Sie sind auf Seite 1von 12

markt (2009) 48:29–40 der markt

DOI 10.1007/s12642-009-0003-1
Journal für Marketing
METHODOLOGICAL PAPER

The formalization of case study research in international business


Pervez N. Ghauri · Rebecca Firth

Received: 6 May 2008 / Accepted: 31 October 2008 / Published online: 27 March 2009
© Springer-Verlag 2009

Abstract Qualitative case study research has frequently Keywords Case studies · CAQDAS · Nvivo · Analysis ·
been criticised for being less systematic and objective than Linkages
other approaches. Ambiguity in the way case studies are
selected, empirically conducted and analysed, often exists
and can hinder the positive dissemination of research. Zusammenfassung Gegenüber qualitativen Fallstudien
This paper aims to outline how case study research can wird vielfach der Vorwurf eingebracht, diese wären zu
be undertaken using the example of the authors’ current wenig systematisch und objektiv in ihrem Ansatz. Es
case study research into interfirm services linkages between gibt eine Reihe von Unklarheiten darüber wie Fallstu-
UK multinational subsidiaries and local firms in China and dien ausgewählt werden, empirisch durchgeführt und
Korea. A particular emphasis is placed on the analysis of analysiert werden. Dies behindert die positive Aufnahme
case studies and the benefits of computer aided qualitative des Fallstudienansatzes in der Forschungsgemeinschaft
data analysis software (CAQDAS). An integrative approach sowie die Verbreitung entsprechender Forschungsarbeiten.
is taken through the use of Nvivo software. It is argued that Dieser Artikel stellt einen Strukturierungsansatz vor, wie
the use of Nvivo can help researchers systematise and order Fallstudien systematisiert werden können. Dabei wird
data, enabling a more thorough and reliable analysis. If auf eine aktuelle Untersuchung zurückgegriffen, in der
qualitative case study research is to become more accepted zwischenbetriebliche Firmenverknüpfungen („linkages“)
and publishable, its transparency must be improved. The zwischen Tochtergesellschaften multinationaler Firmen aus
use of a standardized type of data analysis software may dem Vereinigten Königreich und lokalen Firmen in China
go some way towards helping researchers achieve this und Korea betrachtet werden. Besonderes Augenmerk
goal. wird dabei auf die Vorteile der Nutzung computerbasierter
Datenanalysesoftware (CAQDAS) gelegt. Wir stellen einen
integrativen Ansatz vor und nutzen das Nvivo Software-
paket. Es wird argumentiert dass Nvivo ForscherInnen in
der Systematisierung und Analyse von Daten hilft, und
This paper draws on Ghauri PN (2004) Conducting and analysing case
studies in international business. In: Piekkari R and Welch C (eds) einen Beitrag zur Verlässlichkeit der Untersuchung liefert.
Handbook of qualitative research methods for international business. Wir sehen eine Zielsetzung darin qualitative Fallstudien
Edward Elgar, Cheltenham als ein zuverlässiges und attraktives Instrumentarium für
P. N. Ghauri (✉) Forschung und Forschungsdokumentation zu etablieren.
King’s College London, The Department of Management, Daher ist es unabdingbar die Transparenz in der Analyse
Franklin-Wilkins Building, 150 Stamford Street, und Dokumentation der Untersuchung zu erhöhen. Eine
SE1 9NH London, UK Strukturierung und Systematisierung hilft dabei dem Ziel
e-mail: pervez.ghauri@kcl.ac.uk
Tel.: +44-20-78484122 näher zu kommen.
R. Firth
Manchester Business School, Schlüsselwörter Fallstudien · CAQDAS · Nvivo · Analyse ·
Booth Street West, M15 7PB Manchester, UK Firmenverknüpfungen

