A Scientific Methodology for MIS Case Studies

Author(s): Allen S. Lee

Source: MIS Quarterly , Mar., 1989, Vol. 13, No. 1 (Mar., 1989), pp. 33-50
Published by: Management Information Systems Research Center, University of
Minnesota

Stable URL: https://www.jstor.org/stable/248698

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Management Information Systems Research Center, University of Minnesota is collaborating

with JSTOR to digitize, preserve and extend access to MIS Quarterly

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

A Scientific Introduction
There is a strong case-study tradition in the aca-
Methodology for MIS demic field of management information systems
Case Studies (Benbasat, et al., 1987; Fulk and Dutton, 1984;
Kling, 1978; Kling and lacono, 1984; Kling and
Scacchi, 1982; Kraemer, et al., 1987; Laudon,
1974; Leonard-Barton, 1987; Markus, 1983;
By: Allen S. Lee 1986). At the same time, case researchers in
502 Hayden Hall general are still attempting to clarify the
Northeastern University methodological basis upon which to conduct
Boston, MA 02115 case studies (Benbasat, et al., 1987; Datta,
1982; Dukes, 1965; George and McKeown,
1985; Herriot, 1982; Hersen and Barlow, 1976;
Huberman and Crandall, 1982; Louis, 1982; Lu-
thans and Davis, 1982; Miles, 1979; 1982; Yin,
1981a; 1981b; 1982a; 1982b; 1984). The ob-
jective of this article is to present a scientific meth-
odology with which to conduct case studies of
Abstract management information systems (MIS). In
doing so, the article applies and builds upon con-
A methodology for conducting the case study
cepts that pertain to case-study methodology
of a management information system (MIS) is
and that the author developed in his previous
presented. Suitable for the study of a single
research (Lee, 1985; 1986; 1987b; forthcoming).
case, the methodology also satisfies the stan-
dards of the natural science model of scientific In order to provide a practical demonstration of
research. how these methodological concepts are usable
in case studies of management information sys-
This article provides an overview of the tems, the article illustrates them with extensive
methodological problems involved in the studymaterial taken from an actual, published MIS
of a single case, describes scientific method, case study - Markus' (1983) "Power, Politics,
presents an elucidation of how a previously pub-and MIS Implementation."
lished MIS case study captures the major fea-
In particular, this article (1) provides an over-
tures of scientific method, responds to the prob-
view of the methodological problems involved
lems involved in the study of a single case, and
in the study of a single case, (2) offers a de-
summarizes what a scientific methodology for
scription of scientific method, (3) elucidates how
MIS case studies does, and does not, involve.
the MIS case study by Markus fits this descrip-
The article also has ramifications that go beyondtion, (4) responds to the problems involved in
matters of MIS case studies alone. For MIS re- the study of a single case, and (5) summarizes
searchers, the article might prove interesting for what the article's scientific methodology for MIS
addressing such fundamental issues as whether case studies does, and does not, involve.
MIS research must be mathematical, statistical,
or quantitative in order to be called "scientific."
For MIS practitioners, the article's view of sci-
entific method might prove interesting for em- What is science?
powering them to identify, for themselves, the
The formulation of a particular scientific meth-
point at which scientific rigor is achieved in an
odology for conducting MIS case studies and
MIS research effort, and beyond which further
identifying the methodological problems associ-
rigor can be called into question, especially if ated with this type of research depends on what
pursued at the expense of professional is meant by "science" in the first place. In deter-
relevance.
mining this meaning, there are numerous models
Keywords: Information systems, case studies, of science from which to select. Indeed, philoso-
research methods, research design, phers of science - the scholars who make it
organizational impacts their job to observe scientists and to explain what
it is that scientists do - have not yet settled,
ACM Categories: K.4.3, K.6.0, K.6.1 among themselves, on a single model of what

