Journal of Operations Management 29 (2011) 329–342

Contents lists available at ScienceDirect

Journal of Operations Management

journal homepage: www.elsevier.com/locate/jom

Qualitative case studies in operations management: Trends, research outcomes,

and future research implications
Mark Barratt, Thomas Y. Choi ∗ , Mei Li
Department of Supply Chain Management, W. P. Carey School of Business, Arizona State University, Tempe, AZ 85287-4706, United States

a r t i c l e i n f o a b s t r a c t

Article history: Our study examines the state of qualitative case studies in operations management. Five main operations
Received 31 July 2009 management journals are included for their impact on the field. They are in alphabetical order: Decision
Received in revised form 13 June 2010 Sciences, International Journal of Operations and Production Management, Journal of Operations Management,
Accepted 23 June 2010
Management Science, and Production and Operations Management. The qualitative case studies chosen
Available online 1 July 2010
were published between 1992 and 2007. With an increasing trend toward using more qualitative case
studies, there have been meaningful and significant contributions to the field of operations management,
Keywords:
especially in the area of theory building. However, in many of the qualitative case studies we reviewed,
Case studies
Research methods
sufficient details in research design, data collection, and data analysis were missing. For instance, there
Inductive are studies that do not offer sampling logic or a description of the analysis through which research out-
Deductive comes are drawn. Further, research protocols for doing inductive case studies are much better developed
Qualitative compared to the research protocols for doing deductive case studies. Consequently, there is a lack of
Theory building consistency in the way the case method has been applied. As qualitative researchers, we offer suggestions
Theory on how we can improve on what we have done and elevate the level of rigor and consistency.
Testing Published by Elsevier B.V.

1. Introduction tion techniques (Gupta et al., 2006; Rungtusanatham et al., 2003;

Scudder and Hill, 1998).
Since the early 1980s there have been calls for empirical As an alternative to survey-based research, other OM scholars
research methods in response to the over-reliance on the pre- have promoted the use of qualitative case study research (Lewis,
dominant analytical research paradigm in operations management 1998; McCutcheon and Meredith, 1993; Meredith et al., 1989; Voss
(OM) (Buffa, 1980; Chase, 1980; Flynn et al., 1990; Meredith et al., et al., 2002). We define a qualitative case study as an empirical
1989; Swamidass, 1991; Wood and Britney, 1989). More recently, research that primarily uses contextually rich data from bounded
there have been calls for more relevance and rigor when con- real-world settings to investigate a focused phenomenon (adapted
ducting empirically based research (Boyer et al., 2005; Eisenhardt from Benbasat et al., 1987; Bonoma, 1985; Meredith et al., 1989;
and Graebner, 2007; Fisher, 2007; Roth, 2007). Academics in the Meredith, 1998; Roth, 2007; Yin, 1994). This approach has appealed
OM field responded with predominantly deductive survey-based to researchers, as the field of OM has many emerging areas of
empirical studies (Scudder and Hill, 1998). Recently there have research such as the integration of OM with other functional areas
been a number of reviews of empirical research, focusing on specific of the supply chain (e.g., Hines et al., 2002; Pagell, 2004). The intent
topics such as operations strategy (e.g. Boyer et al., 2005), inter- is to build and extend theories (Eisenhardt, 1989; Yan and Gray,
disciplinary and inter-organizational research (e.g., Buhman et al., 1994) and to explore and better understand emerging, contempo-
2005), sustainability (e.g., Kleindorfer et al., 2005), new product rary phenomena or issues in their real world settings (Flynn et al.,
development (e.g., Krishnan and Loch, 2005), quality management 1990; Meredith, 1998).
(e.g., Schroeder et al., 2005), and supply chain management (e.g., This paper seeks to examine the state of and research outcomes
Kouvelis et al., 2006). There have also been other studies that from qualitative case studies in the OM field, as captured by the
reviewed the state of survey research methods and data collec- inductive and deductive articles published in five main OM journals
(Barman et al., 2001, 1991; Olson, 2005). There have been simi-
lar papers in disciplines outside the OM field. For instance, Dubé
∗ Corresponding author. Tel.: +1 480 965 6135; fax: +1 480 965 8629. and Paré (2003) reviewed the rigor and quality of case studies pub-
E-mail addresses: [email protected] (M. Barratt), [email protected] lished in leading information systems journals during the period
(T.Y. Choi), [email protected] (M. Li). 1990–1999. We are not aware of any similar papers in the OM dis-

0272-6963/$ – see front matter. Published by Elsevier B.V.

doi:10.1016/j.jom.2010.06.002
330 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

cipline, and we intend to fill this void. The purpose of our paper 1998; Rothlisberger, 1977); the research is exploratory and there-
is four-fold. (1) It provides a summarizing review of the numer- fore calls for case research to build theories (Meredith, 1998; Yin,
ous guidelines for undertaking inductive-based case study research. 1989); the research is explanatory (i.e., asking “how” and “why”
(2) It captures the state of the scarce literature for undertaking types of questions) and the context and experiences of actors are
deductive-based case study research. (3) It provides a review of critical (Benbasat et al., 1987; Bonoma, 1985), especially the expe-
the current state (1992–2007) of qualitative case studies and per- riences of managers so as to increase the practical relevance of the
forms a gap analysis between the guidelines and the current state. findings (Fisher, 2007).
The journals include four US-based journals and one European-
based—Decision Sciences (DS), International Journal of Operations and 2.1.2. Research focus and specification of unit of analysis
Production Management (IJOPM), Journal of Operations Management When attempting to build theory from case studies, researchers
(JOM), Management Science (MS), and Production and Operations should have a clear focus to collect specific data in a systematic
Management (POM).1 (4) It provides a review of theoretical contri- manner (Mintzberg, 1979). This focus helps to define the research
butions made by the use of qualitative case studies and identifies question, the types of data to be collected and the types of organi-
subsequent opportunities to improve such theoretical contribu- zations to be approached (Leonard-Barton, 1990; Pettigrew, 1990).
tions. Although research questions may evolve over time and constructs
We begin by reviewing the literature on qualitative case-study may be modified (Eisenhardt, 1989; Voss et al., 2002), there must be
methods, for both inductive and deductive approaches, focusing focus which helps maintain consistency throughout data collection
on the various research frameworks and guidelines that have been and analysis (Benbasat et al., 1987).
proposed to facilitate increased usage and level of rigor. After this, Once the research focus has been specified and the research
we present our research methodology and the results of our anal- questions have been articulated, the unit of analysis must then be
yses. We then focus on the details of contributions made by the clearly specified (Yin, 1989; Dubé and Paré, 2003). When the unit of
qualitative case studies, relative to the existing guidelines. We end analysis is unclear, this influences the research questions and out-
with a general discussion and implications for future qualitative comes (Yin, 1989). Dubé and Paré (2003: 610) suggest that clearly
case studies. defining the unit of analysis “is critical if we want to understand
how the case relates to a broader body of knowledge.” Markus
2. Literature review (1989: 23) further suggests that in exploratory research, clearly
stating the unit of analysis “helps to define the boundaries of a
Across various management disciplines there has been recog- theory which in turn set the limitations in applying the theory.” A
nition of the importance of bringing clarity and increased rigor to clearly stated unit of analysis can help identify applicable extant
theory building and testing from case studies. A number of articles literature that can help clarify the phenomenon under investiga-
have attempted to provide guidance as to how to undertake such tion.
research from a variety of disciplines—management (Bitektine,
2008; Eisenhardt, 1989; Eisenhardt and Graebner, 2007; Harris and 2.1.3. Research purpose and role of existing theory
Sutton, 1986; Langley, 1999; Yin, 1989, 1994), information sys- Case studies are used primarily to develop new theories (e.g.,
tems (Benbasat et al., 1987; Cavaye, 1996; Lee, 1989), marketing Benbasat et al., 1987; Gersick, 1988; Harris and Sutton, 1986; Van
(Bonoma, 1985; Hillebrand et al., 2001; Johnston et al., 1999), and de Ven, 1989). Researchers employ an inductive logic, utilizing a
operations management (Meredith et al., 1989; Stuart et al., 2002; variety of methods to collect primarily qualitative data from which
Voss et al., 2002). Based on this extant literature, we identify the key to develop relevant and testable theories (Eisenhardt and Graebner,
areas of consideration (i.e., inductive or deductive) and method- 2007; Fisher, 2007; Roth, 2007; Voss et al., 2002).
ological approaches (i.e., sampling, data collection, and analysis). An important question arises then as to the role of existing
We have divided the literature review into two sections—one on theories in this theory-building process. On the one hand, the
inductive qualitative case studies and the other on deductive qual- grounded-theory approach, proposed by Glaser and Strauss (1967),
itative case studies. To the best of our knowledge, we are the first to is based on pure inductive logic, where the new theory is derived
explicitly consider both inductive and deductive case approaches strictly from the data. On the other, Eisenhardt (1989) has sug-
in a single review. In general, there has been much more extensive gested that this “clean slate” approach has generated confusion
literature discussing the former, and our review reflects this state. over role of extant literature and existing theories in the use of
case studies for theory building purposes. She has proposed that
2.1. Inductive use of qualitative case studies this approach, as implied by the grounded theory approach, is
impractical, since the study’s purpose, site selection, and data gath-
Much has been written about how to conduct an inductive ering require some rationale or preconceived ideas. Subsequently,
case study (e.g., Eisenhardt, 1989; Meredith, 1998; Yin, 1989). We a number of articles have suggested the use of a priori constructs to
looked for basic requirements that are common to the articles that help shape the initial design of theory building research (Bourgeois
have provided such guidance. and Eisenhardt, 1988; McCutcheon and Meredith, 1993; Voss et al.,
2002). However, such a priori constructs are only to be considered as
2.1.1. Justification of research approach tentative and may not be in the resultant theory (Eisenhardt, 1989;
An important consideration for undertaking theory building McCutcheon and Meredith, 1993). In the end, the higher the level
case studies is to clearly articulate the rationale behind why such of consistency between the emergent theory and existing theory,
research is being conducted (Eisenhardt and Graebner, 2007). the higher the external validity achieved.
Justifications can include: there is a gap in existing theory that
does not adequately explain the phenomenon under investigation 2.1.4. Sampling issues, case selection, and number of cases
(Benbasat et al., 1987; Eisenhardt and Graebner, 2007; Meredith, Instead of statistical sampling from the defined population, case
study researchers utilize a theoretical or biased sampling approach
where cases are chosen for theoretical reasons (Glaser and Strauss,
1
We also considered Manufacturing and Service Operations Management (MSOM)
1967; Meredith, 1998; Eisenhardt, 1989; Yin, 1989). Cases are cho-
as a sixth potential journal, but it had not, since its inception in 1999, published any sen that either predict similar results or contrary results (Yin, 1989).
qualitative case study papers that met our sampling criteria. The use of polar extreme-types has also been suggested where cases
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 331