13
30 P. N. Ghauri, R. Firth

Introduction plain the unique features of the case, as well as to point out
the characteristics that are common to several cases. Finally,
“Using case studies for research purposes remains one of this approach relies on the integrative powers of research:
the most challenging of all social science endeavors” (Yin the ability to study an object with many dimensions and then
2003, p. 1). This is an opinion which still holds largely true to draw the various elements together in a cohesive interpre-
today as it did at the time of the initial writing of this classic tation (Selltiz et al. 1976).
text on case studies. Even though case studies do not have This paper deals with the stages of a case study project,
to be strictly qualitative, they often involve at least some the first being the decision to use the case study method, and
qualitative work. With the growth of qualitative research the last concerning how to analyse data that has been col-
(Flick et al. 2004) and an increasing number of journals lected through cases and how to analyse them using Nvivo.
dedicated to the subject, such as “qualitative inquiry”, Examples from a recent academic project we have under-
“qualitative market research journal”, and “qualitative taken will be used to illustrate common dilemmas faced by
research”; there is a need to understand how qualitative case the practising case study researcher. The hope is that this
studies in particular can be designed and conducted. There article will form a useful introduction for those embarking
has been a call for clearer documentation of qualitative upon case study research project and inform those looking
research in order to counter criticisms such as lack of to use Nvivo from first use to advanced use. Moreover, we
objectivity, generalisability and transparency. The use of aim at suggesting various ways in which case studies can be
CAQDAS and in particular Nvivo in the analysis of qual- conducted in a more systematic way by adopting the use of
itative case studies may go some way toward addressing Nvivo or CAQDAS in general.
these criticisms (Sinkovics et al. 2008). In this article, we
firstly tackle the issues of when to use case studies and how
to conduct them and secondly we undertake an in-depth Deciding to use a case study
look at how to analyse qualitative case studies using Nvivo
with a current empirical example from recent research into Whether to use a case study or not is an important first ques-
inter-firm linkages. tion as the suitability of the research to this approach will ul-
In business studies, we normally study cases to provide timately determine its quality. As Yin (2003) has famously
insight into an issue, a management situation or new theory. said, case studies are a preferred approach when “how” or
Case studies have been used to investigate many IB topics, “why” questions are to be answered, when the researcher has
such as international business negotiations, international little control over events and when the focus is on a current
joint ventures, market entry processes and headquarter- phenomenon in a real-life context. Also influential in the
subsidiary relationships. A case study is a useful method development of case study research is Eisenhardt’s (1989,
when the area of research is relatively less known, and the pp. 548–549) argument that case studies are:
researcher is engaged in theory building (Eisenhardt 1989). Particularly well-suited to new research areas or research
This is perhaps the most frequently used approach for thesis areas for which existing theory seems inadequate. This type
and dissertation research in business at postgraduate level of work is highly complementary to incremental theory
and is often applied as a pretext to quantitative survey-based building from normal science research. The former is useful
work (Jean et al. 2008; Jean 2007). However, we will argue in early stages of research on a topic or when a fresh
that case studies are in fact a flexible research approach that perspective is needed, while the latter is useful in later
is suited to a range of different types of research questions stages of knowledge development.
(Ghauri 2004). Case studies have also been combined with
a variety of different epistemological and methodological
However, this does not mean that case study research
approaches, from positivist to phenomenological, using
is only suited to exploratory and descriptive research.
surveys to interviews. In this paper, we avoid entering into
Case studies can in fact be used in all types of research:
philosophical debates and instead maintain the stance that
exploratory, descriptive or explanatory (Bonoma 1985;
case studies can be used in both types of research.
Ghauri and Grønhaug 2002; Yin 2003).
Case studies can involve data collection through multiple
sources such as surveys, verbal reports, personal interviews,
observation and written reports (such as financial reports, When to use which research approach ultimately depends
archives, budget and operating statements including mar- upon the type of research questions being posed. When
ket and competition reports). The main feature is the depth research questions concern only “what” (for example, “what
and focus on the research object, whether it is an individ- is the impact of forming service sector linkages on host
ual, group, organisation, culture, incident or situation. We country suppliers?”), the objective of the study is to develop
need to have sufficient information to characterise and ex- hypotheses or propositions for later testing. For such an

13
The formalization of case study research in international business 31

exploratory study, a range of research strategies can be used. a follow-up study (Hyder and Ghauri 2000). As most case
If the questions relate to “how many?” or “how much?” studies are done through a review of existing historical
survey or archival strategies are favoured. For example, material and records plus interviews, the case study method
compiling a report on the population statistics or overview is quite similar to historical review, but it is different
of a particular industry, i.e. how many companies operate in the sense that here we have the possibility of direct
in the industry there, how many of these are domestic or observation and interaction. The second strength of case
foreign and what are their market shares etc. However, if studies is, as Yin (2003) has observed, their contextuality.
“how” is used on its own, for example, how do linkages in In business studies case research is particularly useful
the service sector change with host country firms over time? when the phenomenon under investigation is difficult to
Then a range of data collection methods might be used study outside its natural setting – as typically occurs in
but generally, when “how” and “why” questions are asked, international business research, since researchers are often
a case study method as a research strategy is most useful studying the impact of different national contexts. This
(Ghauri and Grønhaug 2002). means that when we wished to understand the extent of
Over the years case studies have been used for a variety linkage formation in China and Korea and what led to their
of “what”, “how” and “why” questions, covering a diverse formation amongst firms case studies were an effective way
range of topics in international business. For example: what of understanding the links between macro-environmental
measures of international competitiveness are relevant to factors, industry-level relationships and firm decision-
the performance of SMEs (Coviello et al. 1998)? How does making. A third advantage of this method is that the level of
a negotiation process in a domestic setting differs from depth with which each case is investigated allows for theory
a negotiation process in an international setting (Ghauri building, not just theory testing. Fourthly, case studies are
1983)? How do international joint venture relationships holistic, thus permitting the investigation of a phenomenon
develop over time (Hyder and Ghauri 1989, 2000)? How from a variety of viewpoints, covering a period of time, and
do headquarter-subsidiary relationships in the MNE change crossing the boundaries between different factors (Ghauri
over time (Ghauri 1992)? Or, how do companies enter 1983). Researchers can cross check data in a relatively
unfamiliar foreign markets (Ghauri and Holstius 1996; unbiased way. Concepts and variables under study are often
Hadjikani et al. 2008) difficult to quantify since there are too many variables to
The case study method is particularly well suited to be considered, which makes experiment or survey methods
international business research, where data is collected inappropriate to use (Bonoma 1985; Ghauri and Grønhaug
from cross-border and cross-cultural settings. Survey 2005; Yin 2003).
and/or experiments raise serious questions about equiva- The decision of whether to choose a case study should
lence and comparability of data collected from different not be taken lightly or seen as an “easy” or “soft” option
countries (Sinkovics et al. 2005). In spite of rather so- because as will be explained their conduction and analysis
phisticated methods of questionnaire translations and can be extremely complex. A case study may very well
cross-translations, the understanding and interpretation of involve quantitative methods or even be entirely quantita-
questions by respondents and of answers and findings by tive. Whether a case study is used as a research method
researchers are very difficult to compare and often lead to depends upon the research question, objectives and the
misleading conclusions. Often surveys and questionnaires research setting/situation. Once this approach has been
are simply thrown away because foreign respondents do chosen, we need to think of the next steps such as how to
not fully understand their purpose. The case study method select individual cases and to conduct and analyse them.
provides excellent opportunities for respondents and re- The rest of the article will deal with these issues.
searchers to check their understanding and keep on asking
questions until they obtain sufficient answers and desired
interpretations (Sinkovics et al. 2008). In-depth interviews Selecting cases
are particularly suitable when researcher wants to under-
stand behaviour of decision-makers and/or complicated How to select and decide what constitutes a case can be
topics in other countries. a puzzling issue. How to select the cases is perhaps the most
Case studies have the potential to deepen our understanding important issue in this type of research. As in other methods
of the research phenomenon, firstly because they allow of data collection, it is important to decide the target pop-
us to take a longitudinal approach. For example, in Hyder ulation that is to be used for the investigation. It includes
and Ghauri (1989, p. 26) it is argued that joint ventures are those firms, individuals, groups or elements that will be rep-
best understood and studied as a “historical process”. The resented in the study. The next stage is to assess the accessi-
researchers were able to extend the time period under study ble population, the population to which we can have access
by later returning to the same joint ventures to conduct (Cooper 1984). Out of this accessible population we have