MIS Quarterly/March 1989 33

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

science is.' The model used in this article is whatby case researchers to formulate theories (Ben-
social scientists call the "natural science model" basat, et al., 1987; Filstead, 1970; Garfinkel,
of social science (Behling, 1980, p. 483; Burrell 1967; Geertz, 1973; Kirk and Miller, 1986; Lee,
and Morgan, 1979, p. 4; Daft, 1983, p. 539; forthcoming; Louis, 1982; Sanday, 1979; Taylor,
Schon, et al., 1984, p. 9; Susman and Evered, 1979; Van Maanen, 1979; Yin, 1981a; 1981b;
1978, pp. 582-583). According to this model, natu- 1982b; 1984) may therefore still be applied in
ral science is the ideal on which social science addition to the methods specified by the natural
should model itself. There are three reasons for science model. Thus, this article respects and
selecting this particular model of science. preserves the traditional function of case stud-
ies in suggesting hunches and generating theo-
First, among management researchers, the natu- ries for later testing - a function recognized by
ral science model is a well-known and widely scholars of all persuasions. Specifically, the meth-
accepted model for conducting studies in social odology formulated in this article (1) allows MIS
science.2 As such, the natural science model sug- case researchers to continue using the tools they
gests itself as a useful device for introducing schol- have traditionally used, and (2) enables MIS
ars, unfamiliar with case studies, to this type of case researchers to conduct case studies that
research.
test theories by using the natural science model.
Second, many of the criticisms directed against
case studies are voiced from the perspective of
the natural science model. It is the critics of case Methodological Problems
studies, not scholars already working in the case-
study tradition, who need to be convinced of the
Raised by the Study of a
legitimacy of case studies. Recognizing this, the Single Case
article demonstrates the legitimacy of case stud- In this article an MIS case study refers to the
ies by using the standards of the critics examination of a real-world MIS as it actually
themselves.3 exists in its natural, real-world setting. In hold-
ing MIS research to the standard of the natural
Third, the article recognizes that a scientific meth-
science model, four problems can be identified
odology, which applies the natural science in MIS research that is conducted in the form
model, actually complements and supports the
methods traditionally associated with case stud-
ies. The natural science model is primarily a 3 On the one hand, the natural science model of social
model for testing theories, not formulating theo- science represents a view of science taken by many
ries in the first place. Methods traditionally used (if not most) social scientists; see the quotation in
the previous footnote, as well as Burrell and Morgan
(1979, p. 4), Daft (1983, p. 539), Schon, et al. (1984,
"'Is scientific advance registered by increasing prob- p. 9), and Susman and Evered (1978, pp. 582-583).
ability (Carnap), by discrete verifications (the logical On the other hand, the philosophy of science would
positivists) or falsifications (Popper), by revolutions recognize the natural science model to be a descen-
(Kuhn), by growing consensus (Polanyi), by progres- dant of logical positivism - a view of science that
sive versus degenerating research programs, con- philosophers of science themselves had originated,
ducted over long periods (Lakatos), or what? The but subsequently abandoned (Bernstein, 1978; 1983;
debate goes on" (Tice and Slavens, 1983, p. 418). Schon, 1983, pp. 37-49). Whereas it is the article's
responsibility to point out the disparity in viewpoints
of these two communities of scholars (philosophers
2 In his article, "The Case for the Natural Science
of science on the one hand, and many (if not most)
Model for Research in Organizational Behavior and
social scientists on the other), it is beyond the arti-
Organization Theory," Behling (1980) advocates the
natural science model: "Research methods similar cle's scope to investigate the ramifications of the dis-
parity, much less to resolve it. For a detailed exami-
to those used in the natural sciences have long been
nation of this matter, see Lee (1987b; forthcoming).
the norm in organizational behavior and organiza-
tion theory" (p. 483). He adds: "Clearly, the authors Practically speaking, the audience for this article -
of mainstream texts in organizational behavior and management scholars critical of, or unfamiliar with,
organization theory accept the natural science modelcase studies - largely subscribe to the natural sci-
of good research. Those who include research meth-ence model. For this reason, an effective strategy
ods chapters ... clearly follow this approach andfor approaching this audience would be to proceed
generally appear to owe an intellectual debt to Ker- with a framework they already find familiar and ac-
linger, a strong proponent of the natural scienceceptable - namely, the framework of the natural sci-
model" (p. 483). ence model.

34 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

of case studies. These problems are discussed wrong?"(Miles, 1979, p. 590; emphasis in the
in the next four sections. original).

