Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
The Grounded Theory Method: Field Work, Analysis, and Reaching Results
Markus M. Mäkelä Helsinki University of Technology
ABSTRACT
In this paper, we continue the discussion on building grounded theory from case studies in the domain of entrepreneurship, an effort we started in a companion paper (Mäkelä and Turcan, 2004). We discuss procedures of field work and data analysis, linking observations from the data and the emerging theoretical framework with prior literature, and formulating propositions as embodiments of the findings. We introduce an analysis of a set of recent top-tier entrepreneurship articles that illustrate the uses of the method under varying circumstances.
Keywords: grounded theory method, entrepreneurship research, field work, analysis, results
2
INTRODUCTION
In this paper, we build on a companion paper (Mäkelä and Turcan, 2004),
referred to throughout this paper. We carry on our discussion on building grounded
theory from case studies in the domain of entrepreneurship. We start by discussing
work at the field. We also review data analysis, linking observations from the data and
the emerging theoretical framework with prior literature, and thereafter the
formulation of propositions that transparently and vividly articulate the findings.
Throughout the paper, we continue our custom from the companion paper,
discussing the conduct of the research phases in recent entrepreneurship articles that
have employed the grounded theory or a similar method. In an Appendix of this
paper, we detail how we selected a set of eight such papers for illustrative purposes
and how we analyzed these papers. The Appendix introduces a table by which we
seek to illustrate the many faces of the method in publications of top-tier international
outlets. By presenting the table, we also hope to make it easy to compare authors’
choices and decisions across papers and to gain a holistic view of these exemplar
papers.
DATA COLLECTION
In this section, we review general requirements for a field investigator, sources
of evidence and methods for data collection, conducting a pilot study, and making use
of overlap between data collection and analysis.
3
Requirements for the Field Investigator
There are a number of requirements for the field researcher. Prior to entering
the field, a researcher must be cognizant of the various types of data and how they can
be employed to achieve the benefits of triangulation. While in the field, she must also
be able to independently act according to the situation, making decisions about the
data being collected so as to maximize the quality of the study (Yin, 1994). She needs
to recognize places where thorough questions are warranted to enable the study to dig
deeper into issues that are of interest. That is, she needs to be adaptive to the situation
and flexible to the requirements – such as schedule –of informants, and a good
listener (Yin, 1994).
Importantly, the researcher has to bear the initial setup of the study – prior
literature, the research problem, and the potential a priori constructs – clear in mind.
She typically also needs to know a lot about the studied case or other environment in
advance, if material has been available. In many empirical settings, there is public
material about the object of study and its history, such as news items and www sites
of organizations. It is often beneficial to be able to show in an interview situation that
one masters public knowledge so that interviewees do not feel that their time would
be inefficiently used. While the investigator should have a sound grasp of information
about the case and its surroundings, she should be free of bias, not seeking to
substantiate a preconceived view of study results (Eisenhardt, 1989; Yin, 1994;
Strauss and Corbin, 1998).
Sources of Evidence and Data Collection Methods
Yin (1994) presented a thorough description of potential sources of evidence
in case studies (see also Brundin, 2004). Key sources, a part of which can also be
4
conceived as data collection methods, are interviews, documents, observations, and
physical artifacts1. Depending on the definition of the above elements, however, not
all sources of a more exotic nature may be covered by the list, such as videotapes,
psychological testing, or life histories (Yin, 1994; Smith, 1994). All sources of data
collection used in a study should be separately mastered by the researcher.
Yin (1994, p. 80) presented a vivid tabular illustration of the strengths and
weaknesses of the sources of evidence, pointing out that no one of them is superior
with a complete advantage over the others. Strengths of interviews include that they
can potentially be effectively targeted to suitable informants, and allow the researcher
dwell into theoretically interesting topics that provide added insight. Weaknesses
include that poorly constructed questions, the purposeful conduct of informants, or
poor recall of prior events can bring about response bias. For instance, interviewees
may be reluctant to reveal worse aspects of their or their organization’s history and
may even dress up what happened – a behavior that scholars sometimes term ‘post
hoc rationalization.’ The informant may also be biased to provide such answers that
he believes the interviewer wants to obtain.
Benefits of documentation include that it can be repeatedly reviewed, is
unobtrusive in that it has not been created in the course of the study, and provides
exact and potentially broad coverage of the states of nature. On the downside, access
to such data, retrievability may be low, and different types of biases may be present.
1 Yin presented a more detailed division describing (1) documents and (2) archival records where we aggregate these under the label of ‘documents’. By archival records, Yin referred to documents that are held in archive-like storages, and stated that the usefulness of such archival records varies much more from study to study than that of documentary evidence.
Moreover, Yin distinguished between (1) direct observations and (2) ‘participant-observation.’ Direct observation refers to observation-making in general, while participant-observation was identified as a special mode in which the researcher may assume a variety of roles within the study situation and may participate herself in the events being studied. Participation, implying the key ingredient of ‘participatory action research’ (Gummesson, 1991; Kemmis and McTaggart, 2000), is (already by definition) the key attribute separating participant-observation from other observational data collection.
5
Upsides in using observations include that they are highly contextual,
covering events in real time. If observations are made in a ‘participant-observation’
mode, where the researcher actively participates events that relate to the study, she
may gain thorough insights into interpersonal behavior and motives of action.
Weaknesses of observation include that the may be time-consuming if conducted in a
large scheme, and negative selectivity may be involved. Furthermore, analogically to
interviews, things may proceed as they do because they are being monitored,
exhibiting reflexivity. Reflexivity can also be brought about by the researcher’s own
manipulation of events when the observation is participatory. Finally, physical
artifacts include objects such as a device or a computer printout in a study of
microcomputer use (see Yin, 1994, p. 90). According to Yin (1994), they have less of
potential relevance in a typical case study. Artifacts can provide insights into, for
instance, cultural and technological features, questions, and operations. Downsides
include availability and selectivity.
Triangulation, as discussed in the companion paper, is an important issue and
tool in qualitative research (Huberman and Miles, 1994; Altheide and Johnson, 1994),
many aspects of which closely pertain to the fieldwork phase and data collection as an
important area of manifestation of its benefits (Eisenhardt, 1989; Yin, 1994).
Triangulation refers to the combination of multiple methods, empirical materials,
observers, or perspectives in a single study. The use of triangulation is an attempt to
obtain a deep understanding of the studied phenomenon that may also add rigor and
breath (Jick, 1979; Denzin and Lincoln, 1994). Triangulation does not represent a
form of validation but is an alternative to validation (Flick, 1992; Denzin and Lincoln,
1994). In all grounded theory research, triangulation provides important means to
reduce the likelihood of misinterpretation, and thus its use is generally recommended.