13
32 P. N. Ghauri, R. Firth

to select one or few cases, objects or firms, for study. This to participate in an academic study. It tends to be consider-
selection should be based on criteria that are consistent with ably easier to get in touch with people who have themselves
the research problem. The cases should correspond to our participated in the issues being investigated. Hence, select-
theoretical framework and the variables we are studying. For ing the appropriate type of organisation is a balancing act
example, for our project we had to first select service sector between certain advantages and disadvantages.
firms who had undertaken FDI in China or Korea who used To illustrate the case selection process, at the outset of our
suppliers for at least some inputs. Once we have selected study on inter-firm linkages in China and Korea we had ini-
a firm, the informant has to be a manager who is involved tial criteria for the selection of cases (e.g. the parent firm
in the process of working with suppliers. An interview with must be a UK headquartered service firm who has made
the firm’s public relations manager or an accountant may not significant direct investments in China and Korea). These
provide us with the information we are looking for (Ghauri criteria were then altered by both pragmatic considerations
and Grønhaug 2005; Ghauri 2004). In order to select firms (above all, the level of access that we could obtain with or
within the service sector we used the Thomson One Banker without personal contacts of colleagues and friends) and by
Deals database to see which firms had made acquisitions and the emergent theory. Thus, when choosing the final cases,
then wrote to the CEO’s of these companies asking for the some UK headquartered service firms were ruled out if they
names of relevant managers we could interview. had not made large enough direct investments which would
The time available for the study, financial resources for mean that they would not have a large supplier network that
travelling, personal contacts and other practical issues are of we also needed to interview. However, important considera-
great importance. For example, depending upon how much tions in the selection process were not only differences but
time we have to perform the study, the type of organisation also similarities; with all the cases sharing some features that
or company we select might be different. Consequently, the made them comparable (e.g. service industry, size and type
researcher is likely to be required to mix theory with prag- of the project, in our case acquisitions).
matism when selecting cases. During our trips to China and How many cases should be included in a study? The an-
Korea we spent about two weeks in each country and had swer to this question is very difficult as there is no upper or
set up virtually all the interviews prior to the trip, therefore lower limit to the number of cases to be included in a study.
a high level of organisation was needed. The most appro- Many times only one case is enough (Mintzberg 1979). It is
priate company in theoretical terms may be too far away, or the research problem and the research objectives that influ-
may simply refuse access. Whilst studying the linkage ef- ence the number and choice of cases to be studied.
fects of foreign firms on local firms and economies we used Single cases are appropriate when a particular case is
a variety of avenues in order to gain access. This ranged a critical case and we want to use it to explain or ques-
from using personal contacts, university alumni, writing in- tion an established theory. It is a critical case because it
troduction letters and sometimes even cold calling. The first meets all the conditions necessary to confirm, challenge
meeting with the potential interviewee or company repre- or extend the theory. For example, in Marschan-Piekkari
sentative is likely to be decisive in terms of whether we and Ghauri (1998) Researchers studied regional control in
are granted access or not. In our experience, the final de- headquarter-subsidiary and inter-subsidiary relationships
cision about access is largely determined by the ability of in a single, Finnish multinational corporation. Given the
researchers to impress the interviewee during the very first limited number of earlier studies in the field, the aim was
meeting. We conducted case studies in many different coun- to extend existing theory on regional management in multi-
tries during our study of linkage formation and our experi- nationals by examining in depth how control is exercised
ence revealed that often being a foreigner is not necessarily through regional centres. Another situation is when a single
a disadvantage when negotiating access. Particularly when case is an extreme or a unique case; for example, particular
having to rely on a second language, expressions tend to organisations may be of interest because they represent
be simpler and the overall communication tends to be more “outstanding successes’ or “notable failures” (Patton 1990,
straightforward and direct. p. 169). Finally, a single case design is appropriate when
Case selection should also take into account the type of a case is revelatory. This means that we can observe and
organisation being studied. Based on our own experiences study a phenomenon which was previously not accessible
of conducting case studies including SME’s such as suppli- and which can provide useful insights. We can also use sin-
ers and MNCs (Ghauri et al. 2008; Ghauri and Fang 2001), gle case design in other situations, such as in a pilot study or
we would argue that research problems tend to be richer in an exploratory study that serves as a first step to a later, more
MNCs, but it is likely to be harder to negotiate access and comprehensive study (Ghauri and Grønhaug 2005; Yin
identify the right informants who have personally been in- 2003).
volved in the phenomenon under study. In SMEs, on the In comparative or multiple case studies, we ask or study
other hand, the informants may be eager to learn and willing the same questions in a number of organisations and com-