Problem 1: Making controlled Problem 3: Allowing for replicability

observations The third problem concerns how to allow for repli-
The first problem concerns how to make con- cability. Research in the natural sciences is rou-
trolled observations. In testing for relationships tinely replicated as a means of assuring the ob-
theorized to exist among different factors, jectivity of the research. However, the MIS case
natural scientists routinely observe the influence researcher is unlikely to observe the same set.
of one factor on another factor, where the po- of events - namely, the same configuration of
tentially confounding influences of all other lac- individuals, groups, social structure, hardware,
tors are somehow removed or "controlled for." and software - unfold again in the same way.
Laboratory experiments in the natural sciences The non-replicability of the same observations
accomplish this through the use of control would clearly hinder subsequent attempts by in-
groups and treatment groups. In statistical ex- dependent investigators wishing to verify the find-
periments, it is accomplished with the help of ings of a particular case study.
statistical controls, such as those available
through a multivariate regression analysis. Un-
fortunately for the MIS case researcher, (1) the Problem 4: Allowing for
study of a real-world MIS in its real-world set- generalizability
ting precludes, by its very nature, the laboratory
controls of laboratory experiments, and (2) the The fourth and last problem concerns how to
study of a single case commonly yields more allow for generalizability. An often admired qual-
variables than data points - a situation that ren- ity of theories in the natural sciences is their ap-
ders inapplicable the statistical controls of sta- plicability to a range of settings. (In this sense,
theories in the natural sciences are said to be
tistical experiments (Yin, 1981b).
"nomothetic," as opposed to "idiographic.") How-
ever, the fact that the study of a single case
is marked by unique and non-replicable events
Problem 2: Making controlled would make the study vulnerable to charges that
deductions its findings cannot be extended to other settings.
The second problem concerns how to make con-
trolled deductions. Making controlled or logical The following section describes a scientific meth-
deductions with mathematical propositions - as
odology for MIS case studies - a methodology
is commonly done in the natural sciences- is that follows from the natural science model. Use
a standard, non-controversial practice. However, of this methodology allows the final section of
since it is rare (though certainly not undesirable) this article to address the problems (identified
for a case study to be quantitative, the MIS case above) associated with case studies.
researcher is typically denied the methodological
convenience of working with numerical data and
mathematically stated propositions. Instead, the
case researcher must somehow manage with A Description of Scientific
qualitative data and verbally stated propositions. Method
Making controlled deductions with verbal propo- In modeling MIS case studies on natural science,
sitions (i.e., qualitative analysis), while certainly we must ask: How does inquiry in natural sci-
possible, is more problematic: "For quantitative ence proceed? Specifically, what is meant by
data, there are clear conventions the researcher "the natural science model"?
can use," such as the widely accepted and well-
known rules of algebra through which the valid- In his classic text, Introduction to Logic, Copi
ity of mathematical deductions is known, "but (1986) provides a lucid description of the logic
the analyst faced with a bank of qualitative data of reasoning used in the natural sciences. "Few
has very few guidelines for protection against propositions of science," he explains, "are di-
self-delusion. . . How can we be sure that [a rectly verifiable as true. In fact, none of the im-
qualitatively deduced] finding is not, in fact, portant ones are. For the most part they con-