6
In top grounded theory research, interviews and documents are typically used
as sources of evidence. It is, in fact, rather rare not to use interviews, but such studies
naturally are possible (see, for instance, Amit and Zott, 2001). We also think that
documentary and observational evidence are in reality used in many papers that do
not explicitly acknowledge their use but just state that they use interviews.
Pilot Study
A pilot study may be a very helpful means of guiding data collection – as well
as analysis – to the appropriate direction. Such investigation can pivotally help avoid
costly pitfalls by letting researchers reformat their research protocol so that empirical
findings can have their share in the important design of grounded theory research.
Both the content of data and procedures of investigation may be revised.
A limited number of cases should be selected for the pilot study, and just one
case can provide important insight into the phenomena at hand. The selection of the
pilot case may follow different criteria than the selection of the actual research cases,
because one may, for instance, find a case where informants or documents are
unusually accessible, or the geographical location is more convenient (e.g. Yin, 1994).
Yin (1994, p. 74) succinctly emphasized that a pilot study is not a ‘pretest’ in
that it should not attempt to actually start the process of generating theory, but just
help to develop suitable lines of questions, potentially clarifying the essence of the
study to the researchers as well. The pilot study can even be a much broader than the
ultimate data collection scheme and can entail much less focus. It can help design
both methodological and substantive issues (Yin, 1994).
One will note that while a pilot study is not a ‘prestudy,’ its benefits still can
occur also in the process of adjusting data collection and analysis in course or
7
research, a key ingredient of grounded theory inquiry that is discussed in the
following subsection. However, due to availability of resources, a separate pilot study
does naturally stand the best chances of obtaining significant benefits.
Entrepreneurship studies that we analyzed poorly stand up to their label of
‘exemplars’ in that the use of pilot study is rare in them. According to our subjective
observations from the review of more than 200 entrepreneurship articles that
employed grounded theory or a similar method, the analyzed papers would appear to
relatively well represent the population of entrepreneurship grounded theory articles
in top scholarly outlets. Only two of the exemplar papers have elements of a pilot
study visible: Uzzi (1997) used two pilot interviews and Amit and Zott (2001)
reported that they changed their questionnaire some time after the beginning of data
collection.
Overlap of the Collection and Analysis of Data
Generally, an important feature of many qualitative research efforts is that the
data collection and analysis phases should overlap to enable refocusing the study and
conducting appropriate theoretical sampling of cases and within cases in the course of
research. Interview templates and questionnaires can also be amended (Glaser and
Strauss, 1967; Eisenhardt, 1989; Strauss and Corbin, 1998). Such freedom to adjust
the working protocol is considered an important advantage of case research
(Eisenhardt, 1989), representing ‘controlled opportunism in which researchers take
advantage of the uniqueness of a specific case and the emergence of new themes to
improve resultant theory’ (p. 539). Such opportunism is legitimate because the
researcher is attempting to understand each case separately in extensive depth
8
(Eisenhardt, 1989). An addition benefit of overlap is that it also gives the investigator
a head start in the laborious task of analyzing data.
Field notes are a central tool for the field researcher. Eisenhardt (1989)
discussed field notes as an important means of achieving overlap between data
collection and analysis. At best, they can be an account of both what happened and
how the researcher herself reflected on that on the spot, representing not only
observation but also analysis, with these two aspects preferably separated. They can
thus provide a ‘stream-of-consciousness’ commentary about events of the study (Van
Maanen, 1988; Eisenhardt, 1989). Ideas recorded in field notes competitive advantage
be, for example, hunches about causal relationships, informal observations, anecdotes,
or cross-case comparisons.
We now proceed to review the phase of research that overlaps data collection.
DATA ANALYSIS
A variety of types of qualitative data analysis has been suggested in the
literature. For instance, Yin (1994) advocated ‘pattern-matching,’ ‘explanation-
building,’ ‘time-series analysis,’ and ‘program-logic models’ with a few ‘lesser
modes’ of analysis. Miles and Huberman (1994) presented a large set of display
vehicles for data analysis, such as arrays, flowcharts, timelines, matrices, and
frequency tabulations.
Strauss and Corbin (1990; 1998) advocated a rather prescriptive set of strict
coding methods that are likely to offer good support for new researchers but at the
same time were argued to lose the sense of emergence and open-ended character and
represent too complex a ‘maze of techniques’ (Charmaz, 2000). The presentation of
their methods of ‘open, axial, and selective coding’ in their 1998 work was perceived
9
as an improved and more accessible version by Charmaz (2000). Glaser (1992)
viewed their 1990 work as ‘forcing’ researchers through too strict and binding a set of
procedures for allowing emergence. Glaser maintained that analysis should be just
about iterative and constant comparisons.
In their analysis, different authors typically use a limited number of methods
that are selected from a larger pool, even though many methods have different names
for different authors. For instance, components of Yin’s methods that were listed
above are present in the work of Eisenhardt (1989), Miles and Huberman (1994), and
Strauss and Corbin (1998) with partly different labels.
Qualitative analysis in grounded theory attempts to conceptualize, reduce,
elaborate, and relate data and categories to integrate them as novel theory. While
various qualitative analysis methods are introduced in detail in the various works
discussed above, we concentrate on one of the often-used types of data analysis. Our
selection is the account of Eisenhardt (1989) that discusses analysis as divided into
the within-case and cross-case phases. In this discussion, we wish to avoid the ‘maze
of techniques’ (Charmaz, 2000) of the more prescriptive approach of Strauss and
Corbin (1990; 1998).
Regarding the ‘Methods’ sections of papers, Eisenhardt (1989) noted the same
pattern in papers of the 1980s that are apparent in those from 1996-2003 that were
reviewed by us for the analysis of the Appendix: papers generally give very little
space for describing analysis. According to the papers’ short descriptions, papers have
much variation in how much their authors have done for the analysis, and there seems
to be a degree of confusion that attests to the view that the process quality in the
production of grounded theory has much variation even in these high-level
publication outlets.
10
Still, it is important for reliability of the study and for readers of qualitative
research to know how the analysis was conducted. Good exemplars can be found in
Eisenhardt’s own studies that provide the reader with deep information about the
research setting, data collection, and analysis (e.g. Brown and Eisenhardt, 1997), with
often a few pages allocated for these issues.
In the following, we first review within-case analysis and then its companion
for multiple-case research, cross-case analysis. Our review is based on Eisenhardt’s
(1989) presentation.