13
The formalization of case study research in international business 33

pare them with each other to draw conclusions. The purpose one method. The main advantage of triangulation, however,
of data collection in comparative case study method is to is that it can produce a more complete, holistic, and con-
compare (replicate) the phenomenon (e.g. strategy forma- textual picture of the object under study. In the case study
tion) in a systematic way, to explore different dimensions of method, it is particularly important as we need to check and
our research issues or to examine different levels of research validate the information we receive from various sources
variables. In this approach it should be clear that every case and examine it from different angles (Ghauri 2004). For ex-
has to serve a particular purpose in the study. In other words, ample, we can check the performance of a firm or a project,
the researcher has to justify the selection of each case. How- claimed by the interviewee, from annual accounts, archives
ever, case study design is often flexible and can be changed, or by interviewing another manager or company represen-
modified or revised during the study with proper justifica- tative. In essence, we use triangulation to reduce the likeli-
tion. For example, in our study of inter-firm linkages we hood of misinterpretation. We employ various procedures to
selected 8 cases (4 in China and 4 in Korea) in order to increase our understanding and explanation. It helps us clar-
compare the cases not only by company but also by coun- ify meaning by identifying different ways the phenomenon
try in which they had invested to see the differences in pat- is seen (Denzin 1989; Flick 1992).
terns of linkage formation between subsidiaries and local In our own research, we have tended to build up case data
suppliers. through interviews and written documentation. When plan-
A related question is what is meant by “a case”? For ning personal interviews, we have made a special effort to
example, if we are trying to understand the process of include multiple viewpoints in the data set by, for example,
decision-making in a firm, we may study multiple decisions interviewing managers from the HQ in the UK, managers
on different issues, important/unimportant, novel/routine from subsidiaries in China and Korea and managers in local
decisions etc. in the same organisation. This will provide supplying companies. In another study both joint venture
variability along important factors (see e.g. Campbell 1975). partners and joint venture managers were interviewed
Thus, in Ghauri and Fang (2001) a single MNE (Ericsson (Hyder and Ghauri 2000); internal company informants
in China) was studied, but in the context of this one unit, as well as external local authorities, officials and staff at
the researchers were able to analyse multiple instances of foreign embassies (Hadjikhani et al. 2008). In our project
the phenomenon under study, namely international business on linkages, to supplement these interviews, we were able
negotiations. A case in the linkages study was defined as to analyse company files related to each case. The files
the FDI context i.e. all firms associated with the process of incorporated such material as correspondence between the
FDI from headquarters, subsidiaries and suppliers. subsidiary and suppliers, contract drafts as well as the final
The use of a particular case study method depends also contract, and relevant government rules and regulations. It
upon the type of study we are doing, whether it is inductive was therefore possible to compare data between interview-
or deductive, and also upon whether we are looking for spe- ees within the same firm, between different firms, and in
cific or general explanations. In the situation of an inductive comparison with the written records of each company.
approach and specific explanations, we may use the single Moreover, the comparison of archival and interview data
case design. However, if we are looking for general expla- is not the only option. As part of the case study reported
nations, we should use a multiple-case design. On the other in Ghauri and Fang (2001), the researchers shadowed
hand, if we are doing a study with a deductive approach a top manager for a couple of years in the early 1990s.
we can use the case study at an early stage to develop our This method, which was very much built on trust and
hypotheses or propositions. A case study is a less recom- mutual respect between the top manager and researcher,
mended method if we are aiming at generalisations. How- provided real insight into face-to-face situations and in-
ever, case studies are justifiable if we know the outcome that creased the researchers own understanding of the research
we are aiming at from the very beginning and clearly state questions. At the same time, the manager found it useful
them so as not to promise over inflated outcomes. to jointly brainstorm and analyse various situations we had
observed.
There are some problems with triangulation. Sometimes
Conducting case studies it can be difficult to judge the accuracy if the results from
different methods and sources are not consistent. For ex-
Triangulation can be one of the defining features of a case ample, the MNE headquarter may say it was responsible
study. It refers to the collection of data through different for initiating linkages whereas suppliers may indicate oth-
methods or even different kind of data on the same phe- erwise. A second problem arises when the different meth-
nomenon. The use of multi-methods or triangulation is not ods employed provide contradictory results. For example, if
new and can be traced back to Campbell and Fiske (1959), a manager reports no conflicts in the relationship while the
who argued that to ensure validity one should use more than archives demonstrate the existence of a number of conflicts,

13
34 P. N. Ghauri, R. Firth

it is important to have this clarified during the interviews Table 1 Case study analysis
with the managers by referring to the incidents mentioned
Techniques for case Explanation
in the archives. However, there is a general tendency that
study analysis
we need not evaluate interview responses or written docu-
ments as true or false reports on reality. “Instead, we can Chronologies Narratives of the events that took place,
organised by date
treat such responses as displays of perspectives and moral
forms” (Silverman 1993, p. 107). One conclusion is that tri- Coding Sorting data according to concepts and
themes
angulation or the usage of a multi-method approach on the
same study object can be useful even if we do not get the Clustering Categorising cases according
to common characteristics
same results. It can lead us to a better understanding or to
Matrices Explaining the interrelationship
new questions that can be answered by later research. Doc-
between identified factors
uments can be tremendously useful. We have relied on them
Decision tree modelling Grounding a description of real-world
when preparing sharp and to-the-point interview questions. decisions and actions coherently
We have also used them when checking information men- by using multiple cases
tioned in interviews. Hence, documents tend to be rich and Pattern matching Comparison between a predicted and
have a lot to offer to the researcher both in pre-interview and an empirically based pattern
post-interview situations.
Source: Ghauri 2004