MIS Quarterly/March 1989 35

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

cern unobservable entities, such as molecules whose school of thought is a rival to Popper's,
and atoms, electrons and protons, chromosomes expresses that "no field is potentially a science"
and genes" (p. 483). As a result, the manner unless its theories are cast according to "Sir
of verification in the natural sciences is indirect Karl's demarcation criterion" (p. 245). The now
rather than direct. "The pattern of indirect test- common characterization of theories, in both natu-
ing or indirect verification consists of two parts. ral science and social science, as falsifiable, re-
First one deduces from the proposition to be futable, testable, or disconfirmable, is an indica-
tested [the proposition being the theory] one or tion of the widespread extent to which the
more other propositions capable of being tested deductive testing of theories is practiced.
directly [these latter propositions being the pre-
Falsifiability is just one requirement that a theory
dictions]" (p. 486). In the terminology of logic,
must satisfy in order to be scientific. There are
a theory's predictions are its conclusions. "Then
these conclusions are tested and are found to three additional requirements, which are all as-
be either true or false." The researcher then com- sociated with the concept of the deductive test-
pares what the theory predicts and what is actu-
ing of theories (Popper, 1968, pp. 32-33). One
of these requirements is logical consistency: as
ally observed. "If the conclusions are false, any
long as the different predictions that may be de-
proposition that implies them [namely the theory]
must be false also. On the other hand, if the duced from the theory are not mutually contra-
conclusions are true, that provides evidence for
dictory, the theory can be said to be logically
consistent. Another requirement is that the
the truth of the proposition being tested, which
theory must be at least as explanatory, or pre-
is thus confirmed indirectly" (p. 486).
dictive, as any competing theory. The last re-
Karl Popper (1968) describes the same proce- quirement is that the theory, while falsifiable,
dure in The Logic of Scientific Discovery, where must survive the actual attempts made at its
he calls it the deductive testing of theories (pp. falsification.
32-33, p. 60, pp. 109-111). It is deductive in the
In the way that scientific method appears in the
following sense. The natural scientist applies a
natural science model, the notion of controlled
theory (for example, "All men are mortal") to a
set of facts or initial conditions ("Socrates is a observation comes into play in the last step
man"), from which a conclusion or prediction is
(namely, the step where the researcher makes
deduced ("Socrates is mortal"). It is the predic-
a comparison between what is predicted and
tion - as a deduced statement - that is then what is observed). In this step, the researcher
must be able to show that the observed effect
tested against an observation statement (for ex-
can be attributed to the factor being tested and
ample, "Socrates dies").
that the potentially confounding effects of other
In this procedure, an observation that contra- factors have been removed or "controlled for"
dicts a prediction would be sufficient to cast (Campbell and Stanley, 1963). This article has
doubt on (perhaps to the point of falsifying) thealready mentioned laboratory controls and sta-
theory from which the prediction follows. On the tistical controls as examples of how observations
other hand, an observation that confirms a pre-can be made in a controlled way.
diction is never regarded as conclusively estab-
lishing the theory's truth. The reason is that a
different set of empirical circumstances, or initial
conditions, to which the same theory may be An Exemplar for Scientific
applied would result in yet another prediction MIS Case Studies
(e.g., "Plato is mortal" or "Superman is mortal"),
which in turn would open up the same theory Markus' (1983) "Power, Politics, and MIS Im
to yet another opportunity for its falsification. plementation" captures the major features of sci
Thus, the ever-present possibility for contradic- entific method that Copi and Popper describe
tory evidence to surface in a subsequent test As such, the MIS case study by Markus ma
requires that a theory always be regarded asbe regarded as an exemplar for scientific MIS
falsifiable. Indeed, falsifiability is the demarca- case studies in general, where the meaning of
tion criterion that Popper uses to distinguish sci- "scientific" is the one embodied in the natural
ence from non-science (pp. 40-42). science model.

Scientific method, in the form of the deductive Markus' research is the intensive study of a
testing of theories, is widely known. Kuhn (1970),single case, involving the entire configuration of

36 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

individuals, groups, social structure, hardware, describes the interaction theory in the following
and software in the setting of an organization ways:
that she calls Golden Triangle Corporation
(GTC). Through interviews and documents, ... resistance is explained as a product of
Markus observes events in the way they un- the interaction of system design features
folded in their natural setting at GTC. with the intraorganizational distribution of
power, defined either objectively, in terms
By conducting her research in this way, Markus of horizontal or vertical power dimensions,
invokes the severe methodological problems men- or subjectively, in terms of symbolism (p.
tioned earlier. They are the problems of: (1) how 432).
to make controlled observations, (2) how to
Resistance-generating conditions are mis-
make controlled deductions, (3) how to allow for
matches between the patterns of interac-
replicability, and (4) how to allow for tion prescribed by a system and the pat-
generalizability.
terns that already exist in the setting into
which the system is introduced (p. 438).
Despite these problems, the MIS case study by
Markus still succeeds in crafting a theory about The primary assumption ... is that informa-
MIS implementation that conforms to the require- tion systems frequently embody a distribu-
ments of falsifiability, logical consistency, pre- tion of intraorganizational power among the
dictive power exceeding that of competing theo- key actors affected by its design. Intraor-
ries, and survival of the empirical tests aimed ganizational power is an attribute of indi-
at falsifying it. viduals and subgroups ....