Conduct of the Analysis
Within-case analysis is an important help for qualitative researchers as they
are trying to cope with a staggering volume of data. There is no standard format for a
within-case analysis, but a typical procedure in deeply studying one case includes a
write-up of the case or a thorough examination of the events on the timeline. The
write-ups and timelines may be purely descriptive. Even description can be important
for generating insight. All sources of evidence that are in use in the study should be
well exploited. The key objective is to reach a state of intimate familiarity with the
states of events in each case separately. Such familiarity and knowledge of the unique
aspects of each case can facilitate a successful cross-case analysis.
As discussed in the companion paper, a multiple case study often is the
appropriate structure for a grounded theory setting, because the only appropriate
single case design for grounded theory research is that of an extreme or unique case
(see Yin, 1994). For multiple case studies, cross-case investigation is a critical phase
in analysis. Its ideas can, however, be applied in some analysis work in the level of
comparing embedded units in the sublevel of a single case.
11
Eisenhardt (1989) vividly points out that a rigorous cross-case analysis is
important because people are poor processors of information; they are overly affected
by some findings from data, by status of some informants, and may inadvertently
ignore disconfirming evidence. In cross-case analysis, the data should be scrutinized
from several viewpoints. This will counteract the weaknesses of people in processing
information. Eisenhardt’s articulation of typical qualitative cross-case analysis
vehicles materialized in a threefold classification: clustering of categories and
dimensions, identification of similarities and differences between pairs of cases, and
dividing data according to the source of evidence.
In the first method, conceptual categories or dimensions are selected and then
grouped for identifying intragroup similarities and intergroup differences. The
selection of the categories or dimensions should be based on the research problem or a
review of the data that could be called ‘open coding,’ that is, the identification of
concepts and their properties from the data. If categories are selected from existing
literature, the researcher will have to be especially vary about not letting prior
knowledge guide the analysis to counterwork the spirit of grounded theory.
Continuous scales can be used in plotting the positions of cases or their subunits. In
many cases, a useful extension of clustering dimensions is to use a multidimensional
display (for instance, a two-dimensional matrix for representation of two dimensions
or multiple two-dimensional matrices for representation of three dimensions).
The second vehicle of comparing pairs of cases enables investigators to find
insights beyond simplistic constructs and relationships. The search of similarities
from seemingly different cases or differences from seemingly similar ones can bring
about sophisticated insights, and new dimensions and concepts may emerge. More
than two cases can naturally be grouped in the use of this tool.
12
The third tool is to cluster data according to the source of evidence.
Sometimes, different sources such as interviews, observations, and documents
facilitate different insights, and a type of triangulation of data source can be made use
of here to produce deep insights and help increase the quality of the study. If results of
the analysis conflict, they can at times be reconciled so that a form of ‘literal
replication’ or ‘theoretical replication’ (see the companion paper) is achieved. ‘Literal
replication’ can be achieved if the results show similar results and ‘theoretical
replication’ if there is a credible reason that accounts for the difference. When such
replication cannot be established, the analysis has helped identify a spurious
relationship.
The general objective in cross-case analysis is to analytically and procedurally
seek insights from beyond initial expressions and potential delusions that are often
inherent in human thought. The use of diverse tools in the task can bring about
increased quality and corroboration of the resultant theory’s fit with data.
A very important part of analysis is to know when the iteration can be stopped.
This meticulous interplay between the three components (Strauss and Corbin, 1994)
demands much resources and researchers should not overdo it, because other things
being equal, this will work to reduce their attention to the quality of analysis. At some
point in time, there is saturation in the emerging propositions so that further iteration
would be unlikely to provide significant incremental learning of new aspects of the
framework (Eisenhardt, 1989). Similarly, in the earlier-phase issue of case sampling,
adding cases should be stopped when there is saturation. One should be cognizant,
however, of the fact that the identification of saturation can be difficult in adding
cases and, indeed, even in knowing when there is saturation to the results.
13
IDENTIFYING CONNECTIONS WITH LITERATURE AND
FORMULATING PROPOSITIONS
Analysis and even data collection are bound to the phases of identifying
connections to prior literature and formulating propositions; the two last phases also
significantly overlap each other. Concepts and relationships between them have begun
to emerge while analysis has progressed, and the researcher can move on to
sharpening constructs and the model. We base our introduction on identifying
literature links and formulating propositions mainly on Eisenhardt’s (1989) work.
In regard to the formulation of propositions, we review Eisenhardt’s
suggestions of modifying and sharpening constructs and verifying fit. ‘Sharpening of
constructs’ refers to making needed revisions to the definition of a construct and
gathering and reviewing evidence for measuring that construct. In other words, it can
be viewed as establishing construct validity. Making constant comparisons in an
iterative fashion is once again the key. The accumulation of evidence from multiple
sources provides support for the construct. Eisenhardt specifically points out the
similarity of developing a construct from several types of evidence to the
development of a single measure from multiple indicators in theory-testing research.
While no method for collapsing multiple indicators, such as factor analysis, is
available, the researcher will have to rely on the stringent iteration inherent to
grounded theory building. Tabular displays (e.g. Miles and Huberman, 1994) are of
key use in this collapsing, reducing, or summarizing of data.
Tables can support the presentation of constructs and findings generally in
important ways, because they offer the author a way to condense the major
components of their results in an easily-accessible form. Perhaps more importantly,
tables can offer a way to condense the support of the data for the emerging framework
14
of propositions, showing the reader how causal dynamics operate in the data when
values of key factors are varied. One example of a useful practice is to describe
detailed findings from the cases in separate tables for each construct of the emerging
model, and then condense the case data to a summarizing attribute and present these
summary attributes for all constructs in a summary table. Such variance-
representation table will draw together the support from the data for the model and
allow the reader to quickly grasp it.
The discussion on variance-representation tables is important for multiple-case
studies that make use of the replication logic, but may also be exploited in the study
of dynamics within a single case. Tabular displays are also oftentimes useful in
describing the interview set or other data: how many informants were there in the
cases, what were their positions and backgrounds, which documentation was obtained
of the cases, when were steps of data collection performed, et cetera. Likewise, tables
can effectively be used in describing the history or present condition of the cases.
Good examples of the use of tables can be found in the work of, for instance,
Eisenhardt (Eisenhardt and Bourgeois, III, 1988; Brown and Eisenhardt, 1997), and in
qualitative research on entrepreneurship and related topics (Birkinshaw, 1997; Uzzi,
1997; Amit and Zott, 2001). Good tables are an important support for constructs and
arguments for causal relationships. Tables can importantly help readers apply their
own standards for judging the study.