Analysing case studies: the traditional approach


and the data collection processes. In case study research,
Interpreting and analysing qualitative data is perhaps the inter-weaving data collection and data analysis right from
most difficult and time consuming task while doing case the first case/interview is the best policy (Miles and Hu-
study research. We cannot be satisfied merely with “telling berman 1994). This allows theory to develop alongside the
convincing stories”, to use Silverman’s (1989) chapter ti- growing volume of data, allowing the research problem
tle. In qualitative research trustworthiness and authenticity to be formulated or even reformulated at the same time.
rather than reliability are the main issues (Sinkovics et al. This often leads to new questions and new data collection,
2008). The idea is to present an “authentic” understanding and there is no definite phase of data analysis (Ghauri and
of people’s experience. This means not just understanding Grønhaug 2005; Grønhaug 1985; Miles and Huberman
the point of view of the individuals and groups being stud- 1994). Preferably, a second case should not be started
ied; in addition, data has to be interpreted against the back- unless the data collected through the first case has been
ground of the context in which they are produced (Hammer- analysed. It will reveal the blind spots and deficiencies of
sley and Atkinson 1983). But how can researchers ensure data collected and the researcher can improve his/her data
– and demonstrate – they have produced an authentic and collection techniques in the subsequent cases. Moreover,
trustworthy interpretation rather than a misguided one? As early analysis reduces the risk that the researcher might
put by Miles (1979, p. 591): simply drown in the sheer volume of data. It is often difficult
for case researchers to filter or discard irrelevant data before
“The most serious and central difficulty in the use of analysis commences.
qualitative data is that the method of data analysis is While analysis may not form an isolated process, never-
not well formulated . . . the analyst faced with a bank theless it does involve some distinct stages. The first step
of data has very few guidelines for protection against of analysis is to construct a case description and explana-
self-delusion, let alone the presentation of unreliable tion. This will help us understand “how” things are develop-
or invalid conclusion to scientific or policy-making de- ing and “why” things occur the way they do. First we have
cisions. How can we be sure that an “earthy”, “undeni- to describe, i.e. make complicated things understandable in
able”, “serendipitous” finding is not, in fact, wrong?” their component parts, and explain, i.e. show how their com-
ponent parts fit together according to some rules (Bernard
Part of the answer lies in making sure that data analysis and 1988). However, it is hard to describe and explain something
collection are closely interconnected during the lifecycle satisfactorily unless you understand what this “something”
of the case study research. Many researchers put a lot is. It is thus important to start with simple “story telling”
of effort into data collection and keep on collecting data about a situation and progress in chronological order. Con-
through case studies, hoping to do data analysis at the end sequently, we can construct a map and locate different ele-
of the process, sometimes months or years after the data ments and variables. This will finally lead us to build a the-
collection. However, this will weaken both the analysis ory or a model, i.e. how the variables are connected together

13
The formalization of case study research in international business 35

and how they influence each other (Miles and Huberman


1994). Step by step we can advance along the “ladder of
abstraction”. We start by trying to code and categorise text,
then identify trends and establish findings. Finally, we inte-
grate the data into an explanatory framework (Ghauri 2004;
Carney 1990; Gherardi and Turner 1987). These different
techniques are summarised in Table 1.
The first stage of analysis, “story-telling”, can be done by
writing chronologies or biographical histories of the organ-
isation(s) or individual(s) under study. This is particularly
important when the researcher is attempting to develop lon-
gitudinal explanations that track a phenomenon over time.
For example, in our study, the first step in our analysis was
to construct a simple narrative of each interview and then
Table 2 An example of a matrix display in case study analysis
other interviews on the same case, to see the whole picture.
The second stage of analysis is also a sifting process. This
means rearranging the data that has been collected, but into
more conceptual rather than chronological categories. To sures and specific quoted examples (Miles and Huberman
analyse data we (often) have to code them so that they can 1994).
be broken down, conceptualised, put together and presented Another possible step in the analytical process of answering
in an understandable manner. Sorting the data in this our research questions is to test our propositions. This is
fashion is typically done through coding, in other words, done by cross-checking for commonality and integrating
classifying the data. This coding and categorisation will the data in one single framework through a meticulous case
help us to interpret the data and to relate the information comparison. Developing a good case comparison in a mul-
to our questions and frameworks. This will also enable us tiple case study is essential but not very simple, as different
to locate different categories when we are analysing data cases might tell different stories and lead to different vari-
to find conclusions. As qualitative studies quite often help ables. It is also necessary to know/establish relevance and
in building theories, coding requires extra care, and a bal- applicability of our findings to similar situations (Firestone
ance between creativity, rigour and persistence has to be and Harriott 1983). The most important reason, however, is
achieved. to enhance our understanding and explanation, as this will
There are several software programs available that can help help us explain under what circumstances/conditions our
researchers in coding, sorting and analysing qualitative data. propositions/model work and under what conditions they
We will subsequently look more closely at the uses and ben- do not. We can then conclude under what conditions the
efits of this type of software. model is most likely to work (Glaser and Strauss 1970). We
An important part of the sifting process is searching for have noticed that case study researchers have considerable
common or conflicting themes in data, and looking for difficulties in drawing conclusions that clearly stem from
themes and trends related to our research questions. This the case data. The way in which we have done this (using
will allow us to identify relationships between different Nvivo) is to directly relate text from the interviews to our
themes and research questions. Moreover, it will enable conceptual framework and compare this information in each
us to detect gaps in our data and to write initial analytical case.
reports. The requirement is that the pattern has to be As illustrated by the above example, we can find answers to
sufficiently systematic. Many researchers suggest that case our questions by comparing different cases. In some cases
comparison can be done by forming types or clusters. the best and worst comparisons are most suitable (Churchill
This technique involves inspecting cases and trying to put 1991). In the same manner we could compare performance
them into groups or clusters that share similar patterns of of several units/branches of a company by comparing them
configurations. These clusters can also be sorted according with each other.
to some dimensions, e.g. international versus domestic Another approach is to use the strategy suggested by Miles
firms (Lofland and Lofland 1984). Another way is to look and Huberman (1994, p. 176): “stacking comparable cas-
for themes or variables that cut through cases. For exam- es”. According to this strategy, we can write up each of a se-
ple, Eisenhardt (1989) found evidence for the construct of ries of cases, using more or less standard variables. Then
“power centralisation” by looking at data on CEO behaviour we can use matrices and other displays to analyse each case
in 10 micro-computer firms: her matrix display included in depth, and explore the interrelationship between different
objectives describing decision style, quantitative mea- factors. Once each case is well understood, we can “stack”