Markus presents three alternative theories on an ... When the introduction of a computerized
equal footing and then compares the deductions information system specifies a distribution
(the predictions) of each against observations of power which represents a loss to certain
made in the setting at GTC. All are theories participants, these participants are likely to
about resistance to MIS implementation efforts. resist the system. Conversely, when the dis-
The case itself involved people's resistance to tribution of power implied in the design of
GTC's newly computerized Financial Information an information system represents a gain in
System (FIS). power to participants, these participants are
likely to engage in behaviors that might sig-
The people-determined theory involves "factors nify acceptance of it ....
internal to the person" (p. 431). When people
factors such as human nature, cognitive styles, ... [A necessary condition for resistance to
or personality traits are incompatible with the re- the implementation of a system is that]
quirements of a computerized information people perceive the system to represent a
system, the system's intended users, according power loss ... (p. 442).
to the people-determined theory, will resist its
utilization.
All three theories refer extensively to the exis-
tence of phenomena that are neither directly ob-
The system-determined theory involves "factors
servable nor easily discernible: human nature,
inherent in the application or system being im-
cognitive styles, personality traits, factors inter-
plemented" (p. 431). Markus cites the following
nal to the person, user friendliness, technical de-
as examples of system factors that incur resis-
ficiency in a system, ergonomics, horizontal and
tance: lack of user-friendliness, technically defi-
vertical dimensions of intraorganizational power,
cient systems, and poor ergonomic design (p.
power in terms of symbolism, and (perhaps most
431). According to the system-determined
important and most directly unobservable of all)
theory, when such factors are present, the
the distribution of power implied in the design
system's intended users will resist its utilization.
of an information system. None of the three theo-
The interaction theory, the most sophisticated ries is therefore directly verifiable as true.
of the three theories, involves people factors as
well as system factors. However, Markus says, In this context, Copi's remarks on scientific
"This explanation identifies neither the system method, already quoted above, are worth repeat-
nor the organizational setting as the cause of ing: "Few propositions of science are directly
resistance, but their interaction" (p. 431). She verifiable as true. In fact, none of the important