Shaping propositions continues with the verification of their fit with the data
from the cases. At this stage, the researcher should keep iterating between the data
and emerging framework, adjusting the new framework so that its fit with the data
grows, increasing the empirical validity of the model. Some relationships will prove
not to have backing from data, and they will have to be discarded. In fact, as
15
Eisenhardt pointed out, the difference of this verificational phase to theory-testing
research is just that the checking is done for each case separately – a manifestation of
the replication logic – and, as opposed to quantitative testing, does not involve
statistical procedures.
In the replication logic, cases that confirm proposed relationships enhance
confidence in the validity of findings. Conflicting cases can provide an opportunity to
refine and extend the emerging theory. They can for instance, lead to the
identification of new dimensions with which a model has to be amended to improve
explanation. One should remember, however, that cases need not display similar
results, especially if there is a theoretical reason to believe that they should not
because of the influence of a factor external to the proposition. Such a theoretical
reason will account for ‘theoretical replication’ (Yin, 1994, p. 46). The researcher will
have to rely on analysis and the stringent rotation between data, the emergent
framework, and existing literature, and thereafter conclude whether the analysis
converges to certain propositions. Replication can be claimed whenever two or more
cases support a proposition, and established replication will enhance the belief in the
validity of the results. Replication can be claimed also when there is such conflict that
can be explained by the variation of an external factor (see, e.g. Yin, 1994).
Analogously to conflicting case results, finding of conflict with earlier
literature may reveal places for improvement in existing knowledge but may also
create a need to do more iteration to distinguish the contribution of the present work
and to see if its results are valid. One must keep in mind that the result of a grounded
theory inquiry may be disappointing, leading to no new insights of significant value.
All insights from the emerging framework may turn out only to replicate prior theory
or turn out not to converge (e.g. Eisenhardt, 1989).
16
A critical aspect to internal validity is understanding the underlying theoretical
reasons for why something happens, and qualitative data has the advantage of
providing rich opportunities for this. This understanding will facilitate understanding
if an identified relationship could be just a spurious one or perhaps caused by an
external variable that has not been recognized before. Overall, the strengths and
consistency of the propositions have to be examined both within and across cases.
In grounded theory research, it can at times be very challenging to distinguish
one’s results from results already achieved by others. When the first initial results
have emerged from data analysis, the researcher can begin comparing these with prior
literature.
Comparison with existing literature is a key part of the rotation between data,
emergent framework, and theory that is inherent to the grounded theory process. A
large enough body of literature needs to be considered and similarities and
contradiction sought so that new contributions can be distilled. Similarly with
contrasting cases, contrasting literature can even be helpful in producing
framebreaking new insights to factors such as new dimensions or constructs. In the
best scenario, new views can be gained to the emerging framework, to the conflicting
literature, and to the domain of external validity. However, as in contrasting cases,
counteracting literature may also just reduce confidence in the validity of the findings.
Contrasting literature may point out to incorrect results, that is, limited internal
validity, or to case-idiosyncracy of the results, or in other words, limited external
validity.
Literature of analogous phenomena that nevertheless does not pertain to the
very same phenomena can be valuable, because it can show similarities that are
typically not analyzed in connection to each other, and new insights can, once again,
17
result. Improved levels of conceptual representation and external and internal validity
can follow from such comparison. Typically, these three benefits are the key
advantages that can accrue from rigorous comparison of the results with earlier
literature.
Finally, we discuss a more practical issue that has some relevance: the
selection between explicit and embedded presentation of propositions in the research
report. This refers to the selection on whether the propositions are stated as separate
full sentences that often are labeled with a preceding ‘Proposition 1:’ or equivalent
label, or if the reader will have to pick them up by reading relevant parts of the report
text.
Advantages of presenting propositions explicitly as separate sentences include
that they will be easily accessible for glancing readers and also more quickly
recognizable for all. Explicit statements will provide the presentation with a more
strict structure, and help to concretize the contributions of the research and make them
stand out. However, in some cases, a touch of generality and a recognition of potential
application areas might perhaps be lost if the thinking of the reader is guided by
ready-formatted statements. This might make it optimal to present propositions
embedded in the text.
In our review of a large number of articles, we found that authors typically
ground their results relatively well to existing literature, addressing the findings’ links
to prior knowledge at the latest in the results section. The ‘contribution to theory’ of
contemporary papers, however, is often rather topical in nature, that is, studies still
mostly present ‘theories about specific phenomena’ as typical outcomes were
described by Eisenhardt (1989, p. 547). In the analyzed papers, both explicit and
embedded presentation of propositions are frequently employed choices. Roughly,
18
about a half of those papers that we reviewed for selection to our analysis explicitly
presented propositions in a sentence format.
DISCUSSION AND CONCLUSIONS
In this paper and in its companion, we have distinguished the presentation
from many of prior grounded theory review by two focal means. First, in the
beginning of the companion paper, we have presented an explicit discussion on the
paradigmatic position of grounded theory – that is, we have identified how the
method’s epistemology and ontology can be and have been viewed, and how these
views affect the use of the methodology. We have organized our discussion of
grounded theory’s position by acknowledging each important dimension that is
relevant for the total positioning of a method. We have noted the possible paradigms
from which the method can be viewed and have taken an explicit stand on that we
view it from the perspective of the postpositivistic paradigm.
One notably important aspect and subissue of our paradigmatic discussion
pertains to the interrelation of ‘case study research’ and ‘grounded theory research.’
We have explicitly defined these terms, not leaving their meaning and mutual
relationship blurred or vague. Several definitions can be proposed for case study
research, and in our opinion, it is advantageous – and, in fact, the only appropriate
way to represent the actual meaning of the term ‘case study’ – that the case study be
viewed as a selection of the object of study, not method. ‘Case research’ is, indeed,
often used wrong, for instance as an euphemism for qualitative research (Locke,
2001). We wish to reiterate to emphasize this important definitional point. We have
taken the definition of case study as a choice of object of investigation in the papers.
In this position, the research method or strategy should be viewed as another, separate
19
dimension. The twin papers review the intersection of using the grounded theory
method and investigating cases as objects of study. In other words, several different
methods can be used in the study of cases. This fact together with the meaning of the
term ‘case study’ implies to us that it is misleading to talk of case study as a method.
While it would be possible to adopt such practice for the method’s label, the label
would be a misnomer.
Second, we particularly discuss grounded theory as a method for research in
the entrepreneurship domain. We have conducted a review of more than 200
entrepreneurship articles from the years 1996 through 2003 that use the grounded
theory or a similar method, and we have selected eight articles for a detailed analysis.