have sharply contrasting characteristics (Miles and Huberman, is within-case analysis, where a single case description is offered and
1984; Pettigrew, 1990; Yin, 1989). Leading companies have been the emerging constructs and their relationships are delineated. At
used for the usefulness of the results for benchmarking purposes this stage, detailed, descriptive write-ups are created. Despite being
(Choi and Hong, 2002; Fisher, 2007). In this regard, when building descriptive, such case write-ups are core to the creation of insights
theory from case studies, the selection of cases should be carefully (Gersick, 1988; Pettigrew, 1990), although there are no standard-
thought out rather than opportunistically derived (Benbasat et al., ized formats for such write-ups (Yin, 1989). Case write-ups are
1987). deemed to be analysis as a result of the decisions that researchers
A question then arises as to the number of cases that researchers make as to what the emerging issues are and how they should be
should select. Voss et al. (2002), recognizing this dilemma, suggest captured in the write-ups.
that the fewer the number of cases, the greater the opportunity Cross-case analysis is the act of comparing and contrasting the
for depth of observation. However, multiple cases can augment patterns emerging from the detailed case write-ups (Benbasat et
external validity and help guard against observer bias. In partic- al., 1987; Eisenhardt, 1989; Yin, 1989). Eisenhardt (1989) warns to
ular, for theory building purposes, the use of multiple cases is guard against leaping to conclusions based on limited data. Nisbett
likely to create more robust and testable theory than single case and Ross (1980) and Miles and Huberman (1984) also warn against
research (Eisenhardt, 1989; Eisenhardt and Graebner, 2007; Yin, allowing the vividness or status of respondents to unfairly influence
1994). Eisenhardt (1989: 15) specifically suggested that in the the case write-ups. Several articles (e.g., Eisenhardt, 1989; Miles
range of 4–10 cases “usually works well.” She cautioned that if and Huberman, 1984; Yin, 1989) articulate how to engage in cross-
less than four it may become difficult to capture the complexity case analysis and overcome these pitfalls. The researchers should
of the real world and if more than 10 it may become difficult for select two cases at a time and compare them noting the differences
the researchers to cognitively process the information. Dyer and and similarities and repeat this procedure until all cases have been
Wilkins (1991) countered this suggestion by arguing that single considered. Alternatively, they may select a few constructs based
case studies enable the researcher to capture in much more detail on the extant literature that describes the phenomenon of inter-
the context within which the phenomena under study occur. Single est and then look for the evidences that address these constructs.
case studies may be useful for longitudinal research (Narasimhan Ultimately, the focus is on looking for similar patterns. But when
and Jayaram, 1998; Voss et al., 2002) and can be used if they are differences occur they are not discarded until the uniqueness of
extreme exemplars or opportunities for unusual research access the situation is clarified as having contributed to the difference.
(Yin, 1994). Another approach is to divide the data by its source (Bourgeois and
Eisenhardt, 1988; Eisenhardt and Bourgeois, 1988). Findings from
2.1.5. Data collection and analysis one form of data source are then corroborated by similar findings
There are several data sources: interviews either structured from another form of data.
(interview tool remains fixed) or semi-structured (interview tool
is updated based on emerging data), observations (e.g., plant tour, 2.1.7. Presentation of research outcomes
attendance at meetings), and archival sources (e.g., documents, One of the ongoing challenges with case study research is how
historical records, organizational charts, and production statis- to present the research outcomes (Eisenhardt and Graebner, 2007;
tics). While some researchers have used only one method (e.g., Miles and Huberman, 1984)—more specifically, how to draw and
observation in Gersick, 1988), others have used multiple methods validate conclusions from data analysis (Miles and Huberman,
for the purpose of “triangulation” of data from different sources 1984). It is not an easy task for researchers to present their data
(e.g., Eisenhardt, 1989; Choi and Hong, 2002). Using multiple and defend the process, in the form of a “chain of evidence” which
data sources (Jick, 1979) provides increased reliability of data the reader can “readily follow” (Benbasat et al., 1987) as to how the
(Benbasat et al., 1987; Boyer and McDermott, 1999; Hyer et al., researchers arrived at their research outcomes from the data that
1999; Leonard-Barton, 1990) and stronger substantiation of con- was collected.
structs and propositions (Benbasat et al., 1987; Eisenhardt, 1989; Miles and Huberman (1984) have proposed numerous tactics
Voss et al., 2002). Another form of triangulation is the use of multi- for drawing case conclusions and to present findings. However, as
ple investigators (Dubé and Paré, 2003; McCutcheon and Meredith, far as the documentation of the report is concerned, many authors
1993). Benbasat et al. (1987) and Eisenhardt (1989) suggested that agree that there is no standard form (Miles and Huberman, 1984;
the use of multiple investigators leads to a better ability to handle Yin, 1989, 1994). For single-case studies it has been suggested
the richness of the contextual data and more confidence in research that researchers present a detailed narrative supported by quota-
findings. tions from key informants and other forms of evidence (Yin, 1989).
At the core of theory building is data analysis (Dubé and Paré, The narrative is thought to be closely intertwined with the theory
2003; Eisenhardt, 1989; Glaser and Strauss, 1967; Stuart et al., (Eisenhardt and Graebner, 2007). For multiple cases this challenge
2002; Yin, 1989). It needs to occur simultaneously and incremen- becomes even more critical and difficult. It requires a careful craft-
tally with data collection (Glaser and Strauss, 1967). Obtaining ing and presentation of data to make the outcome self-evident to
overlap between data collection and analysis allows the researchers the readers. The use of tables and visual displays is often promoted
to capture the reality that the data bring (e.g., McCutcheon and as the way to convey and summarize the rich empirical evidence
Meredith, 1993). Constructs and their relationships are adjusted as within case studies (Eisenhardt and Graebner, 2007; Miles and
data are collected. Such adjustments may come from the addition Huberman, 1984; Voss et al., 2002).
of cases to pursue a particular emerging theme (e.g., Gersick, 1988),
the addition of questions to an interview protocol (e.g., Harris and 2.2. Deductive use of qualitative case studies
Sutton, 1986), and the addition of data sources in existing case
studies (e.g., Burgelman, 1983; Sutton and Callahan, 1987). Although the majority of the OM case studies have taken the
inductive, theory-building approach, a small number of authors
2.1.6. Organization of results have proposed their use for deductive, theory-testing purposes
The biggest challenge behind data analysis is to demonstrate (McCutcheon and Meredith, 1993; Meredith, 1998; Voss et al.,
the objectivity of the process through which the data and field 2002). This proposal to test existing theory is in line with other
notes are developed into conclusions (Eisenhardt, 1989; Miles and studies in the general business disciplines: management (Bitektine,
Huberman, 1984; Van Maanen, 1988). The first step in this process 2008; Bryman, 1988; Eisenhardt, 1989; Langley, 1999; Pinfield,
332 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