MIS Quarterly/March 1989 37

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

ones are. For the most part they concern unob- was installed, informants in the [resisting] divi-
servable entities, such as molecules and atoms, sions still spoke of FIS [Financial Information
electrons and protons, chromosomes and System]. Many felt strongly that the system
genes." Hence, because direct verification is not should be replaced..." (p. 438). Markus reports
feasible, theoretical propositions are tested indi- no mitigation in the resistance - a mitigation
rectly instead. This manner of testing involves that the system-determined theory predicts.
what Popper calls the deductive testing of theo- These observations falsify the system-deter-
ries, the details of which were described earlier mined theory.
in the article.
The interaction theory predicts that nei-
As it turns out, Markus provides an exemplary ther changing the people (by removing
demonstration of how an MIS case study is ca- them, by educating them, or by attempt-
pable of carrying out the deductive testing of theo-
ing to coerce them), nor changing techni-
ries. She tests for the truth of the three theories cal features of the system will reduce re-
(and for the presence of the unobservable phe- sistance as long as the conditions which
nomena) in the following way. Whereas the three
gave rise to it persist. [The prediction is
theories refer to phenomena that are not directly that there will be resistance as long as
observable, they nonetheless yield predictions there are] mismatches between the pat-
of events that (if the given theory is true) would
terns of interactions prescribed by a
be observable. Thus Markus' strategy is to use system and the patterns that already exist
the contrary theories to make contrary predic- in the setting into which the system is in-
tions about what would happen in the same set- troduced (Markus, 1983, p. 438).
ting. The theory that emerges unfalsified in this
competition would be judged scientific. Markus describes the interaction pattern that
was already in place at GTC as consisting of
"The people-determined theory leads to the pre- autonomy experienced by the divisional account-
diction that replacing individual resistors or co- ants and dependence experienced by the cor-
opting them by allowing them to suggest improve- porate accountants. The divisional accountants
ments to the system might reduce or eliminate controlled their own data (often in thick, manu-
resistance" (p. 437). However, resistance to the ally maintained ledger books) and could there-
new information system persisted despite GTC's fore reconcile unusual situations before releas-
practice of job rotation and mobility. Markus ing reports. The corporate accountants had to
gives the example of an accountant, one of the go through the divisional accountants to obtain
designers and advocates of the system, who had financial data, which was "a valued resource"
originally been working in corporate accounting (p. 438). This conflicted with the interaction pat-
and then became the controller in one of the
tern prescribed by FIS, under which "all finan-
divisions that had resisted the system all along. cial transactions were collected into a single [com-
He subsequently came to resist the system him- puterized] database under the control of
self. This observation falsifies the people- corporate accountants .... At any time, corpo-
determined theory. rate accountants had the ability to 'look into'
the database and analyze divisional perform-
"The system-determined theory predicts that ance" (p. 438). Markus' observation showed re-
fixing technical problems eliminates resistance" sistance to FIS, just as predicted by the interac-
(p. 437). However, the resistance continued de- tion theory.
spite corrective actions taken to address a
number of major technical problems. Signifi- The interaction theory satisfies the four require-
cantly, the problems were identified by a task ments that Popper observes to be satisfied by
force whose members were (as characterized all scientific theories. First, it is falsifiable (e.g.,
by Markus) "resistors." The corrective actions the interaction theory would have been falsified
were the installation of a larger computer with if Markus had observed acceptance of FIS de-
a more powerful operating system, a change in spite the difference between the interaction pat-
processing mode from batch to online, and a tern that was in place and the one that FIS pre-
simplified method in the software by which man- scribed). Second, its logical consistency is
agers could create new accounts. Markus (1983) known through the mutual compatibility of the
writes, "when data were collected for this study, different predictions that Markus considers (pp.
about one year after the last of these changes 437-438). Third, it is confirmed, not falsified, by

38 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

the observations in the GTC case study. Fourth, How to make controlled
and most important, its predictions succeed,
whereas the predictions of its rival theories -
observations
the people-determined theory and the system- A critic of cate research may point out, correctly,
determined theory - fail. In crafting the interac- that Markus fails to utilize either laboratory con-
tion theory so that it satisfies the four require- trols or statistical controls when making obser-
ments, Markus not only attains her specific re- vations to test the three theories. However,
search goal of explaining resistance to MIS Markus solves the problem of how to make con-
implementation efforts, but also demonstrates trolled observations by utilizing natural controls.
how an MIS case study is able to capture the
major features of scientific method in the way
that scientific method is embodied in the natural
A simple but clear example of this (referred to
science model. earlier) is Markus' test of the people-determined
theory, in which a particular accountant, upon
It must be emphasized that the conclusions moving from his position in corporate account-
drawn by Markus in her case study are only ten- ing to controller in one of the divisions, changes
tative at best. After all, (1) there may exist (or from being an advocate of FIS to one of its re-
come into existence) some corporate account- sistors. This particular test "controls for" or
ants at GTC who, when transferred into one of "holds constant" the people factors by focusing
the divisions, will continue to accept, and never on just one person (the accountant), and "varies"
come to resist, FIS; (2) the improvements in the or "treats" the situation external to the person
technical features of the system may not yet by observing his move from corporate account-
have reached the threshold at which the resis- ing to a division. Thus, Markus is able to cleanly
tance would diminish observably; and (3) the re- attribute the accountant's new behavior (resis-
sistance may persist even when the interaction tance to FIS) to the "treatment" (the change in
pattern, required by FIS, comes to be more like the situation external to the accountant) rather
the interaction pattern already in place. These than to the "control" (the people factors - the
possibilities, however, do not weaken Markus' factors internal to the accountant). Indeed, by
case study, but actually strengthen it by empha- making this controlled observation, Markus fal-
sizing the extent of the interaction theory's falsi- sifies the people-determined theory, which pre-
fiability and allowing the case researcher to im- dicts no change in behavior where there is no
prove the theory by pointing out where later change in people factors.
surprises may occur. No scientific explanation
- whether Markus' interaction theory or a theory In utilizing natural controls and treatments to test
of physics - may ever be conclusively proven predictions, the case researcher must do more
true. According to the logic of the deductive test- than wait passively for desired controls and treat-
ing of theories, a theory can only be shown to ments to materialize. Rather, the case re-
be false, or not (yet) false. In scientific research, searcher must actively apply his or her ingenu-
further tests are always in order. ity in order to derive predictions that take
advantage of natural controls and treatments
either already in place or likely to occur. For ex-
ample, in Markus' prediction concerning the
people-determined theory, the control (the hold-
ing constant of people factors) is already in place
Response to the Problems by virtue of focusing on just one person, and
Involved in the Study of a the treatment (the variation in the environment)
takes advantage of the person's move from one
Single Case part of the organization to another. In general,
This section discusses the four severe it is incumbent upon the case researcher to scan
methodological problems associated with case the empirical material for the presence of natu-
research in the way they are manifested inral the
controls and treatments that may be incorpo-
case study by Markus. The discussion estab- rated into the formulation of a prediction. (This
lishes the context in which the article in its final
is no different from the activity of the statistician
section, will articulate what a scientific method-
who, in utilizing a multiple regression analysis
ology for MIS case studies does, and does not, to analyze 1980 census data, scans the data
involve. to identify what factors might serve as the ap-