This analysis is summarized in the Appendix of this paper and verbally reflected
throughout the texts of this and the companion paper. We have used the analysis to
illustrate potent ways of employing the method. We have also pointed out places that
call for improvement. Furthermore, we have illustrated the flexibility of the method
by pointing out how the it can be fitted into different environments and which
variations in its use are possible, appropriate, or advantageous.
Feasibility, Process, and Outcomes of Grounded Theory Research
In the presentation of the twin papers, strongly combined and synergistic
between the papers in design, we have first discussed the position of grounded theory
research, its history, potential outcomes from it, and its feasibility in different
environments. We then have reviewed phases of the grounded theory process,
introducing key design issues, requirements for researchers and a pilot study, data
collection and sources of evidence, and conducting the analysis. Finally, we have
20
discussed the formation of propositions and identifying connections with prior
literature.
A grounded theory approach is often feasible and appropriate when the
research question pertains to a field where not much prior theory exist, and in which
theories of other fields may lack validity or conflict. Qualitative research can also
provide a fresh perspective on a topic that has already received scholarly attention
(Hitt et al., 1998).
Case study in general has been suggested to be appropriate when the study
focuses on contemporary events over which the researcher has little control (Yin,
1994). Furthermore, qualitative research often stands much better chances that other
research to grasp the complexity of social phenomena (Dougherty, 2002). Qualitative
research can also drill deep into phenomena where obtaining reliable quantitative data
might be troublesome.
Grounded theory researchers should begin by formulating a research problem
and making sure they have relevant background information of prior literature and
theory that might be applicable or provide directions of potential a priori constructs or
formulating the research procedure, including sampling. Sampling should be
theoretical in nature, seeking to obtain variation in key variables. Decisions between
single and multiple case structures as well as holistic and embedded design has to be
made. The replication logic in multiple-case designs is an important means for
supporting external validity. Researchers can importantly benefit from the richness of
qualitative data, and triangulation in terms of, for instance, data types, researchers,
and sometimes even the use of complementary methods outside the domain of
grounded theory.
21
The overlap of data collection and analysis is an important feature of
qualitative research that facilitates improving quality of research. Within- and cross-
case analysis phases and various more detailed tools provide a structure for analysis.
Iterative rotation between data, literature, and emerging propositions is a key feature
of grounded theory research. It is important to know when the results have reached
saturation and the iteration can be stopped.
Grounded theory research can have many outcomes (Eisenhardt, 1989; the
present paper’s Appendix). It can lead to, among others, ‘causal theory,’ wherein
relationships of mutually interacting constructs are explained, or ‘process theory,’
wherein the explanation specifically focuses on sequences of temporally evolving
action, changes in which can be traced to structural and environmental changes
(Strauss and Corbin, 1998, p. 163). Potential outcomes also include mere building
blocks of theory, for instance, concepts, typologies, and suggestions for facilitating
statistical research, such as ideas for potentially valid measurement items.
The Future of Grounded Theory Research
Dougherty (2002) succinctly raised some avenues where researchers might
work to enhance the quality of grounded theory in the future, and the first one of them
concerns research problems. She argued for better framing of research questions and
for the related task of better arguing the purpose of investigation. She pointed out that
this framing is challenging because it is difficult to know enough of earlier theory but
still not let it guide the process. Genuine fit of the method with the research question
is of essence.
Both Dougherty (2002) and our results point out that data representation and
its display should be improved. Readers would be able to much better recognize the
22
support of the actual data for the presented propositions and theoretical implications if
appropriate tables would be used. There are a huge number of types of displays (Miles
and Huberman, 1994) and different types fit different presentations. In many cases of
grounded theory research, however, a table representing the variances of an
underlying factor and a factor affected by it would provide excellent illustration and
support for the results’ validity. If possible, tables can be used to provide detailed
description of data underlying each factor, and then a summary table could tie
together the effects for all factors of the proposed model.
Dougherty (2002) also suggested that report structures be given more thought.
A common way to report literature in reports is to introduce it prior to the method,
data, and results – that is, grounded theory reports follow the same format in this than
deductive papers typically do. This would appear to be a fine choice, even though it
remains possible to first introduce the data and then prior literature, a choice that
might better reflect the origin of grounded theory that is in data and not in hypotheses
from prior literature.
A further improvement to grounded theory efforts would be enhancing their
contributions to existing base of theory. We specifically mean that in addition to
creating new, context-specific theory, the level of abstraction should be raised and
contributions to more established theories should be identified more often in research.
Currently, as illustrated by our analysis, many contributions of articles of the first tier
journals are ‘essentially theories about specific phenomena’ that lack the sweep of
grand theory; the situation is alike in the 1980s, when Eisenhardt could use these
words (1989, p. 547). As we take the positivistic or postpositivistic perspective to
viewing research, we advocate seeking an elevated level of abstraction so that
enhanced generality can be reached.
23
We hope that our papers help entrepreneurship scholars to clearly state their
method and details of analysis, including the paradigmatic position, and well
articulate the contents of the propositions and issues of quality of study.
For a proportion of research problems in the domain of entrepreneurship,
grounded theory potentially presents a very fruitful methodological selection. The
accumulation of knowledge requires constant iteration between theory-creation and
testing, much like grounded theory requires iteration between data, the emerging
framework, and literature. Good grounded theory is likely to have the benefits of
novelty, validity, and testability, and providing an early conception of new research
domains. As pointed out by Eisenhardt (1989), this type of research importantly
complements incremental theory building via empirical testing and extension.
Bergmann Lichtenstein and Brush (2001) noted the increasing use of theory-building
with multiple case studies in entrepreneurship. We wish to support the further use of
the method and hope to have contributed to the spread of knowledge on its
employment.
24
REFERENCES
Altheide, D. & Johnson, J. M. 1994. Criteria for assessing interpretive validity in
qualitative research. In Denzin, N. K. & Lincoln, Y. S. (Eds.), Handbook of
Qualitative Research: Thousand Oaks, CA.: Sage.
Amit, R. & Zott, C. 2001. Value creation in e-business. Strategic Management
Journal, 22(6/7): 493-520.
Barringer, B. R. & Greening, D. W. 1998. Small business growth through geographic
expansion: A comparative case study. Journal of Business Venturing, 13(6): 467-492.
Bergmann Lichtenstein, B. M. & Brush, C. G. 2001. How do "resource bundles"
develop and change in new ventures? A dynamic model and longitudinal exploration.
Entrepreneurship Theory and Practice, 25(3): 37-58.
Birkinshaw, J. 1997. Entrepreneurship in multinational corporations: The
characteristics of subsidiary initiatives. Strategic Management Journal, 18(3): 207-
229.