13
36 P. N. Ghauri, R. Firth

the case-level display in a “meta-matrix”, which is then fur- in our linkages project we began with a conceptual frame-
ther condensed, permitting systematic comparison (Ghauri work which was derived from the literature and sought to
2004). Eisenhardt (1989) called this strategy a “replication include all of the main units which would influence linkage
strategy”. formation and the variables associated with each one. The
In case studies where data analysis leads to the development text was then coded with the aim of trying to link parts
of propositions, the aim of the analysis should be a link of the text with the variables included in the conceptual
between data and propositions. Campbell (1975) and Yin framework. The precise method of analysis will in addition
(2003) discuss “pattern matching”, where several pieces of depend on the type of study being undertaken. In descriptive
information from one or several cases are related to a priori research, the researcher has to work with specific research
assumptions. If we can find a systematic or unsystematic problems, propositions or hypotheses. While an exploratory
pattern, we can accept or reject our propositions. In such study is characterised by flexibility (see e.g. Ghauri 1983),
studies, statistical tests are not necessary to establish a descriptive study can be considered more rigid. It requires
a pattern and there are no precise ways of testing or setting a clear specification of the who, what, why and how of the
criteria for interpreting these findings. The pattern has research problem.
to be sufficiently systematic to accept certain propos-
itions.
As case analysis is a difficult task, we recommend that Analysing case studies using CAQDAS
researchers should use a mixture of the above mentioned
strategies: case comparison based on pattern seeking, It has been argued that with globalisation and increasingly
clustering, matrices and other techniques displayed in more complex business environments; both managers and
Table 2. Most often, we start with a preliminary model or researchers are being bombarded with an overload of infor-
set of variables based on existing literature, then develop mation (Sinkovics et al. 2005). This trend is not reversible
and refine a model as the analysis progresses. For example, and therefore, a way in which information and data alike can

Table 3 The advantages of using Nvivo in IB research

Characteristics of IB research Implications for analysis of case studies Advantage of using Nvivo

Data saturation – advances in technology Case studies can become increasing large All types of data can be neatly stored and
mean the researcher can collect more and less manageable organised with a single programme increasing
data on a global scale clarity and focus of analysis

Research may be undertaken by a variety Interpretations of cases may differ between Nvivo offers a single platform in which all
of different researchers based in different researchers and ideas may not be communi- researchers can work with individual passwords.
countries cable cross borders in real-time in order Data can then easily be emailed so researchers
to cross check the data can view/alter others ideas in an iterative process
which is less subjective

Language barriers/difference in language Case studies may not be comparable/validity If researchers and multi lingual, newer versions
of data collected of data may be compromised of Nvivo will accept other languages and charac-
ters than English. Different codes and categories
can be used to show which language was used
and retain equivalence of the data

The unit of analysis often differs Case studies must recognise the differences Nvivo allows categorisation of unit of analysis
i.e. HQ/Subsidiary/Local Firm in units of analysis and compare issues and direct comparisons on a single screen
according to separate accounts from between them are possible
individual units

Time constraints of cross border research Case study analysis may be put on hold Nvivo allows initial analysis of the data even if
until ALL data is collected, compromising collection is not totally complete as alterations
the quality of analysis can easily be made at a later date, rather than
the analysis having to be repeated

IB Research is multidisciplinary and likely Case studies can be rich with data however, Nvivo can store all sources of data including text,
to make greater use of a variety of sources linking different sources may be difficult webpage’s, electronic files, videos, pictures and
of data audio files and link them together in terms
of their coding

Source, based on Sinkovics et al. 2005; Ghauri and Cateora 2006

13
The formalization of case study research in international business 37

be more systematically organised must be utilized for bet- criticism levelled at them that qualitative research is too
ter ease of understanding. Computer aided qualitative anal- subjective, biased and not at all replicable, without compro-
ysis software (CAQDAS) has been developed for this exact mising the usefulness of qualitative data in describing and
purpose and is being used more frequently owing to the in- creating meaning and answering the “how” and the “why”
creasing popularity of qualitative methods as a whole. Nvivo questions.
is a qualitative data analysis software which has to a large
extent replaced NUD* IST as the most popular. Nvivo al-
lows for the storage, editing and organisation of data through Nvivo: a practical example
a process of meaningful coding. Researchers are often disap-
pointed to learn that unlike some quantitative data analysis We undertook a large IB project looking at the formation
tools, Nvivo does not generate propositions or “results” for and impact of inter-firm linkages between UK service
them. It is primarily an organisational tool which provides multinationals and local firms in host economies in which
a cleaner and more accessible way of looking at data. How- we utilized Nvivo in order to analyse the data. The aim
ever, we argue that this fact does not diminish its usefulness; was to find out what influenced linkages between firms
on the contrary, it encompasses the reliability of systema- to be formed and whether these generally deepened or
tising data without losing rich textual narratives. Nvivo is widened the longer a company operated in a specific market
in fact particularly useful in International Business cultural and whether there was any positive or negative effect of
research (see Table 3). these long term relationships with multinationals for locally
When one considers all the disadvantages of paper based operating firms. A comparative case study method was used
analysis; storage of multiple large documents, ensuring by which we selected a number of UK headquartered firms
confidentiality of data in storage, shuffling through to look in the service sector who had undertaken significant FDI
for each and every new argument, manual highlighting and in a number of emerging markets. The focus was primarily
cutting out of the text, Nvivo seems like a very attractive on two Asian markets: China and Korea. There were 8 case
alternative. There is some disagreement on a theoretical studies in total, 4 for each market. Semi structured personal
level as to whether the use of Nvivo may make qualitative interviews were undertaken with elite managers in the UK
data more quantitative in nature. However; this is for the headquarters for each firm and then a snowball method
individual researcher to decide when enough structuring (Hornby and Symon 1994; Patton 1990) was used to identify
is enough. In general, Nvivo helps researchers to counter the equivalent managers in the subsidiaries in each country

