researcher during the case
study process was validated by the senior leadership team of the case companies.
 The structured -closed ended survey questionnaire was run through a trial pilot phase
with the senior executives approximately 25 companies to arrive at the final validated
questionnaire for executing quantitative process.

4.3.4 Ethical Issues in research

The general ethical issues that apply to respondents at the data collection state, involve
obtaining their informed consent; keeping their identities private and their answers confidential
(Saunders, et al., 2007: 180). During the research process, researchers undertake a great
responsibility for ethical issues such as the interest of interviewees’, sensitivity and privacy
and ensure no harm is done do their physical, social and psychological well-being.
For this whole research, the author considered four key ethical issues viz., harm to
participants, lack of interviewees’ consent, invasion of privacy and deception, as identified by
Bryman and Bell (2011). The interviews were documented only after taking interviewees’
consent. In the case of survey questionnaire, the respondents were free to decide if they
intended to participate or not. The research aim and objective were explained clearly to the
interviewees, they were assured of data confidentiality and that it would be used only for this
research and not elsewhere. Confidentiality and anonymity were ensured during the research
process.

4.4 Profile of survey companies

This section presents the profile of the survey companies in the aspects, the company size (sales
revenues), the industry segments they belong to, their number of employees, the products they
sell, the profile of the channels by which their products get sold and the profiles of survey
respondents.
4.4.1 Company Size (Sales Revenues)

The size of the survey companies that provided useful responses are as below in figure 4.13.
While above 60 % of companies that responded were having a turnover of > 500 crores, the
data includes a good mix of companies of all sales turnover buckets. The responses from <100
crores companies have been very useful across all industry segments. The section 5.0 ‘Data
Analysis’ further presents the various aspects of this research from a company size perspective
as well.

Figure 4.13: Size of survey companies

Since there were fewer companies in the turnover segments ‘less than 50 crores’, ‘50-
100 crores’ and ‘101-500 crores’, therefore, the researcher clubbed them under one category
as ‘less than 500 crores’. The visual below (figure 4.14) presents them as three groups.

Figure 4.14: Size of survey companies by three turnover groups

4.4.2 Number of Employees

The data presented in the figure 4.15 goes to show that majority, 72 out of 85, i.e. about 84 %
of companies had more than 200 employees.

Figure 4.15: Number of employees in survey companies

4.4.3 Profile of Survey Respondents

The primary respondents to the survey questionnaire were as presented in table 4.6 below. To
maintain anonymity, their names are withheld.

Table 4.6: Profile of survey respondents

Industry Segment
Bags Beauty
Designation/Department and and Consumer Food
of Respondents Apparel Luggage Health Electronics Products Footwear Total
CEO 3 - 3 - 3 - 9
CFO 2 - 4 2 2 - 10
Managers and Senior 1 4 2 8 3 2 20
Managers-IT
Managers and Senior - 2 3 3 - 2 10
Managers-Marketing
Managers and Senior 3 3 1 6 - 4 17
Managers-Operations
Department
Managers and Senior - - - - - 5 5
Managers-Retail and E-
Commerce
Managers and Senior - 1 - 1 2 - 4
Managers-Sales and
Distribution
 Industry Segment
Bags Beauty
Designation/Department and and Consumer Food
of Respondents Apparel Luggage Health Electronics Products Footwear Total
Managers and Senior - - - 2 - - 2
Managers-After Sales
Managers and Senior - 1 - 2 4 1 8
Managers-Supply Chain
85

The respondents were of the job titles CEO (Chief Executive Officer), CFO (Chief
Finance Officer), Vice President, Director, Senior Manager and Manager. Since the
respondents are senior-level executives who have a wider domain (job responsibility) and
administrative knowledge, it implies high reliability of responses received.

Figure 4.16: Departments of respondents

4.4.4 Competition

When asked about the number of companies in the Indian Market that they consider as their
direct competitor, figure 4.17 presents their responses. Only 2 % of the respondents felt that
they have less than 5 companies as their competitors in India whereas the rest have greater than
5. Greater the number of competitors, nimbler must be a company in this digital age to survive
and remain profitable.
 Figure 4.17: Competitors to survey companies

Figure 4.18: Competitors to survey companies, an industry segment wide view

Another representation of the number of competitors of survey companies by industry segment

is presented above in figure 4.18. There is tremendous competition in all the industry segments
and the companies do acknowledge and realize that.