Brown, S. L. & Eisenhardt, K. M. 1997. The art of continuous change: Linking
complexity theory and time-paced evolution in relentlessly shifting organizations.
Administrative Science Quarterly, 42: 1-34.
Brundin, E. 2004. Strategies for collecting empirical material for real-time studies. In
Neergaard, H. & Ulhøi, J. P. (Eds.), Handbook of Qualitative Research Methods in
Entrepreneurship, forthcoming in 2005: Edward Elgar.
25
Charmaz, K. 2000. Grounded theory: Objectivist and constructivist methods. In
Denzin, N. K. & Lincoln, Y. S. (Eds.), Handbook of Qualitative Research: 163-188.
Thousand Oaks, CA.: Sage.
Denzin, N. K. 1978. The Research Act: A Theoretical Introduction to Sociological
Methods. New York: McGraw-Hill.
Denzin, N. K. & Lincoln, Y. S. 1994. Introduction: Entering the field of qualitative
research. In Denzin, N. K. & Lincoln, Y. S. (Eds.), Handbook of Qualitative
Research: Thousand Oaks, CA.: Sage.
Dillman, D. A. 1978. Mail and Telephone Surveys. New York: Wiley.
Dougherty, D. 2002. Grounded theory research methods. In Baum, J. A. C. (Ed.), The
Blackwell Companion to Organizations: Padstow, UK.: Blackwell.
Eisenhardt, K. M. & Bourgeois, L. J., III 1988. Politics of strategic decision-making
in high-velocity environments. Academy of Management Journal, 31(4): 737-770.
Eisenhardt, K. M. 1989. Building theories from case study research. Academy of
Management Review, 14(4): 532-550.
Flick, U. 1992. Triangulation revisited: Strategy of validation or alternative? Journal
for the Theory of Social Behaviour, 22: 175-198.
Garud, R., Jain, S., & Kumaraswamy, A. 2002. Institutional entrepreneurship in the
sponsorship of common technological standards: The case of Sun Microsystems and
Java. Academy of Management Journal, 45(1): 196-214.
26
Glaser, B. G. 1992. Basics of Grounded Theory Analysis: Emergence vs. Forcing.
Mill Valley, CA.: Sociology Press.
Glaser, B. G. & Strauss, A. L. 1967. The Discovery of Grounded Theory: Strategies
for Qualitative Research. Chicago, IL.: Aldine Publishing Company.
Gummesson, E. 1991. Qualitative Methods in Management Research. Newbury Park,
CA.: Sage.
Hitt, M., Harrison, J., Ireland, R. D., & Best, A. 1998. Attributes of successful and
unsuccessful acquisitions of US firms. British Journal of Management, 9(2): 91-114.
Huberman, M. & Miles, M. B. 1994. Data management and analysis methods. In
Denzin, N. K. & Lincoln, Y. S. (Eds.), Handbook of Qualitative Research: Thousand
Oaks, CA.: Sage.
Jick, T. D. 1979. Mixing qualitative and quantitative methods: Triangulation in action.
Administrative Science Quarterly, 24(4): 602-611.
Kemmis, S. & McTaggart, R. 2000. Participatory action research. In Denzin, N. K. &
Lincoln, Y. S. (Eds.), Handbook of Qualitative Research: Thousand Oaks, CA.: Sage.
Kirby, D. A. & Kaiser, S. 2003. Joint ventures as an internationalisation strategy for
SMEs. Small Business Economics, 21(3): 229-242.
Lincoln, Y. S. & Guba, E. G. 1985. Naturalistic Inquiry. Beveerly Hills, CA.: Sage.
Locke, K. 2001. Grounded Theory in Management Research. Thousand Oaks, CA.:
Sage.
27
Mäkelä, M. M. & Turcan, R. V. 2004. Grounded theory in entrepreneurship studies:
Feasibility of the method and the design of research. PrePrint Publication, Software
Business and Engineering Institute, Helsinki University of Technology.
Miles, M. B. & Huberman, A. M. 1984. Qualitative Data Analysis. Beverly Hills,
CA.: Sage.
Miles, M. B. & Huberman, A. M. 1994. Qualitative Data Analysis: An Expanded
Sourcebook. Thousand Oaks, CA.: Sage.
Mintzberg, H. 1979. An emerging strategy of "direct" research. Administrative
Science Quarterly, 24: 582-589.
Pettigrew, A. 1992. The character and significance of strategy process research.
Strategic Management Journal, 13: 5-16.
Smith, L. M. 1994. Biographical method. In Denzin, N. K. & Lincoln, Y. S. (Eds.),
Handbook of Qualitative Research: Thousand Oaks, CA.: Sage.
Steier, L. & Greenwood, R. 2000. Entrepreneurship and the evolution of angel
financial networks. Organization Studies, 21(1): 163-192.
Strauss, A. 1987. Qualitative Analysis for Social Scientists. New York: Cambridge
University Press.
Strauss, A. & Corbin, J. 1994. Grounded theory methodology: An overview. In
Denzin, N. K. & Lincoln, Y. S. (Eds.), Handbook of Qualitative Research: Thousand
Oaks, CA.: Sage.
28
Strauss, A. & Corbin, J. 1998. Basics of Qualitative Research: Techniques and
Procedures for Developing Grounded Theory. Thousand Oaks, CA.: Sage.
Strauss, A. L. & Corbin, J. 1990. Basics of Qualitative Research: Grounded Theory
Procedures and Techniques. Newbury Park, CA.: Sage.
Uzzi, B. 1997. Social structure and competition in interfirm networks: The paradox of
embeddedness. Administrative Science Quarterly, 42: 35-67.
Van Maanen, J. 1988. Tales of the Field: On Writing Ethnography. Chicago:
University of Chicago Press.
Yin, R. K. 1981. The case study crisis: Some answers. Administrative Science
Quarterly, 26: 58-65.
Yin, R. K. 1984. Case Study Research. Beverly Hills, CA.: Sage.
Yin, R. K. 1989. Case Study Research: Design and Methods. Newbury Park, CA.:
Sage.
Yin, R. K. 1994. Case Study Research: Design and Methods. Thousand Oaks, CA.:
Sage.