1986; Yin, 1994); information systems (Benbasat et al., 1987; ateness of competing theories (Johnston et al., 1999; Keil, 1995;
Cavaye, 1996; Darke et al., 1998; Lee, 1989); and marketing Pinfield, 1986) and utilizing a longitudinal approach (Anderson,
(Bonoma, 1985; Hillebrand et al., 2001; Johnston et al., 1999). How- 1983; Bitektine, 2008; Eisenhardt, 1989). Both approaches adopt
ever, while we agree that qualitative case studies can be used for the logic of confirmation/falsification of the appropriateness of a
deductive purposes, it should be noted that despite the number of theory discussed above. They entail articulating a theory into a
authors that propose and support the use for such purposes only a set of hypotheses and then comparing them against data either
few offer insights as to how to actually undertake such research to confirm or falsify them. Key here is to devise ways to main-
(Bitektine, 2008; Pinfield, 1986; Johnston et al., 1999). As such, tain objectivity and to guard against researchers’ personal bias. The
the literature for deductive use compared to the inductive use of competing theories approach helps researchers minimize personal
qualitative case studies is in need of further development. bias that may enter into the analysis by selectively looking for evi-
dence that fit the hypotheses. The presence of competing theories
2.2.1. Standing criticisms and potential solutions would force the researchers to choose one theory over another. The
There have been criticisms for using qualitative case studies for longitudinal approach is similar to the “prospective case design,”
deductive, theory-testing purposes (Bitektine, 2008; Hillebrand et which is borrowed from the medical field (Bitektine, 2008). Here,
al., 2001; Johnston et al., 1999). These criticisms may have simply hypotheses are formulated first in a prospective manner and then
arisen from the lack of familiarity of qualitative methods (Bitektine, the qualitative data are collected at in different points across time.
2008; Roth, 2007); nevertheless, many researchers trained in posi- Pinfield (1986) demonstrated an approach to enable the com-
tivist traditions have criticized theory-testing based on qualitative parison and evaluation of two theoretically-derived perspectives of
case studies on the grounds of “ambiguity of inferred hypotheses” the organizational decision-making process (see also Keil, 1995).
and the “selective bias” (Bitektine, 2008: 161). Here, the concern This approach was loosely based on two earlier approaches used
is over the degree of freedom that a researcher has to formulate by March and Olsen (1976) and Kagan (1978). Following the care-
hypotheses and the natural inclination to peek into the data. An ful description of the two competing decision-making perspectives
additional concern is the risk of selectively looking for evidence (e.g., structure vs. anarchic) across multiple dimensions, Pinfield
that fit the a priori stated hypotheses. (1986) collected data from multiple (four) sources to avoid inter-
The over-arching approach that has been proposed for the pretive bias. Utilizing a single case study of a complex decision
deductive use of qualitative case studies is that of confirmation (or making process within a single organization, five dimensions were
falsification) of the appropriateness of a theory (Bonoma, 1985; drawn from the theoretically derived perspectives and considered
Bryman, 1988; Johnston et al., 1999; Ross and Staw, 1993; Yin, in the analysis—decision definition, goals and technology, partic-
1994). Johnston et al. (1999) proposed three main requirements for ipation, contextual dependence and time. In doing so the author
using qualitative case studies for such confirmation purposes: (1) made qualified recommendations as to the most appropriate the-
the case study must begin with an existing theory for the develop- oretical perspective.
ment of research hypotheses; (2) a systematic and logical research The use of longitudinal data builds on the application of the
design should be followed; and (3) researchers should implement principles of prospective study design adapted from the field of
evaluation criteria to independently assess potential biases and to medicine where cases are used to investigate suspected ailments.
ensure the methodological rigor. These requirements are founded It uses “a comparison of a pattern of observed outcomes (on sev-
on the assertion that case studies are not to be viewed as “sampling eral variables) with some pattern of expected values derived from
units” in inferential statistics but rather as “individual studies” that a given theory” (Bitektine, 2008: 162). This approach is akin to
are used to confirm or falsify a theory (Cavaye, 1996; Yin, 1994). In Popper’s (1968) approach to falsifying theory—using a proposition
other words, lack of generalizability to the sampling population is under consideration to “predict outcomes for specific cases and
not of main concern. What is important is the contextual data from subsequently investigate these cases to see whether the theory
case studies that are used to confirm or falsify a theory. holds true for them” (Hillebrand et al., 2001: 652). The “falsifica-
Following the development of the hypotheses, the systematic tion” of the theory arises when the theory under consideration does
research design should incorporate: the clear definition of the unit not hold true for the predicted outcomes. This pattern-matching
of analysis, the careful selection of appropriate cases (Johnston et technique (Campbell, 1966; Yin, 1994) allows for “outcome eval-
al., 1999) and triangulated data sources driven by the nature of the uation on multiple dimensions, where as little as one actual
specific research questions (Bonoma, 1985; Yin, 1994). Concern- observation for a given dimension is available” (Bitektine, 2008:
ing the selection of cases, the authors recognize that while a single 162).
case is possible, multiple cases are more compelling and make the In fact, both approaches of competing theories and longitudinal
research more robust. They also suggest that the cases be cho- data utilize pattern matching (Campbell, 1966; Yin, 1994). All that
sen to complement each other, in terms of similar contexts and or the pattern matching requires is “a theoretical pattern of expected
polar extreme types or to specifically investigate rival hypotheses outcomes, an observed pattern of effects, and an attempt to match
(Johnston et al., 1999). the two” (Trochim, 1989, p. 360). In essence, the researchers are
To overcome potential researcher bias, the issues of internal and looking for patterns in the emerging data and then comparing the
external validity need to be considered together with reliability and patterns against the theoretically derived hypotheses. Because the
objectivity (Johnston et al., 1999; Yin, 1994). One possible approach focus is on these patterns involving a priori determined constructs,
that can be utilized here is the use of multiple researchers (see also there is less opportunity for making Type 1 error (i.e. false positive).
Dubé and Paré, 2003; McCutcheon and Meredith, 1993), similar to In competing theories, emerging patterns are compared against
the suggestions by Benbasat et al. (1987) and Eisenhardt (1989) multiple theories, and in use of longitudinal data, emerging pat-
for inductive research, which leads to a better ability to handle the terns are compared against the prescribed theories over time. For
richness of the contextual data and more confidence in research example, Keil (1995) in examining the escalation of commitment
findings. in information systems tested three theories of escalation and, in
doing so, used the theories as a template (i.e., stated hypotheses)
2.2.2. Two proposed approaches: use of competing theories and for pattern matching. Also, Lee et al. (1996) in testing a model of
longitudinal data voluntary employee turnover used pattern matching when the the-
Eisenhardt (1989) suggests that the confirmation approach orized essential features for a given decision path are judged to
can take two subsequent forms: namely examining the appropri- occur across multiple cases.
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 333

Fig. 1. Article sampling criteria and their application using Decision Sciences as an example.

2.3. Summary of the literature review of qualitative case studies (i.e., International Journal of Operations
and Production Management).
It is clear from the literature reviews in Sections 2.1 and 2.2 that
the methodology for inductive case study purposes is significantly
3.1. Time horizon and article sampling
more developed and comprehensive than that for deductive case
study purposes. This disparity between the levels of sophistication
The time horizon for our research is 16 years, beginning in 1992,
of the two research approaches is further evidenced by our anal-
when POM was inaugurated, to 2007. We reviewed all the publi-
ysis of the deductive articles in our sample that appear to have
cation issues of the five journals during the specified time period.
adopted an inductive logic and applied this to case studies for the-
There were in total 5526 articles published from 1992 to 2007. We
ory testing purposes. For instance, we found only three papers that
excluded any editorial articles or corrections to earlier articles. Fig. 1
partially followed the longitudinal approach and two papers that
illustrates the process of sampling using DS as an example. In this
partially followed the competing theories approach out of a total of
process, we first looked to see if the paper used some form of qual-
35 deductive papers. Therefore, we have classified the research out-
itative methods. We then applied our definition of qualitative case
comes of these articles based on their stated approach and claimed
studies as articulated previously. Initially, we found 461 case-based
outcomes rather than based on the emerging protocols we learned
articles that used either qualitative or quantitative or a mixture of
from our literature review. Such outcomes from the published arti-
both methods (for references, see under Supplemental Materials at
cles range from “confirmation/falsification” to “revised frameworks
http://www.journaloperationsmanagement.org/). By utilizing our
or hypotheses” to “descriptive insights.”
sampling criteria we ended up with the total of 204 qualitative
case studies—two from DS, 150 from IJOPM, 32 from JOM, 11 from
POM, and nine from MS (for references, go to the same web address
3. Methodology
shown above).
Our goal is to report on the state of qualitative case study
research. The published articles in five journals that met our defi- 3.2. Coding
nition of qualitative case studies became our data source. The five
journals listed were selected on the basis of quality and impact (e.g., We have coded all 204 qualitative case studies by the cod-
Journal of Operations Management, Management Science, etc.). Con- ing criteria shown in Table 1. Two of the researchers coded
siderations were also given to geographic coverage (i.e., we wanted all of the articles and then the third researcher reviewed all
to include a leading European journal) and their stated acceptance of the coding. Through the coding process, inter-coder agree-
334 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

Table 1
Coding criteria.

Coding criteria Description of criteria

Year of publication In what years were the articles published?