Fig. 1 Process of Nvivo use in


linkages project

13
38 P. N. Ghauri, R. Firth

Fig. 2 Example of the coding


process using nodes in Nvivo

and subsequently suppliers or distributors of the sub- • There is a lot of technical language involved in a study
sidiaries in the host countries were interviewed. In total, 29 of linkages and therefore interviewees used long descrip-
interviews were conducted in total, typically 3 or 4 for tions and stories to exemplify linkages in their particular
each case. All interviews were digitally recorded and subse- organisation.
quently transcribed. The use of Nvivo can begin even before
Nvivo Solution: – Textual data could be thoroughly
the interviews are transcribed by using the notes or memos
searched and categorised quickly even when larger quota-
function to immediately record thoughts and observations
tions or explanations were given by managers.
about the interview in hand, the following process is typical
(see Fig. 1). • Interview transcripts had to be combined with informa-
A conceptual framework was derived from existing litera- tion from company websites, secondary documentation
ture and these provided the basis for the initial codes. There given to the interviewers during interviews, audio files,
were approx. 24 codes in total which in Nvivo are called memos and news articles.
nodes. These are the categories under which we sought to
link parts of the text. The first stage of this process was Nvivo Solution: – stores and links together all data
to try to fit as much of the text as possible into these pre- sources, increased comparability of different data sources.
defined codes, breaking these down further at a later stage. • Different units of analysis used i.e. Headquarter,
Figure 2 shows an example of the coding process in our subsidiary and local firm.
linkages project using a selection of nodes which are effec-
tively variables from our conceptual framework. This pic- Nvivo Solution: – Units of analysis can be compared to
ture shows the coding at a later stage when the codes have al- back up findings on a particular assertion that can enhance
the reliability.
ready been broken down into “Tree” nodes and “free” nodes,
sometimes called “parent” nodes and “child” nodes. The
sorting of nodes into different levels indicates the relation-
ships between them and although in our project we used Displaying findings using Nvivo
only two different levels, depending on the depth of analysis
needed, the data can be ordered into many different levels of One very evident problem in using Nvivo is how to display
nodes. the patterns found. Many researchers claim they have used
Nvivo was suitable for use on the linkages project for Nvivo for the analysis but fail to show specifically how
a number of reasons: their propositions have been arrived at as a direct result
of the Nvivo analysis. Typically, the modelling tool within
• There was a large amount of data (amounting to approx.
Nvivo can be used for this and the traditional method
377 double spaced A4 pages for transcripts alone).
of storytelling and using meaningful coded quotations
Nvivo Solution: – Enabled collation, organising and link- from transcripts. Nvivo can also be used to increase
ing of different data sources in a user-friendly manageable reliability when using traditional methods of analysis
format. such as matrices, tables and pattern matching amongst

13
The formalization of case study research in international business 39

others. In the linkages project the first stage of analysing nent scholars are also calling for case studies to provide us
the data was to do descriptions, storytelling and pick out with a more in-depth understanding and insights into cul-
important quotations from the interviews using as many turally embedded problems of our field. Yet we do not see
examples as possible. We did this per code, outlining case studies widely published in top class journals such as
the findings for each code or node. In a sense this is Journal of International Business Studies, Journal of Mar-
a form of template analysis or pattern matching as the keting or Academy of Management Journal. As researchers,
text was compared to a pre-defined pattern which was we would strongly encourage case study researchers to ex-
our conceptual framework. Following this a series of ercise a little self-criticism and move beyond “just telling
Matrices were created in order to compare answers on stories’. It is pivotal that case study research transcends the
each code or node from the headquarters, subsidiaries and researchers own orientations, in order to be objective and to
domestic firms. A further attempt at structuring the data provide a clear picture of phenomena as they actually are
in a more quantitative way was then made as we tried rather than as we see them.
to give each code/node a number which we subjectively
determined by judging that codes relative importance in Acknowledgement The authors would like to thank Economic and
Social Research Council (ESRC-UK) for funding the study reported in
the case as a whole. As a secondary measure we added this paper.
up what percentage of the text referred to that particular
code, also to give an indication of importance. However, it
must be noted that this latter measure should not be used References
alone because percentage of coverage of the text does not
necessarily reflect importance as there may be other factors Bernard HR (1988) Research methods in cultural anthropology. Sage,
impacting on this such as length of interview as a whole or Newbury Park
simply the talkative nature of the interviewee compared to Bonoma TV (1985) Case research in marketing: opportunities, prob-
lems, and a process. J Marketing Res 12:199–208
another.
Campbell DT (1975) Degrees of freedom and the case study. Comp
Nvivo can also generate a number of reports which give Polit Stud 8(2):173–193
statistical style data on the coding and organising that has Campbell DT, Fiske W (1959) Convergent and discriminant validation
been done. This was also incorporated into the written anal- by the multitrail-multimethod matrix. Psychol Bull 56:81–105
Carney TF (1990) Collaborative inquiry methodology. University
ysis. The final stage in displaying data analysed in Nvivo is
of Windsor, Division for Instructional Development, Windsor,
the use of the modelling tool in which we reformulated our Ontario
conceptual framework to include concepts proved to be im- Churchill GA (1991) Marketing research: methodological foundations,
portant. This was also used on a smaller scale in the compar- 5th edn. Dryden Press, Chicago
Cooper HM (1984) The integrative research review: a systematic ap-
ative part of the analysis to show close relationships and ties proach. Sage, Beverly Hills
between codes. Visualisation of the data as well as descrip- Coviello NE, Ghauri PN, Martin KAM (1998) International compe-
tion is an important aspect of analysis which is sometimes titiveness: empirical findings from SME service firms. J Int Mar-
overlooked. keting 6(2):8–27
Denzin NK (1989) The research act, 3rd edn. Prentice Hall, Englewood
Cliffs
Eisenhardt KM (1989) Building theories from case study research.
Conclusions Acad Manage Rev 14(4):532–550
Firestone WA, Harriott RE (1983) The formalization of qualitative re-
search: an adaption of “soft” science to the policy world. Evalua-
The primary aim of this paper has been to relate traditional tion Rev 7:437–466
approaches to case studies, and in particular to their anal- Flick U (1992) Triangulation revisited: strategy of validation or alter-
ysis to more modern computer aided ways of looking at native? J Theor Soc Behav 22:175–197
them. Based on our experiences of conducting case studies, Flick U, Kardorff EV, Steinke I (2004) A companion to qualitative re-
search. Sage, London
it seems that case studies are becoming an increasingly ac-
Ghauri PN (1983) Negotiating international package deals: Swedish
cepted form of research. We assert that the use of CAQDAS firms in developing countries. Almquist and Wiksell, Stockholm
should also be an accepted and preferable way of looking Ghauri PN (1992) New structures in MNCs based in small countries:
at cases. In order for the trend towards using case studies a network approach. Eur Manage J 10(3):357–364
Ghauri PN (2004) Conducting and analysing case studies in interna-
to continue, the academic community must believe in their
tional business. In: Piekkari R, Welch C (eds) Handbook of qual-
worth and understand the intricate way in which they are itative research methods for international business. Edward Elgar,
sometimes conducted. For this to occur the conduction and Cheltenham
analysis processes of case studies must be both systematic Ghauri P, Fang T (2001) Negotiating with the Chinese: a socio-cultural
analysis. J World Bus 36(3):303–325
and transparent. Only this way will case study research be
Ghauri P, Grønhaug K (2002) Research methods in business
truly valued and taken seriously at both at international aca- studies: a practical guide. Financial Times and Prentice Hall,
demic conferences and in top level journals. Several promi- Harlow, UK