29
APPENDIX
Exemplar Entrepreneurship Articles Employing Case Study Approaches to
Grounded Theory or Similar Research
We used the World Wide Web and libraries to survey Administrative Science
Quarterly, Academy of Management Journal, Strategic Management Journal,
Organization Science, Management Science, Journal of Management, Journal of
Management Studies, Organization Studies, Journal of Business Venturing,
Entrepreneurship Theory and Practice, and Small Business Economics, for articles
published since 1996 with a significant entrepreneurship connection and that
employed a grounded theory or similar method in the sense that they produced
propositions forming a theoretical framework of some level. We also searched for the
Frontiers of Entrepreneurship Research – the conference proceedings publication of
the Babson College–Kauffman Foundation Entrepreneurship Research Conference.
Our objective in identifying and analyzing papers was finding out how do such
papers differ and what do they have in common in the various stages of research and
in the aspects of the report, and from more than 200 papers reviewed, we picked eight
papers for the purpose of illustrating the conduct of grounded theory and similar
approaches in practice and especially to illustrate the variations in published research.
Since the papers were selected merely for illustration, we selected them on grounds
that they appeared to relatively well represent what we, based on a subjective and
intuitive estimate, considered the population of such papers to represent. In other
words, we did not attempt to produce a detailed analysis of various types of
qualitative research and its appearance in different scholarly outlets, but rather
browsed the literature for examples that would serve this and the companion paper
30
well by illustrating our points that we wish to emphasize about making grounded
theory research.
We explicitly searched for those papers addressing the high-growth venturing
type of entrepreneurship and not general ‘small business’ literature.
The eight papers have many similarities that are also shared by a majority of
other papers we reviewed: The research problem is often clearly expressed (although
not always explicitly), links to prior literature are presented both in stating the
position and justification for the research and in articulating the value and
contribution of the findings. Review of prior literature was typically presented in the
beginning of the paper as is typical with hypothesis-testing research; Dougherty
(2002) notes that it is also possible to begin the discussion on earlier research only
after presenting the data and analysis in the report.
While all of the articles acknowledge the existence of some earlier theory that
has implications for the research, they build their theory on existing frameworks and
articulate links thereto; Amit and Zott (2001) had the objective of integrating several
earlier theories. Common to the papers was also that they did not make much use of
quotes within text; some made none. Using illustrative quotes within tables was also
quite rare. One can ponder reasons for this and hypothesize that authors do not see
that offering detailed quotes could offer significant value while the number of them
would in any case be small so that quotes cannot convey a broad picture of the whole
data.
Pilot studies were very rare. In fact, only Amit and Zott (2001) and Uzzi
(1997) were found to report such features as we indicate in the column ‘Other
Information.’ All interviews that authors described were semistructured. One
observation which may have important implications if generalizable to a wider
31
population is that papers seldom used tables for illustrating how the causal or process
relationships of the emergent theory were present in the data. Such presentation would
arguably increase the credibility of a grounded theory study (Eisenhardt, 1989; Miles
and Huberman, 1994, Dougherty, 2002).
In a majority of our exemplar papers introduced the main contents of their data
in tables, and some also discussed it in the text. A few papers, typically those with one
or two cases only, included a lengthy discussion of the case. Many of the papers
provided a vague description of the methodology, especially analysis, and in some
papers the discussion was lacking. It was rare that the quality of the research was
discussed in a dedicated section of the paper and with using a thorough set of labels
for aspects of quality, such as construct, internal, and external validity and reliability
proposed by Yin (1994). Many papers did, however, more or less explicitly address
several aspects of validity in their methods section. Finally, a number of names was
used for similar methods, including ‘grounded theory,’ ‘case study,’ ‘inductive
research,’ ‘comparative case study,’ ‘analytical induction,’ and ‘naturalistic inquiry.’
Further details where the papers significantly differ are introduced in the table below.
Table: Examples of case study approaches to grounded theory or similar researcha
Article Journalb Topic Key Findings Method Stated Justification for Method Key
Methodology References
Sampling
(Amit and Zott, 2001)
SMJ Value creation in e-business
‘Efficiency, complementarities, lock-in, and novelty’ identified as key factors underlying value creation potential. Integration of earlier theories sought via offering a business model framework
Inductive; case study; iterative; ‘grounded theory’ mentioned
Existing theories inadequate; ‘lack of prior theorizing about a topic’ (p. 500); ‘grounding of the emerging theory in the data can provide a new perspective on an already researched topic’ (e.g. Hitt et al., 1998, p. 500); boundaries between phenomenon and context not evident (with reference to Yin, 1981); replication sought via multiple case approach
Eisenhardt, 1989; Glaser and Strauss, 1967; Miles and Huberman, 1984; Yin, 1981; 1989
30 US Internet businesses random sampled; all 29 European firms included; authors acknowledged that theoretical sampling can be argued for
(Barringer and Greening, 1998)
JBV Entrepreneurial small businesses’ growth via geographic expansion
Identification of antecedents for effective expansion
Inductive; multiple cases
Little research on the ‘topic’; nature of the case method suited for early stages of theory development
Denzin, 1978; Eisenhardt, 1989; Glaser and Strauss, 1967; Lincoln and Guba, 1985; Strauss, 1987
Five cases theoretically sampled; polar types in terms of success were chosen
(Bergmann Lichtenstein and Brush, 2001)
ETP Development and change of resource bundles in new ventures
Types of salient resources identified; types of change in them identified; a link between the type of change and performance identified
Longitudinal study of three cases; termed exploratory by authors
Lack of empirical ‘validation’ of RBV in new ventures; avoiding retrospective bias; the increasing use of comparative case studies in entrepreneurship
Several, e.g. Denzin, 1978; Eisenhardt, 1989; Miles and Huberman, 1984; Pettigrew, 1992
Three ventures that fitted the scope were chosen; no further justification was provided
(Birkinshaw, 1997)
SMJ Characteristics of subsidiary initiatives in MNEs; corporate entrepreneur-ship
Subsidiary entrepreneurship can enhance local responsiveness and global learning and integration; contextual mechanisms limited in creating differentiated subsidiary roles
Inductive; multiple cases
Quality and richness of data; multiple initiatives per subsidiary to avoid selection bias; understudied topic
Yin, 1984; Miles and Huberman, 1984
39 initiatives in six subsidiaries; variation in key attributes was sought; also some convenience sampling; sampling stopped according to the principle of saturation
33
Article Journalb Topic Key Findings Method Stated Justification for Method Key
Methodology References
Sampling
(Garud et al., 2002)
AMJ Institutional entrepreneurship in a firm’s sponsorship of its technology