MIS Quarterly/March 1989 39

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

propriate independent variables and, hence, as (Namely, she deduces several different, verbally
the statistical controls). expressed predictions from the three different,
verbally expressed theories as applied to the ver-
MIS case researchers who wish to utilize natu-
bally expressed facts of the situation at GTC.)
ral controls will find themselves in good com- With regard to logical form, Markus' deductions
pany. Investigators in some of the natural sci- involving verbal propositions are identical to and
ences, such as astronomy, geology, and human no less valid than, the deduction of the verbal
biology, are also unable to conduct laboratory proposition, "Socrates is mortal" (the prediction)
experiments for obvious reasons and are there- from the two other verbal propositions, "All men
fore also prevented from utilizing laboratory con- are mortal" (the theory) and "Socrates is a man"
trols in order to make controlled observations. (the facts or initial conditions).
Instead, these investigators routinely conduct natu-
ral experiments in which they utilize natural con- Like the situation pertaining to the utilization of
trols, through which they have been able to natural controls, MIS case researchers will find
achieve impressive results (Nagel, 1979, p. 452). themselves in good company with regard to analy-
MIS case researchers who invoke natural con- sis that utilizes the medium of verbal proposi-
trols would therefore be employing a research tions, as opposed to mathematical propositions.
strategy no different from, and no less scientific Consider biology and the theory of evolution. For
than, what is employed by these natural scien- Darwin, th'e medium of logical deduction was
tists. In using natural controls, MIS case research- words and sentences, not numbers and mathe-
ers would therefore be keeping within the natu- matics (Kaplan, 1964, pp. 245-246).
ral science model. To pursue this line of think-
ing would take us beyond the scope of this arti-
cle, but would lead to the conclusion that case
studies can be conducted as a form of natural
How to allow for replicability
experiment, which is already a conventional form How might an independent investigator go about
of research practiced in the "hard" sciences replicating the findings of the MIS case study
(Lee, forthcoming). by Markus?