13
40 P. N. Ghauri, R. Firth

Ghauri PN, Grønhaug K (2005) Research methods in business studies: Jean RJB, Sinkovics RR, Kim D (2008) Information technology and
a practical guide, 3rd edn. Prentice Hall, Harlow, UK organizational performance within international business to busi-
Ghauri PN, Holstius K (1996) The role of matching in the foreign ness relationships: a review and an integrated conceptual frame-
market entry process in the Baltic States. Eur J Marketing work. Int Marketing Rev 25(5):563–583
30(2):75–88 Lofland J, Lofland LH (1984) Analyzing social settings: a guide to
Ghauri PN, Elg U, Tarnavskaya V (2008) Market driving supplier strat- qualitative observational research. Wadsworth, Belmont
egy: a study of Ikea’s global sourcing networks. Int Marketing Rev Marschan-Piekkari R, Ghauri PN (1998) Growing pains in multi-
25(6):674–699 nationals: controlling subsidiaries through regional centers.
Gherardi S, Turner BA (1987) Real men don’t collect soft data. Research Report 98B39. Research Institute Systems, Organisa-
Quaderno 13, pp 1–17. Dipartimento di Politica Sociale, Univer- tion and Management, University of Groningen, Netherlands
sità di Trento Miles MB (1979) Qualitative data as an attractive nuisance: the prob-
Glaser BG, Strauss AL (1970) Discovery of substantive theory. In: lem of analysis. Admin Sci Quart 24:590–601
Filstead W (ed) Qualitative methodology, pp 288–297. Rand Mc- Miles MB, Huberman AM (1994) Qualitative data analysis, 2nd edn.
Nally, Chicago Sage, Thousand Oaks
Grønhaug K (1985) Problemer i empirisk forskning. In: NHH/RSF Mintzberg H (1979) An emerging strategy of direct research. Admin
(ed) Metoder og Perspektiver i Økonomisk-Administrativ Forskn- Sci Quart 24:582–589
ing. Universitetsforlaget, Oslo Patton MQ (1990) Qualitative evaluation and research methods, 2nd
Hadjikhani A, Lee JW, Ghauri PN (2008) Network view of MNCs’ edn. Sage, Newbury Park
socio-political behavior. J Bus Res 61:912–924. Selltiz C, Wrightsman S, Cook SW (1976) Research methods in social
Hammersley M, Atkinson P (1993) Ethnography: principles in prac- relations, 3rd edn. Wiley, New York
tice. Tavistock, London Silverman D (1989) The impossible dream of reformism and roman-
Hornby P, Symon G (1994) Tracer studies. In: Cassell C, Symon G ticism. In: Gubrium JF, Silverman D (eds) The politics of field
(eds) Qualitative methods in organisational research. Sage Publi- research: beyond enlightenment. Sage, Newbury Parke, pp 30–48
cations, London Silverman D (1993) Interpreting qualitative data: methods for
Hyder SA, Ghauri PN (1989) Joint venture relationship between analysing talk, text and interaction. Sage, London
Swedish firms and developing countries: a longitudinal case Sinkovics R, Penz E, Ghauri PN (2005) Analysing textual data
study. J Global Marketing 2(4):25–47 in international marketing research. Qual Market Res Int J
Hyder SA, Ghauri PN (2000) Managing international joint venture 8(1):9–38
relationships: a longitudinal perspective. Ind Marketing Manage Sinkovics R, Penz E, Ghauri PN (2008) Enhancing trustworthiness
29:205–218 of qualitative research in International Business. Manage Int Rev
Jean RJB (2007) The ambiguous relationship of ICT and organiza- 48(6):689–714
tional performance: a literature review. Crit Perspect Int Bus Yin RK (2003) Case study research: design and methods, 3rd edn.
3(4):306–321 Sage, Thousand Oaks

13

Das könnte Ihnen auch gefallen