Article authors Who were the authors of the article?
Journal Which of the five journals was the article published in?
Major focus of article What was the major focus of the article?
Unit of analysis (UofA) What unit of analysis was adopted by the case study(ies) (Yin, 1989)?
Statement of unit of analysis To what extent was the unit of analysis clearly stated?
Justification of research approach To what extent had the authors justified their choice of research approach?
Mode of research Did the authors follow an Inductive or deductive mode of research?
Degree of alignment Bearing in mind the stated research goal, how appropriate were the selected cases?
Theoretical lens Which theories or literatures influenced the authors’ framing of the research?
Methodology Did the authors use a single, multiple or longitudinal case study approach?
Sampling strategy What was the logic behind the case sampling—theoretical or convenience?
Case selection strategy If theoretical sampling was adopted, what kinds of strategies were used to select cases?
Number of cases How many cases were selected for the research?
Data sources Were there multiple sources of data, such as interviews, observations, and/or documents?
Data source triangulation How much data triangulation was adopted and in what form?
Role of existing theories Were existing theories used to develop constructs and/or used to examine the findings?
Data analysis To what extent were within and cross-case analyses carried out?
Research outcomes What types of findings were produced, such as frameworks, propositions or descriptive insights?

ments remained in the 80–85% range. Any issues or exceptions

were discussed and resolved by all of the researchers through
consensus.
Once we were able to ascertain the unit of analysis, we examined
each article and determined whether the article provided sufficient
justification for the adoption of a case study as the selected research
approach. We determined what primary research topic that the
article was focusing on (which is presented in Table 3 below) and
whether the articles had clearly stated the unit of analysis and in
what context it occurred. Regarding the use of existing theories,
we looked to see if any theories were being used to frame the
research, or if no theories were used, then what bodies of literature Fig. 2. Number of qualitative case studies vs. total number of articles.
were being used. We also examined each article to determine what
sampling approach had been adopted. We determined how many
cases were being used. We reviewed the data collection techniques 4. Analysis, results and implications
used and examined the data analysis approaches deployed. Lastly,
we reviewed the research outcomes of each article. Appendix In this section we present the analysis and general trends. We
A contains a more detailed explanation of the evaluation cri- focus on the differences between the inductive and deductive case
teria, the scales that were adopted, and the rationale for the studies, in terms of the research outcomes and the underlying
scales. methodological issues that relate to the rigor of such studies.

3.3. Analysis approach 4.1. General trends for qualitative case studies

We summarized all coding results on a large spreadsheet. Over the period 1992–2007, as evidenced in Fig. 2, qualitative
Classifying qualitative case studies by their research orientation case studies constitute a very small portion of the published papers.
(i.e., inductive or deductive) and the form of their research out- Nonetheless, while the total number of articles shows in general a
comes (i.e. frameworks, propositions, or insights), we reviewed decreasing trend, the number of case studies shows an increasing
each category in depth. We looked for common patterns across trend. The growth has come slowly but steadily from an average of
each of the categories, but whenever a deviation occurred, we 1.4% of the total number of articles published in these journals for
investigated it further (Poole and Van de Ven, 1989). To reduce 1992–1996 to an average of 6.8% for 2003–2007.
all research topics to a few categories, the three researchers
engaged in a Q-sort activity. Each person independently grouped 4.1.1. Qualitative case studies by journal
the individual topics into a smaller number of key categories, and Of the five journals in Table 2, IJOPM has published the largest
then all three discussed discrepancies together. Overall, we found number of qualitative case studies (150), followed by JOM (32), then
about 85% agreement among the sorters. Any issues or excep- by POM (11), MS (nine) and lastly DS (2). The post hoc column in
tions were discussed and resolved by all of the researchers through Table 2 captures articles that took one approach in the main body
consensus. of research and then adopted the other in a post hoc analysis.
The purpose of the present research is descriptive and induc- As seen in Fig. 3, it appears that 2000 was a watershed year for
tive and was not conducive to inferential statistics. Also, the the qualitative case study methodology in IJOPM, when 15 papers
research entailed census rather than survey. We have used a were published. Besides IJOPM, JOM and POM have been taking the
qualitative trend and pattern analysis to develop a greater under- lead in publishing qualitative case studies. JOM has been consis-
standing of contributions from qualitative case studies in OM tently publishing case studies, at least one article every year. POM
and to identify potential opportunities for improvement. We have published qualitative case studies actively from 1996 to 2007, with
presented these analytical results in the forms of tables and the exception of 1999–2000 and 2005–2006 when no qualitative
figures. case studies were published.
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 335

Table 2
Case studies by Journal and Research Orientation.

Journal Total Inductive (%) Deductive (%) Adoption of post hoc analysis (Ind-Ded or Ded-Ind)

DS 2 2 1 0 0.0 None
IJOPM 150 119 58.3 31 15.2 2 Ind-Ded, 6 Ded-Ind
JOM 32 29 14.2 3 1.5 2 Ind-Ded, 1 Ded-Ind
MS 9 8 3.9 1 0.5 None
POM 11 11 5.4 0 0.00 1 Ind-Ded

Total 204 169 82.8 35 17.2

cles with rich areas for theory building in previously well studied
areas.

4.1.3. Qualitative case studies by research outcomes

We examined the qualitative case studies for their research
outcomes, as shown in Table 3. For 169 inductive articles, 38 or
22.5% produced forms of frameworks or models. For example, Wu
and Choi (2005) developed a typology of supplier–supplier rela-
tionship configurations, supported by eight propositions related
to the interactions between competing suppliers. In Danese et al.
(2006) a model of the sequences of improvements in pharmaceuti-
Fig. 3. Number of qualitative case studies by year. cal supply networks was developed. Thirty-four or 20.1% developed
formally stated propositions; for example, Grutter et al. (2002)
developed nine propositions relating to work team performance in
4.1.2. Qualitative case studies by topic South African manufacturers. An additional example can be found
Based on the identification of the primary topic for each article, in Krajewski et al. (2005), where eight propositions were devel-
Table 3 presents the 14 major OM topic areas that used a qualitative oped relating to the reaction strategies adopted by suppliers in
case study method. The area with the most qualitative case study build-to-order supply chains.
publications is manufacturing strategies. Initially, the high number For deductive-oriented qualitative articles, twelve or 34.3% pro-
of qualitative case study publications in the area of manufacturing vided forms of confirmation/falsification of theoretically derived
strategies seemed to be counter-intuitive. Qualitative case studies hypotheses. For example, in Jensen and Szulanski (2007), the
are typically used for exploring an area not previously studied, yet article’s original hypotheses were confirmed (i.e., that the use
the topic of manufacturing strategies in general has been studied of templates increases the effectiveness of knowledge transfer).
intensively for several decades (e.g., Hayes and Wheelwright, 1979; In Lewis (2000), the author refuted their initial hypothesis that
Miller and Roth, 1994; Skinner, 1980). However, a closer look at the becoming lean does not automatically result in improved finan-
focus of study revealed that a large number of qualitative case stud- cial performance for an organization. Ten or 28.6% provided revised
ies in this area were integrative in nature, for instance, combining hypotheses/frameworks as their research outcomes. For example,
theories in manufacturing strategies with other areas such as con- in Bititci et al. (2005), revised hypotheses concerning how existing
tingency theory (Sousa, 2003; Sousa and Voss, 2001), modularity performance measure can be used for measuring performance in
(Salvador et al., 2002), and engineering (Narasimhan and Jayaram, extended enterprises were produced. In Mosey (2005) the author
1998; Voss and Winch, 1996). The integrative studies provided arti- produced a revised framework for understanding how small and

Table 3
Research outcomes by topic and mode of research.

Topic Inductive research outputs (n = 169) Deductive research outputs (n = 35) Overall Totals

Framework Propostions Descriptive Inductive Confirm vs. Revised Descriptive Deductive

insights sub-totals falsify framework or insights sub-totals
hypotheses

n (%) n (%) N (%) n (%) n (%) n (%) n (%) n (%) n (%)

Manufacturing strategy 11 6.5 1 4.1 42 24.9 60 35.5 5 14.3 5 14.3 4 11.4 14 40.0 74 36.3
Org. behavior 4 2.4 2 1.2 6 3.6 12 7.1 3 8.6 0 0.0 3 8.6 6 17.1 18 8.8
Integration 4 2.4 4 2.4 6 3.6 14 8.3 2 5.7 1 2.9 0 0.0 3 8.6 17 8.3
Strategic sourcing 7 4.1 2 1.2 6 3.6 15 8.9 0 0.0 2 5.7 0 0.0 2 5.7 17 8.3
Perf. measurement 1 0.6 1 0.6 10 5.9 12 7.1 0 0.0 2 5.7 1 2.9 3 8.6 15 7.4
Service operations 2 1.2 6 3.6 3 1.8 11 6.5 0 0.0 0 0.0 3 8.6 3 8.6 14 6.9
Demand chain mgmt. 4 2.4 2 1.2 4 2.4 10 5.9 0 0.0 0 0.0 0 0.0 0 0.0 10 4.9
Knowledge mgmt. 1 0.6 2 1.2 5 3.0 8 4.7 1 2.9 0 0.0 0 0.0 1 2.9 9 4.4
Plant mgmt. 1 0.6 1 0.6 6 3.6 8 4.7 1 2.9 0 0.0 0 0.0 1 2.9 9 4.4
Supply chain mgmt. 1 0.6 2 1.2 4 2.4 7 4.1 0 0.0 0 0.0 1 2.9 1 2.9 8 3.9
Environmental mgmt. 0 0.0 4 2.4 3 1.8 7 4.1 0 0.0 0 0.0 0 0.0 0 0.0 7 3.4
Inventory mgmt. 0 0.0 1 0.6 2 1.2 3 1.8 0 0.0 0 0.0 0 0.0 0 0.0 3 1.5
Project mgmt. 2 1.2 0 0.0 0 0.0 2 1.2 0 0.0 0 0.0 0 0.0 0 0.0 2 1.0
Retail strategy 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1 2.9 1 2.9 1 0.5

Totals 38 22.5 34 20.1 97 57.4 169 82.8 12 34.3 10 28.6 13 37.1 35 17.2 204 100
336 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

Fig. 4. Qualitative case study articles—inductive vs. deductive (1992–2007).