4.4.5 Sales Channels

As explained in section 2.1.3 of this thesis, the consumer products in India flow through a tiered
distribution network before it reaches the end consumer. There are multiple intermediaries that
have a role to play in the traditional distribution network viz. carrying and forwarding agents,
whole-sales, distributors, sub-distributors, retailers and mom and pop stores. In addition to
traditional distribution network, the consumer products are also sold through Modern Retail
and E-Commerce channels. While out of 85 companies that had presence in e-commerce, in
the current state only 3 of them are selling through their own e-commerce web sites, while the
rest sell their products through other online intermediaries like Amazon, Flipkart and Snapdeal.
The visualization below in figure 4.19 presents the average sales % across various
channels
 Figure 4.19: Overall average sales (in %) by channels

It can be observed that the survey companies have just started making a foray into E-Commerce
selling. E-Commerce attributes to only 12.5 % of total sales on an average. Traditional
Distribution Network and Modern Retail remain their primary channels of sales. A further
distribution of sales percentage by industry segment is also presented in the below figure 4.20.
Apparel, Bags and Luggage, Beauty and Health segments have almost similar e-commerce
sales characteristics. In fact, two Apparel companies and one Beauty and Health company were
selling their products through company owned e-commerce web sites. Food products are still
way behind in the e-commerce segment. The key point to note here is traditional distribution
still dominates the sales in most segment and modern retail is not way behind. It is now
important to investigate the number of levels of trade overall before the product reaches the
end-consumer.

Figure 4.20: Channel wise sales proportion2

2 the total for each industry does not add to 100 % in some cases due to rounding error
Tiered Distribution Network

Given traditional distribution network attributing to over 45 % of total sales (on an average), it
was important to investigate the maximum number of layers in the distribution network of the
survey companies, before their products reached the end-consumer. The consumer product
companies have many customers at various levels of trade (primary, secondary, tertiary etc.)
that act as intermediaries in the sales process until the product reaches the end consumer. Figure
4.21 below clearly shows that over 93 % of the companies sell their products at least through
three intermediaries before their product finally reaches the end consumer.

Figure 4.21: Number of layers of distribution to end-consumer

Therefore, it is difficult for most companies to get access to end consumer data as they are at
least two-three levels away from them. While this aspect of customer data sufficiency is
investigated in the Data Analysis Chapter 5.0, further in this section we look at the number of
primary, secondary and tertiary customers of the survey companies in the visualization, figure
4.22.
At the first level of trade, most of the companies had greater than 100 customers (that are
intermediaries). At level 2 and level 3 of trades, the majority of the survey companies had over
500 customers.
 Figure 4.22: Number of direct customers at levels 1, 2 and 3 of trade

4.4.6 Profile of Products Sold

The profile of products sold by survey companies are shown in the figure 4.24 below. The
number of active SKUs (Stock Keeping Units) sold by a majority of them is over 1000. In fact,
49 % of them have over 5000 active SKUs. This goes to show that they have a large product
portfolio to market and sell.
If we observe II) in the figure 4.24, 46 % of the survey companies state that about 10-
20 % of their products attributed to over 80 % of revenues. If we combine the data analysis
of I) and II), it can be inferred that the survey companies have a long tail of products that they
maintain, since only 10-30 % of their products attribute to over 80 % revenue.
The other aspect of understanding their products was the frequency of launch of new
products. It can be observed from visualization III that majority of them (48 %) launch new
products in a short period of 3-6 months. In this digital age customers have access to immense
amount of information and their wants and expectations are ever growing dynamically.
Therefore, it becomes very important for companies in this digital age to understand customer
expectations and become customer obsessed to survive and thrive.

Figure 4.23: Profile of products sold by survey companies

4.5 Profile of case companies

The profile of the three case companies that formed a part of this thesis are presented in tables
4.7, 4.8 and 4.9. This covers the sales turnovers, the number of employees, number of products
manufactured, number of active product categories, SKUs (Stock Keeping Units), channels of
sales and percentage share revenues across channels.
 Table 4.7: Sales turnover and number of employees of the case companies
Company Industry Segment Company Turnover No. of Employees
(approx.) in INR crores (approx.)
Company 1 Apparel and home décor 120 300
items
Company 2 Food Products 600 350
Company 3 Apparel 50 60

Table 4.8: Products manufactured, number of active product categories, active SKUs
Company Industry Details of Products Manufactured No. of Active No. of Active
Segment Product SKUs
Categories
(approx.)
Company Home décor, Primarily Apparel and other luxury 12 88,000
1 dining and goods in the categories home décor
Apparel and dining
Company Food Products Chips, Sauces, Spreads, Cake mixes, 40 450
2 Milk shake mixes
Company Apparel Primarily Apparel. They also sell 10 15,000
3 Travel Accessories

Table 4.9: Channels of sales and % share of business from each channel is presented
here
Company Industry Segment What are the various Percentage of business from
channels by which you sell each sales channel
your products? Traditional: Organized
Modern Retail: E-Commerce
Company 2 Apparel a) Traditional Distribution 0:85:15
Network, b) Organized
Modern Retail, c) Ecommerce
Company 2 Food Products a) Traditional Distribution 60:35:5
Network, b) Organized
Modern Retail, c) Ecommerce
Company 3 Apparel b) Organized Modern Retail, 0:60:40
c) Ecommerce