Standards-in-the-making generate seeds of fragmentation; social and political skills identified that a sponsor uses to overcome challenges
Inductive single-case study, ‘naturalistic inquiry’ with an aim to ‘generalize to a theory’ (p. 199); iterative, systematic analysis
Complexity of the phenomenon makes a naturalistic, interpretive, and inductive study feasible
Glaser and Strauss, 1967; Jick, 1979; Lincoln and Guba, 1985; Miles and Huberman, 1984
Explicit discussion on why Sun Microsystems and its Java technology were chosen as study objects
(Kirby and Kaiser, 2003)
SBE JV strategies as means of SMEs to overcome internationalization restrictions
Outline of problems in JV strategy; similarities of experiences to MNEs; JV partner selection and obtaining help thereby is important
Questionnaires from 21 SMEs; descriptive statistics only; 2 mainly illustrative cases
Not explicitly provided; newness of the mentioned; work is rather exploratory but provides ‘loose propositions’
Dillman, 1978 for survey research; no case study references; methods very little discussed
All UK and German SMEs with a JV in China were sought after
(Steier and Greenwood, 2000)
OS Development and evolution of angel financial networks of entrepreneurial firms
Network efficiency is positively associated with entrepreneurs’ dependence of investors; Entrepreneurs avoid dependency by creating redundant networks or by building robust ties to individual investors
Longitudinal single-case study, ‘subjectivist’ approach
The aim is theory development; not much prior theory on the subject; single-cases especially well-suited for identifying processes
Eisenhardt, 1989; Miles and Huberman, 1984; Mintzberg, 1979; Yin, 1984
No devoted discussion; a case with an angel financier network was just picked
(Uzzi, 1997) ASQ Embeddedness and competition in interfirm networks
Embeddedness is a logic of exchange that promotes economic performance up to a threshold and derails it beyond it
Ethnographic ‘grounded theory’ of relationships of 23 entrepreneurial firms (interfirm tie as the unit of analysis)
‘Field methods’ viewed as advantageous due to their providing rich data for in-depth analysis and theorizing
Glaser and Strauss, 1967; Miles and Huberman, 1984
Selection from New York’s women’s ‘better-dress’ firms; variation sought in a number of variables; stated reasons for this were representativeness and controlling variables
34
Article Key Data
Typesc Analysis Use of Displays Presentation of
Propositions Assessment of Quality Other Information
(Amit and Zott, 2001)
Archival documents: www sites, analysts’ and annual reports; IPO prospectuses, SEC’s database; supposedly no interviews were made
The overlap of data collection and analysis exploited; within- and cross-case analyses; field notes; ‘no narratives’; data and researcher triangulation; tabulation of evidence
Model diagram; illustrations of value sources; table linking emerging value constructs to prior theories; links between the value source and business model frameworks
Embedded within article text; presented throughout a majority of the paper
Many aspects of reliability and validity discussed in various terms; external validity also discussed in terms of statistical generalization
Pre-announced attempt to integrate earlier theories; statistical study features present (large number of cases; random sampling; statistical generalization problems discussed); a pilot study-like feature of changing the questionnaire after the beginning
(Barringer and Greening, 1998)
Apparently 2-3 interviews per case; archival documents
Coding with ‘typical’ content analysis procedures; coding used not only emergent categories but also categories directly from prior literature in its first phase; data triangulation noted
Model diagram; case descriptions; degree of case firms’ achievement of success characteristics
Embedded, throughout the latter half of the paper
Aspects of validity discussed in various parts of the paper
(Bergmann Lichtenstein and Brush, 2001)
Interviews – altogether 13 site visits in three cases, 5-15 people interviewed during each visit; documents
Systematic and iterative analysis of content; pattern matching; coding for process identification
Earlier research; data collection; case timelines; exploratory findings; enumeration of change type frequencies in case data
Explicit; in the end Explicit assessment of different types of validity
(Birkinshaw, 1997)
Interviews, apparently 100; questionnaire
Miles and Huberman’s (1984) procedures used for data reduction; case histories were used for identifying propositions; coding not discussed
Prior research; description of interviews and cases; model diagram; questionnaire data; summary of findings
Embedded; in the end
Several aspects of validity discussed in various parts of the paper
Much quantitative data and statistical testing included
35
Article Key Data Typesc Analysis Use of Displays Presentation of
Propositions Assessment of Quality Other Information
(Garud et al., 2002)
Archival documents: online news services; articles, press releases
A chronology constructed; iteration until theoretical saturation; data triangulation mentioned; coding not mentioned
Model diagram; list of events of the case timeline
Embedded; emerge throughout the empirical part but more in the end
‘Procedural adequacy’ and ‘credibility’ emphasized and ‘established by employing steps that Miles and Huberman (1984) suggested’ (p. 199); data triangulation noted; validity assessed; a decision made as to which level of generality to express results at
A long case description along timeline
(Kirby and Kaiser, 2003)
Questionnaire and apparently a few (2-4) interviews
Qualitative analysis not discussed; descriptive statistics drawn from questionnaire data
Displays of descriptive statistics
Embedded Embedded, some discussion on generalizability problems
Cases used just for illustration; cases apparently based on data from one or two interviews only; propositions drawn from descriptive statistics; a vague connection to theory; long case descriptions
(Steier and Greenwood, 2000)
30 ‘formal’ and 16 ‘less formal’ interviews; documents
Methods only briefly mentioned; iteration; analysis ‘not sequential nor structured - - as is normal in subjectivist approaches’ (p. 170)
Interviews; case firm and financial network descriptions
Explicit in the end; discussion part very long
Some aspects of reliability and validity explicitly discussed
A long case description
(Uzzi, 1997) 42 interviews; additional discussions with case-independent industry experts
Iterative rotation at least within data and the emerging framework; use of displays for data reduction
Case characteristics; amount of interviewing; a diagram for illustrating interfirm networks; enumeration of frequencies of emerging constructs in the data; model diagram
Explicit; in the end Many aspects of validity and reliability discussed; external validity discussed in terms of statistical generalization (’23 cases - - can have but moderate generalizability’, p. 38)
Two pilot study interviews used; a large number of propositions: 16
a Abbreviations used in table, excluding journal abbreviations: IPO, initial public offering; JV, joint venture; MNE, multinational enterprise; RBV, resource-based view of the firm; SEC, U.S.
Securities and Exchange Commission; SME, small or medium-sized enterprise. b Abbreviations for the journals are: AMJ, Academy of Management Journal; ASQ, Administrative Science Quarterly; ERD, Entrepreneurship and Regional Development; ETP,
Entrepreneurship Theory and Practice; JBV, Journal of Business Venturing; JM, Journal of Management; JMS, Journal of Management Studies; OS, Organization Studies; SBE, Small Business Economics; SMJ, Strategic Management Journal.
c In practice, we conclude that observation data has been in an important role in all or many papers with interviews even though it was rarely mentioned.