Fig. 6. Deductive articles (by research outcome) fulfilling methodological design
medium-sized enterprises (SMEs) develop a dynamic capability for issues.

new-to-market product development. frameworks provided at least partial or full justification for their
choice of research approach compared to only 50 of 97 (i.e., 51.5%)
4.1.4. Qualitative case studies by research orientation of articles that produced descriptive insights. Fig. 5 would suggest
Of the 204 qualitative case studies included in this study, that, generally speaking, the articles in the “Framework” and
most adopted the inductive approach. Ones that used deductive “Proposition” research outcome categories were methodologically
approach were clearly in a minority. As shown in Table 3 above, 169 more rigorous than the articles in the “Descriptive Insight” research
(82.8%) were inductive (i.e., theory building) and 35 (17.2%) were outcome category.
deductive papers (i.e., theory testing). We note that 31 of the 35
deductive, theory-testing articles came from IJOPM. Nevertheless, 4.2.1.1. Justification of research approach. The majority (102 of 169,
the number of deductive case studies exceeded our expectations, i.e., 60.4%) of inductive case studies provided at least some justifica-
especially given the criticisms surrounding this approach as dis- tion for their choice of the case study methodology as their research
cussed under the literature review. As evidenced by Fig. 4, while approach and why they were undertaking an inductive study, as
the number of qualitative case studies that adopted the inductive recommended by Yin (1989) and Eisenhardt and Graebner (2007).
approach has increased significantly since 2000, the number of case It should be noted that the majority of cases that did not justify
studies that adopted the deductive approach has shown a slow their use of case study method were from IJOPM. Justification of
decline. the case methodology is associated with certain types of research
outcomes. For cases that did not justify the use of case method-
4.2. Inductive vs. deductive case studies: key patterns and ology, about half (i.e., 48.5%) ended with descriptive insights only.
differences For cases that did justify the use of case methodology, only about
a quarter (i.e., 27.8%) ended with descriptive insights. The remain-
We now offer the results of a more detailed analysis, based on ing majority of articles developed either a theoretical framework
Table 4. While the case studies have produced some significant con- or set up formal propositions. This pattern of relationship between
tributions in terms of their research outcomes, all articles have been the justification of case approach and research outcome may have
examined for potential ways to improve quality and rigor. They suggested that case approach justification is one of the salient indi-
are examined in light of the differences between the research out- cators of an overall rigorous case study design which produces
comes and what our earlier literature review informed us about the meaningful results.
requirements of a scholarly case study.
4.2.1.2. Unit of analysis (UOA). Overall, for inductive articles there
are more case studies that did not clearly state their UOA (76.9%)
4.2.1. Inductive qualitative case study papers
than those that did (23.1%). This is seen as a potentially significant
Table 4 is reframed into Figs. 5 and 6. According to Fig. 5, the
area for improvement because clearly stated unit of analysis makes
research outcome categories of the “Framework” and the “Propo-
a difference in terms of the resulting research outcomes (Dubé and
sition” consistently scored higher, in terms of the percentage of
Paré, 2003; Markus, 1989; Yin, 1989). A higher percentage of stud-
articles that met the research design criteria, than the “Descriptive
ies that clearly stated their UOA were able to derive frameworks
Insight” research outcome category over all of the methodolog-
or proposition compared to those who produced only descriptive
ical design issues. For example, in terms of justification of case
insights (56.4% vs. 38.5%).
research approach, 28 of 34 (i.e., 82.4%) of articles that produced
propositions and 24 of 38 (i.e., 63.2%) of articles that produced
4.2.1.3. Theory vs. phenomenon. Overall, a small percentage of case
studies used an existing theory as opposed to a phenomenon occur-
ring in the literature to frame the research. An existing theory (i.e.
transaction cost economics, resource based view, etc.) adds validity
to the conclusions one may draw from the data whether induc-
tive or deductive, whereas a phenomenon pertains to the specific
context in which the case studies are conducted (i.e. service oper-
ations, knowledge management, etc.). According to Table 4, case
studies that focused on a phenomenon occurring in the literature
increased the likelihood of deriving descriptive insights as opposed
to a framework or formal propositions.

4.2.1.4. Sampling approach. Discussion of the sampling approach

Fig. 5. Inductive articles (by research outcome) fulfilling methodological design
is universally important and is explicitly expressed across different
issues. methodologies. For the case articles we reviewed, the majority used
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 337

Table 4
Summary comparison of inductive and deductive research outputs (by methodological issues).

Inductive research outcomes (n = 169) Deductive research outcomes (n = 35)

Framework or Descriptive Total Confirm/falsify Rev frame/hypos or Total

propositions insights descriptive insights

n (%) n (%) n (%) n (%) n (%) n (%)

Justified research approach

Yes 19 11.2 13 7.7 32 18.9 1 2.9 5 14.3 6 17.1
Partial 33 19.5 37 21.9 70 41.4 3 8.6 10 28.6 13 37.1
No 20 11.8 47 27.8 67 39.6 8 22.9 8 22.9 16 45.7

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Unit of analysis
Clearly stated 22 13.0 17 10.1 39 23.1 0 0.0 6 17.1 6 17.1
Not clearly stated 50 29.6 80 47.3 130 76.9 12 34.3 17 48.6 29 82.9

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Theory vs. phenomenon

Theory 8 4.7 7 4.1 15 8.9 0 0.0 4 11.4 4 11.4
Phenomenon 62 36.7 84 49.7 146 86.4 12 34.3 19 54.3 31 88.6
Neither 2 1.2 6 3.6 8 4.7 0 0.0 0 0.0 0 0.0

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Sampling strategy
Theoretical 58 34.3 62 36.7 120 71.0 6 17.1 12 34.3 18 51.4
Convenience 5 3.0 13 7.7 18 10.7 5.7 6 17.1 8 22.9
Random 1 0.6 1 0.6 2 1.2 0 0.0 1 2.9 1 2.9
No logic offered 8 4.7 21 12.4 29 17.2 4 11.4 4 11.4 8 22.9

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Number of cases
1 11 6.5 34 20.1 45 26.6 6 17.1 9 25.7 15 42.9
2 4 2.4 6 3.6 10 5.9 1 2.9 1 2.9 2 5.7
3 8 4.7 13 7.7 21 12.4 2 5.7 2 5.7 4 11.4
4–10 37 21.9 35 20.7 72 42.6 2 5.7 7 20.0 9 25.7
>10 12 7.1 9 5.3 21 12.4 1 2.9 4 11.4 5 14.3

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Triangulated data sources

Yes 56 33.1 68 40.2 124 73.4 10 28.6 14 40.0 24 68.6
No 15 8.9 22 13.0 37 21.9 0 0.0 9 25.7 9 25.7
Not stated 1 0.6 7 4.1 8 4.7 2 5.7 0 0.0 2 5.7

Totals 72 42.6 97 57.4 169 100.0 12 34.3 23 65.7 35 100

Data analysis
Within & cross-case 45 26.6 31 18.3 76 45.0 5.7 7 20.0 9 25.7
Within-case only 12 7.1 41 24.3 53 31.4 6 17.1 10 28.6 16 45.7
Cross-case only 11 6.5 22 13.0 33 19.5 4 11.4 6 17.1 10 28.6
None 4 2.4 3 1.8 7 4.1 0 0.0 0 0.0 0 0.0

Totals 72 42.6 97 57.4 169 100 12 34.3 23 65.7 35 100

theoretical sampling (71%), the remaining used approaches rang- as suggested by Eisenhardt (1989). Of 76 articles that used
ing from convenience sampling (10.6%) to random sampling (0.6%). less than 4 cases, 34 (20.1%) articles used a single case, all
We should note that 29 cases (17.2%) did not mention their sam- of which produced descriptive insights. For 21 (12.4%) arti-
pling logic at all. Of the cases that did not use theoretical sampling, cles that used over ten cases, there was a tendency to
72.4% of these cases only produced descriptive insights. In com- derive more framework/proposition than purely descriptive
parison, of the case studies that adopted theoretical sampling, only insights.
51.7% resulted in producing descriptive insights. All of the 21 arti-
cles (12.4%) under descriptive insights that did not offer sampling
logic came from IJOPM. 4.2.1.6. Data sources. Overall, 124 out of 169 (73.4%) inductive arti-
cles used triangulated data sources. In terms of research outcomes,
the use of triangulated data sources had a big impact on the devel-
4.2.1.5. Number of cases. Seventy-two (42.6%) articles were in opment of formal propositions or frameworks, whereas only 22.2%
line with the ideal number (i.e. between 4 and 10 cases), (16 of 72) of articles did not use triangulation. This percentage is
338 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

higher for cases that produced descriptive insights at 30%, i.e. 29 of 4.2.2.3. Theory vs. phenomenon. Surprisingly, most studies (31 or
97 articles. 88.6%) used an emerging phenomenon derived from the extant lit-
erature. Existing theory was only used in four (11.4%) of deductive
4.2.1.7. Data analysis. For inductive studies that used both within papers with none of them being able to provide confirmation or fal-
and cross-case analysis, the majority of them (45 out of 76 or 59.2%) sification of the selected theory, or even revision of the framework
were able to derive either a theoretical framework or formal propo- or hypotheses drawn from the theory. Instead they all derived only
sitions. This percentage is much lower for cases that used only descriptive insights.
within case or only cross-case analysis.
4.2.2.4. Sampling approach. For deductive case studies, the major-
4.2.1.8. Summary of inductive qualitative case study papers. The ity of articles (18 or 51.4%) used theoretical sampling (51.4%),
inductive articles that were more rigorous with their research followed by convenience sampling (8 or 22.9%) and no logic offered
design managed to produce frameworks or propositions as an out- (8 or 22.9%) and random sampling (1 or 2.9%). No clear patterns
come of their research compared to those that were less rigorous. emerged in terms of the sampling approach and the resulting
When examining the differences between the three main research research outcomes.
outcomes (i.e. frameworks, propositions and descriptive insights),
we found three key drivers: (1) justification for choice of case- 4.2.2.5. Number of cases, data sources and data analysis. For deduc-
based research methodology, (2) clearly stated unit of analysis, tive case studies, 15 (42.9%) articles used a single case, in line
and (3) the use of multiple case studies leading to both within with the notion of confirming or falsifying an existing the-
and cross-case analysis. All of these three drivers represent sig- ory. There are no distinctive patterns relating the number of
nificant opportunities for improving the methodological rigor and cases used and the research outcomes for deductive case stud-
contributions. Along with the three drivers we also found that ies. Overall, deductive case studies appear to have adopted an
case based studies are doing better with issues relating to the use inductive logic for the presentation of their data analysis. While
of theoretical sampling, the use of triangulated data sources, the only a single case is needed for confirmation/falsification pur-
choice of an appropriate number of cases with which to under- poses, additional cases can certainly add further plausibility to
take their research, and the use of theory and phenomena drawn the conclusion. Of the 35 total deductive cases, 9 (25.7%) pre-
from the literature in framing their research. Finally, in terms of sented within-case combined with cross-case analysis, and 10
the research outcomes, the articles that produced only descriptive (28.6%) presented only their cross-case analysis. As such, these
insights represent a significant opportunity for potential improve- articles employed an inductive logic based on replication, rather
ment. While making some basic contribution in terms of the insight than treating each case as a separate “confirming/falsifying”
they provide, further work remains to move the theory building entity.
element of these articles to the point where they can begin to be
tested. 4.2.2.6. Summary of deductive qualitative case study papers. Our
research reveals a more troublesome picture in terms of deduc-
4.2.2. Deductive case study papers tive case-based research. Given the lack of specific guidelines in
A more confusing picture is shown in Fig. 6. We could not the literature, it is surprising to see 35 of 204 papers attempting
detect any patterns between the methodological design categories to utilize case-based research for theory-testing purposes. At the
and the level of rigor. Under justification of research approach, same time, it is unfortunate to see that in many instances arti-
sampling strategy, unit of analysis and theory vs. phenomenon, cles appear to have simply adopted an inductive logic for their
the descriptive insight articles appear more rigorous compared deductive research. This apparent decision manifested itself into
to the other two research outcome categories. Then, the situa- some serious “methodological” problems related to the justifica-
tion is reversed for the remaining three categories—number of tion of the research approach, unit of analysis, the use of theory vs.
cases, data source triangulation and data analysis. Perhaps, this phenomenon, and data analysis.
was to be expected given there are no explicit guidelines pub-
lished for using case studies for deductive, theory-testing purposes. 5. Discussion and implications for future research
We also note that there are articles that unquestioningly apply
an inductive logic to deductive case studies, for instance, using In response to the calls for more qualitative case studies, the OM
within-case and then cross-case analysis rather than treating field has seen a slowly but steadily increasing trend since 1992.
each case as a separate entity for confirmation/falsification pur- The trend is especially notable in IJOPM and JOM, although the
poses. percentage of case articles compared to total articles still remains
small. Based on our observations, we believe the OM field will con-
4.2.2.1. Justified research. For theory-testing, a survey methodol- tinue to see increasing numbers of qualitative case studies, and if
ogy is typically considered a preferred choice and it is clearly not so, it becomes more imperative that the OM field takes stock of
intuitive to use case studies for such a purpose. Therefore, provid- what we have done as a field and clearly delineate the areas of
ing methodological justification for using case studies for deductive improvement.
purposes would be even more important than it would be for induc- We offer what we perceive to be the typical profiles of case stud-
tive research. However, over 45% of the deductive articles did not ies using inductive and deductive approaches. We then conclude by
offer any justification for using case-based research, compared to making suggestions for future research. In particular, we propose
less than 40% for inductive articles. When at least some justification a methodological model for conducting qualitative case studies for
is offered, the tendency is to use the inductive-based logic (e.g., the deductive, theory-testing purposes.
nature of their research questions).
5.1. Typical profiles of inductive and deductive studies
4.2.2.2. Unit of analysis. Overall, 29 (82.9%) articles did not clearly
state their unit of analysis, which represents a very significant The typical inductive paper is focused on describing a phe-
opportunity for improvement. Despite overwhelming majority, the nomenon, using theoretical sampling of multiple cases. There is
issue of whether articles clearly stated their unit of analysis does evidence for some triangulation when conducting data analysis
not seem to have impacted the research outcome. with within and cross-case comparisons. However, it only partially
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 339

Fig. 7. Suggested approach to conducting deductive case studies.

justifies its research idea and may not clearly state its unit of anal- 5.2. Moving forward with case study research
ysis. It ends up offering some insights but falls short of advancing
new propositions or theories. From our examination of 204 inductive and deductive case stud-
The typical deductive paper is focused on revising existing ies published during the period 1992–2007, we have found that the
frameworks/hypotheses or describing a phenomenon, using a sin- use of qualitative case studies has made some contributions to the
gle case with within-case analysis. However, it does not clearly OM field in terms of theory building in new areas and also from inte-
articulate research questions and its unit of analysis. In the absence grating existing theory with new contexts. However, despite these
of clear research protocols, it adopts inductive logic for deductive positive contributions there are some clear lessons that the OM
purposes. For instance, the qualitative data that support hypothe- field needs to heed for theory building and testing purposes, which
ses are compiled inductively and then are used for deductive would increase the rigor and perceived quality of our research, and
means to claim support. It imparts a strong impression that the possibly lessen some of the doubts about the use of this particular
authors selectively chose evidence to justify confirmation of their methodological approach.
hypotheses. Firstly, for inductive, theory building studies, researchers should
Overall, the typical profiles of both approaches lack some follow, and academic journals should seek to encourage researchers
details in how the study is framed and how the analysis is con- to follow, the significant guidelines that have emerged for the
ducted. If so, the basic scientific mode of inquiry that would use of qualitative case studies for theory building. Secondly, for
call for transparency and repeatability could be compromised. theory testing purposes, there needs to be a concerted effort
Nonetheless, moving forward, we believe the OM field is in a within the OM field to develop some standards, or at least some
good position to improve our research practices involving inductive degree of consensus beyond the thoughts of Yin (1989) over
approach as its protocols are well developed. However, the deduc- whether qualitative case can be used for theory testing, and if
tive case studies research protocols are still being developed and so, on what basis and how such case studies should be under-
debated. taken.
340 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

To begin this process, we recognize that the extant literature Appendix A. Evaluation criteria, scales and rationale
points toward two approaches for using qualitative case studies
for theory-testing (Eisenhardt, 1989), based on the overarching Measured items for Rating scales and rationale
form of confirmation of the appropriateness of a theory. Firstly methodological
authors could assess the appropriateness of competing theories (e.g. issues

Eisenhardt, 1989; Pinfield, 1986; Keil, 1995), and secondly, authors Justification for case “Yes”—A statement of why the case method was
could follow an approach based on the collection of longitudinal research adopted appeared in the research together with
a clear explanation of why the case research
data (e.g. Bitektine, 2008). We offer Fig. 7 as a broad framework to
method is appropriate. For example, if “an
take a step toward creating an approach for conducting qualitative exploratory study” was used as the justification
case studies for deductive purposes. of the case method, the research provided a clear
Research questions first need to be stated clearly and the explanation of the gaps in the literature to
unit of analysis identified. Then, the logic of deductive mode validate the “exploratory” claim
Was the reasoning for “Partial”—A statement of why case method was
of inquiry should be presented. Here, it should no longer be using a case research used (for example, an exploratory study)
one that argues exploratory purposes, because the mode is method provided? If appeared in the research but there was no or
theory testing and not theory building. Applicable existing the- so, how well was the only limited explanation of the rationale to
ory or theories should be discussed. If competing theories, reasoning? validate the initial “exploratory” claim
“No”—No attempt of defending the choice of case
competing sets of hypotheses should be developed. If single the-
method appeared in the research
ory or complementary theories, a longitudinal study should be
planned. Unit of analysis “Clearly stated”—The research provided an explicit
statement of the unit of analysis
The competing theories approach is based on the careful
Was the unit of analysis “Not clearly stated”—No explicit statement of the
delineation of multiple dimensions drawn from at least two explicitly stated? unit of analysis was provided in the research
competing theories (Pinfield, 1986). These dimensions are then
evaluated for their appropriateness against data derived from a Theory vs. “Theory”—The research was framed by existing
case study. The evaluation criteria are identified before data col- phenomenon theory such as Transaction Cost Economics,
Resource-based View, etc.
lection, and case study data is collected from multiple sources
Was the research “Phenomenon”—The research was framed by an
to avoid interpretive bias. The theoretically derived dimensions grounded in existing existing stream of literature such as Inventory
are the considered in the analysis through pattern matching (Yin, theory or Management or Project Management
1994) and qualified recommendations are made as to the most phenomenon?
appropriate theory. When discussing the final results, theoretical “Neither”—The research was not framed by existing
theory nor existing literature
generalization, as opposed to statistical generalization, should be
addressed. Sampling strategy “Theoretical”—Cases(s) were selected for
The underlying logic of the longitudinal approach is in essence theoretical purposes, for example, to select polar
similar to that of the competing theories approach in that the extremes where cases exhibited extremely high
data collection and analysis in multiple points in time helps guard or extremely low value on the constructs of
interests
against interpretive bias. However, the longitudinal nature of the
How did the “Convenience”—Case(s) were selected out of
approach gives rise to some fundamental differences (Bitektine, researcher(s) decide convenience of the researchers, for example, the
2008). After formulating research questions and selecting the- on which case(s) to case companies were located within close
ories, an initial case study is identified and the data collection choose? geographical proximity of the researchers
“Random”—Case(s) were randomly chosen
and analysis are conducted. Once the exercise of pattern match-
“No logic offered”—No discussion appeared in the
ing reveals which hypotheses are supported and which are not, research regarding how the case(s) was/were
a set of modified hypotheses may be formulated. Key here is selected
to keep in mind that the researchers are engaged in a deduc-
tive mode of inquiry, and modifying the hypotheses does not Number of cases
mean changing the research questions or constructs. A new set How many cases were We noted the number of cases examined in each
examined in the research article and then grouped them in 5
of evaluation criteria may be developed as well. These steps are research? categories: 1, 2, 3, 4–10 and greater than 10. The
repeated across different points in time until the outcomes have range of 4–10 was drawn from the
answered the research question. In this process, certain aspects recommendation by Eisenhardt (1989)
of the theories may be confirmed while some other may be fal-
sified. Triangulated data “Yes” – More than one source of data was collected
sources and used to validate the findings, for example,
Our study has examined the state of qualitative case studies
company documentation was reviewed in
in OM. The case studies will continue to explore new areas of addition to interviews with key informants
the OM field (i.e. service operations or sustainable supply chains) Was there more than “No”—Only one source of data was collected and
but will also be used to integrate existing topics and theories one source of data used
(i.e. manufacturing strategy) with new theories and perspectives. used to validate the
research findings?
Such approaches will lead to new and significant contributions to “Not stated”—No discussion of the data sources
the OM field. The significantly higher number of inductive case appeared in the research
studies published over the period 1992–2007 is reflective of the
more advanced development of theory-building research proto- Data analysis “Within & Cross-case”—Both within and cross-case
cols (Eisenhardt, 1989; Meredith et al., 1989; Stuart et al., 2002; analyses were provided
How were the research “Within-case”—Only within case analysis was
Voss et al., 2002). At the same time, our study points out how
results presented? provided
researchers in the OM field need to improve on offering sufficient “Cross-case”—Only cross case comparison was
details in research design, data collection, and data analysis when provided
they engage in qualitative case studies. In particular, it calls for a “None”—No within or cross case analysis was
provided
need to develop methodological protocols for deductive case stud-
ies.
 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342 341

Appendix B. Supplementary data Hayes, R.H., Wheelwright, S.C., 1979. Link manufacturing process and product life
cycles. Harvard Business Review 57 (1), 133–142.
Hillebrand, B., Kok, R.A.W., Biemans, W.G., 2001. Theory-testing using case studies:
Supplementary data associated with this article can be found, in a comment on Johnston, Leach, and Liu. Industrial Marketing Management 30
the online version, at doi:10.1016/j.jom.2010.06.002. (8), 651–657.
Hines, P., Silvi, R., Bartolini, M., 2002. Demand chain management: an integrative
approach in automotive retailing. Journal of Operations Management 20 (6),
707–728.
References Hyer, N.L., Brown, K.A., Zimmerman, S., 1999. A socio-technical systems approach to
cell design: case study and analysis. Journal of Operations Management 17 (2),
Anderson, P., 1983. Decision making by objection and the Cuban misslle crisis. 179–203.
Administrative Science Quarterly 28, 201–222. Jensen, R.J., Szulanski, G., 2007. Template use and the effectiveness of knowledge
Barman, S., Hanna, M.D., LaForge, R.L., 2001. Perceived relevance and quality of OM transfer. Management Science 53 (11), 1716–1730.
journals: a decade later. Journal of Operations Management 19 (3), 367–385. Jick, T.D., 1979. Mixing qualitative and quantitative methods: triangulation in action.
Barman, S., Tersine, R.J., Buckley, M.R., 1991. An empirical assessment of the per- Administrative Science Quarterly 24 (4), 602–611.
ceived relevance and quality of POM-related journals by academicians. Journal Johnston, W.J., Leach, M.P., Liu, A.H., 1999. Theory testing using case studies
of Operations Management 10 (2), 194–212. in business-to-business research. Industrial Marketing Management 28 (3),
Benbasat, I., Goldstein, D.K., Mead, M., 1987. The case research strategy in studies of 201–213.
information systems. MIS Quarterly 11 (3), 368–386. Kagan, R.A., 1978. Regulating Justice: Implementing A Wage-price Freeze. Russell
Bitektine, A., 2008. Prospective case study design qualitative method for deductive Sage, New York.
theory testing. Organizational Research Methods 11 (1), 160–180. Keil, M., 1995. Escalation of commitment in information systems development: a
Bititci, U.S., Mendibil, K., Martinez, V., Albores, P., 2005. Measuring and manag- comparison of three theories. Academy of Management Journal 38 (2), 348–352.
ing performance in extended enterprises. International Journal of Operations Kleindorfer, P.R., Singhal, K., Van Wassenhove, L.N., 2005. Sustainable operations
& Production Management 25 (3/4), 333–353. management. Production and Operations Management 14 (4), 482–492.
Bonoma, T.V., 1985. Case research in marketing: opportunities, problems and a Kouvelis, P., Chambers, L.N., Wang, H., 2006. Supply chain management research
process. Journal of Marketing Research 22 (2), 199–208. and production and operations management: review, trends, and opportunities.
Bourgeois, L, Eisenhardt, K., 1988. Strategic decision processes in high velocity envi- Production and Operations Management 15 (3), 449–469.
ronments: four cases in the microcomputer industry. Management Science 34 Krajewski, L., Wei, J.C., Tang, L.L., 2005. Responding to schedule changes in build-to-
(7), 816–835. order supply chains. Journal of Operations Management 23 (5), 452–469.
Boyer, K.K., McDermott, C., 1999. Strategic consensus in operations strategy. Journal Krishnan, V., Loch, C.H., 2005. A retrospective look at production and operations
of Operations Management 17 (3), 289–305. management articles on new product development. Production and Operations
Boyer, K.K., Swink, M., Rosenzweig, E.D., 2005. Operations strategy research in the Management 14 (4), 433–441.
POMS Journal. Production and Operations Management 14 (4), 442–449. Langley, A., 1999. Strategies for theorizing from process data. The Academy of Man-
Bryman, A., 1988. Quantity and Quality in Social Research. Routledge, London. agement Review 24 (4), 691–710.
Buffa, E.S., 1980. Research in operations management. Journal of Operations Man- Lee, A.S., 1989. A scientific methodology for MIS case studies. MIS Quarterly 12 (4),
agement 1 (1), 1–8. 33–50.
Buhman, C., Kekre, S., Singhal, J., 2005. Interdisciplinary and inter-organizational Lee, T.L., Mitchell, T.R., Wise, L., Fireman, S., 1996. An unfolding model of voluntary
research: establishing the science of enterprise networks. Production and Oper- employee turnover. Academy of Management Journal 39 (1), 5–36.
ations Management 14 (4), 493–513. Leonard-Barton, D., 1990. A dual methodology for case studies: synergistic use of a
Burgelman, R., 1983. A process model of internal corporate venturing in a major longitudinal single site with replicated multiple sites. Organizations Science 1
diversified firm. Administrative Science Quarterly 28 (2), 223–244. (1), 248–266.
Campbell, D.T., 1966. Pattern matching as an essential in distal knowing. In: Ham- Lewis, M.A., 2000. Lean production and sustainable competitive advantage. Interna-
mond, K.R. (Ed.), The Psychology of Egon Brunswik. Holt, Rinehart & Winston, tional Journal of Operations & Production Management 20 (8), 959–978.
New York, pp. 81–106. Lewis, M.W., 1998. Iterative triangulation: a theory development process using
Cavaye, A.L.M., 1996. Case study research: a multi-faceted research approach for IS. existing case studies. Journal of Operations Management 16 (4), 455–469.
Information Systems Journal 6 (3), 227–242. McCutcheon, D.M., Meredith, J.R., 1993. Conducting case study research in opera-
Chase, R.B., 1980. A classification and evaluation of research in operations manage- tions management. Journal of Operations Management 11 (3), 239–256.
ment. Journal of Operations Management 1 (1), 9–14. March, J.G., Olsen, J.P., 1976. Ambiguity and Choice in Organizations. Universitets-
Choi, T.Y., Hong, Y., 2002. Unveiling the structure of supply networks: case studies forlaget, Bergen, Norway.
in Honda, Acura and Daimler. Chrysler Journal of Operations Management 20 Markus, M.L., 1989. Case selection in a disconfirmatory case study. In: Cash, J.I.,
(5), 469–493. Lawrence, P.R. (Eds.), The Information Systems Research Challenge: Qualitative
Danese, P., Romano, P., Vinelli, A., 2006. Sequences of improvement in supply net- Research Methods, vol. 1. Harvard Business School Press, Boston, pp. 20–26.
works: case studies from the pharmaceutical industry. International Journal of Meredith, J.R., Raturi, A., Amoako-Gyampah, K., Kaplan, B., 1989. Alternative research
Operations & Production Management 26 (11), 1199–1222. paradigms in operations. Journal of Operations Management 8 (4), 297–326.
Darke, P., Shanks, G., Broadbent, M., 1998. Successfully completing case study Meredith, J.R., 1998. Building operations management theory through case and field
research: combining rigor, relevance and pragmatism. Information Systems research. Journal of Operations Management 16 (4), 439–452.
Journal 8 (4), 273–289. Miles, M., Huberman, A.M., 1984. Qualitative Data Analysis. Sage Publications, Bev-
Dubé, L, Paré, G., 2003. Rigor in information systems positivist case research: current erly Hills, CA.
practices, trends and recommendations. MIS Quarterly 27 (4), 597–635. Miller, J.G., Roth, A.V., 1994. A taxonomy of manufacturing strategies. Management
Dyer, W.G., Wilkins, A.L., 1991. Better stories, not better constructs, to generate Science 40 (3), 285–304.
better theory: a rejoinder to Eisenhardt. Academy of Management Review 16 Mintzberg, H., 1979. An emerging strategy of “direct” research. Administrative Sci-
(3), 613–619. ence Quarterly 24 (4), 580–589.
Eisenhardt, K.M., 1989. Building theories from case study research. Academy of Mosey, S., 2005. Understanding new-to-market product development in SMEs.
Management Review 14 (4), 532–550. International Journal of Operations & Production Management 25 (2), 114–130.
Eisenhardt, K.M., Bourgeois, L., 1988. Politics of strategic decision making in high Narasimhan, R., Jayaram, J., 1998. Reengineering service operations: a longitudinal
velocity environments: towards a midrange theory. Academy of Management case study. Journal of Operations Management 17 (1), 7–22.
Journal 31 (4), 737–770. Nisbett, R., Ross, L., 1980. Human Inference: Strategies and Shortcomings of Social
Eisenhardt, K.M., Graebner, M.E., 2007. Theory building from cases: opportunities Judgment. Prentice-Hall, Englewood Cliffs, NJ.
and challenges. Academy of Management Journal 50 (1), 25–32. Olson, J., 2005. Top-25-business school professors rate journals in operations man-
Fisher, M.L., 2007. Strengthening the empirical base of operations management. agement and related fields. Interfaces 35 (4), 323–338.
Manufacturing and Service Operations Management 9 (4), 368–382. Pagell, M., 2004. Understanding the factors that enable and inhibit the integration
Flynn, B.B., Sakakibara, S., Schroeder, R.G., Bates, K.E., Flynn, E.J., 1990. Empirical of operations, purchasing and logistics. Journal of Operations Management 22
research methods in operations management. Journal of Operations Manage- (5), 459–487.
ment 9 (2), 254–284. Pettigrew, A, 1990. Longitudinal field research on change: theory and practice. Orga-
Gersick, C., 1988. Time and transition in work teams: toward a new model of group nization Science 1 (3), 267–292.
development. Academy of Management Journal 31 (1), 9–41. Pinfield, L., 1986. A field evaluation of perspectives on organizational decision mak-
Glaser, B., Strauss, A., 1967. The Discovery of Grounded Theory: Strategies For Qual- ing. Administrative Science Quarterly 31 (3), 365–388.
itative Research. Wiedenfeld and Nicholson, London. Poole, M.S., Van de Ven, A.H., 1989. Using a paradox to build management and
Grutter, A.W., Field, J.M., Faull, N.H.B., 2002. Work team performance over time: organization theories. Academy of Management Review 14 (4), 562–578.
three case studies of South African Manufacturers. Journal of Operations Man- Popper, K., 1968. The Logic of Scientific Discovery. Harper Torchbooks, New York.
agement 20 (5), 641–657. Ross, J., Staw, B.M., 1993. Organizational escalation and exit: lessons from the
Gupta, S., Verma, R., Victorino, L., 2006. Empirical research published in production Shoreham Nuclear Power Plant. Academy of Management Journal 36 (40), 701–
and operations management (1992–2005): trends and future research direc- 732.
tions. Production and Operations Management 15 (3), 432–448. Roth, A.V., 2007. Applications of empirical science in manufacturing and ser-
Harris, S., Sutton, R., 1986. Functions of parting ceremonies in dying organizations. vice operations. Manufacturing and Service Operations Management 9 (4),
Academy of Management 29 (1), 5–30. 353–367.
342 M. Barratt et al. / Journal of Operations Management 29 (2011) 329–342

Rothlisberger, F.J., 1977. The Elusive Phenomena. Harvard Business School, Division Swamidass, P.M., 1991. Empirical science: the new frontier in operations manage-
of Research, Boston, MA. ment research. Academy of Management Review 16 (4), 793–814.
Rungtusanatham, M.J., Choi, T.Y., Hollingsworth, D.G., Wu, Z., Forza, C., 2003. Survey Trochim, W.M.K., 1989. Outcome pattern matching and program theory. Evaluation
research in operations management: historical analyses. Journal of Operations and Program Planning 12 (4), 355–366.
Management 21 (4), 475–488. Van de Ven, A.H., 1989. Nothing is quite so practical as a good theory. Academy of
Salvador, F., Forza, C., Rungtusanatham, M.J., 2002. Modularity, product variety, Management Review 14 (4), 486–489.
production volume, and component sourcing: theorizing beyond generic pre- Van Maanen, J., 1988. Tales of the Field: On Writing Ethnography. University of
scriptions. Journal of Operations Management 20 (5), 549–575. Chicago Press, Chicago.
Schroeder, R.G., Linderman, K., Zhang, D., 2005. Evolution of quality: first fifty issues Voss, C.A., Winch, G.M., 1996. Including engineering in operations strategy. Produc-
of production and operations management. Production and Operations Man- tion and Operations Management 5 (1), 78–90.
agement 14 (4), 468–481. Voss, C.A., Tsikriktsis, N., Frohlich, M., 2002. Case research in operations manage-
Scudder, G.D., Hill, C.A., 1998. A review and classification of empirical research in ment. International Journal of Operations & Production Management 22 (2),
operations management. Journal of Operations Management 16 (1), 91–101. 195–219.
Skinner, W., 1980. Manufacturing and technological strategy. The Journal of Business Wood, A.R., Britney, R.R., 1989. Production operations management: research and
Strategy 1 (2), 69–72. teaching opportunities in the 1990’s. Operations Management Review 8 (3/4),
Sousa, R., 2003. Linking quality to manufacturing strategy: an empirical investi- 33–43.
gation of customer focus practices. Journal of Operations Management 21 (1), Wu, Z., Choi, T.Y., 2005. Supplier-supplier relationships in the buyer–supplier triad:
1–18. building theories from eight case studies. Journal of Operations Management 24
Sousa, R., Voss, C.A., 2001. Quality management: universal or context dependent? (1), 27–52.
Production and Operations Management 10 (4), 383–404. Yan, A., Gray, B., 1994. Bargaining power, management control, and performance in
Stuart, I., McCutcheon, D., Handfield, R., McLachlin, R., Samson, D., 2002. Effec- Uni. Academy of Management Journal 37 (6), 1478–1517.
tive case research in operations management: a process perspective. Journal Yin, R.K., 1989. Case Study Research: Design and Methods. Sage Publications, New-
of Operations Management 20 (5), 419–433. bury Park, CA.
Sutton, R., Callahan, A., 1987. The stigma of bankruptcy: spoiled organizational image Yin, R.K., 1994. Case Study Research: Design and Methods, 2nd ed. Sage Publications,
and its management. Academy of Management Journal 30 (3), 405–436. Newbury Park, CA.