One way - perhaps the most conceptually

straighfforward way - would be to attempt to
How to make controlled deductions replicate the case study in exactly the way that
Markus performs it. For the independent investi-
In qualitative analysis, as performed by Markus
gator, this would involve the attempt to apply
in her case study, how can deductions be made the same three theories to the same set of in-
in a controlled (i.e., logical) way? In mathemati-
itial conditions in order to deduce the same pre-
cal analysis, the validity of deductions involving
dictions as Markus, and then test these predic-
mathematical propositions can be readily
tions against the same observations made by
checked by turning to the rules of algebra. In
Markus. The obvious difficulty with this proce-
qualitative analysis, there is no corresponding
dure is that, in an MIS case study, any observed
body of rules as succinct or easily applied as
configuration of individuals, groups, social struc-
the rules of algebra for verifying the validity of
ture, hardware, and software in a real-world set-
deductions involving verbal propositions.
ting is highly unlikely to recur and be observed
To respond to this problem, it must first be em- again. Thus, an independent investigator could
phasized that mathematics is a subset of formal not verify the findings of the MIS case study by
logic, not vice versa. Logical deductions in the Markus, at least not through this conceptually
general case do not require mathematics. An straightforward procedure.
MIS case study that performs its deductions
Fortunately, there is at least one alternative pro-
with verbal propositions (i.e., qualitative analy-
cedure. The independent investigator could
sis) therefore only deprives itself of the conven-
apply the same theories as tested in the original
ience of the rules of algebra; it does not de-
case study to a different set of initial conditions
prive itself of the rules of formal logic, to which
(for example, the facts of the situation at AAA
it may therefore still turn when carrying out the
Corporation or XXX Corporation), thereby result-
task of making controlled deductions.
ing in different predictions (for example, if the
Indeed, Markus herself provides examples of con- people-determined theory is true, then individu-
trolled deductions involving verbal propositions. als who share the same people factors at XXX

40 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

Corporation will display no difference in their In other words, generalizability is a quality de-
level of resistance to, or acceptance of, the com- scribing a theory that has been tested and con-
puterized information system at XXX, regardless firmed in a variety of situations, whether such
of the rank and location of their position in the testing is conducted through case research, labo-
organization). In other words, the investigator ratory experiments, statistical experiments, or
would be working with a new prediction, "Plato natural experiments. As such, generalizability
is mortal," as opposed to the original prediction, poses no more, and no less, of a problem for
"Socrates is mortal"; even though the predic- MIS case research than it does for the studies
tion would be different, it would still be the same conducted in the natural sciences. In taking this
theory being tested. Such new predictions would position, the MIS case researcher would, again,
call for observations different from the ones be in step with the natural science model.
made by Markus and would therefore relieve the
independent investigator of the impossible task
of attempting to replicate the observations made
in the original case study. Consequently, even What a Scientific
though the observations in a particular MIS case
study are non-replicable, the case study's find-
Methodology for MIS Case
ings (that a particular theory is confirmed or dis- Studies Does, and Does Not,
confirmed) would be replicable. Involve
In suggesting how MIS case studies might be
carried out, this article has offered a scientific
methodology that involves (1) the deductive test-
How to allow for generalizability ing of theories, where (2) the theory must be
(a) falsifiable, (b) logically consistent, (c) more
The fact that Markus' case study of GTC is predictive than other theories, and (d) not falsi-
marked by unique and non-replicable events ren- fied by the tests it experiences. As such, this
ders it vulnerable to the charge that its findings scientific methodology is no different from, and
cannot be extended to other settings. However, therefore no less rigorous than, scientific meth-
such a criticism, applied to the study of a single odology as it is practiced as the natural sciences.
case, would be misplaced. A comparison to ex-
periments conducted in the natural sciences clari- At the same time, this article takes the position
fies the issue. that the scientific methodology of the natural sci-
ence model does not involve, as objectives, the
Consider a natural science theory that has so utilization of any of the following, even though
far been confirmed in just a single experiment they may often be regarded (in this article's view,
(whether a laboratory, statistical, or natural ex-
improperly) as necessary elements in scientific
periment). Of course, the theory would not be research: laboratory controls, statistical controls,
generalizable on the basis of the single experi- mathematical propositions, and replicable obser-
ment, since the experiment would have tested vations. Instead, each one of these happens to
the theory against just a single set of empirical be a means to an objective in scientific research
circumstances. Instead, the theory would be gen- rather than the objective itself. MIS case stud-
eralizable to other sets of empirical circum- ies are capable of achieving the same scien-
stances only on the basis of actually being con- tific objectives through different means.
firmed by additional experiments that test it
against those other sets of empirical circum- Laboratory controls and statistical controls, for
stances. The same point holds true for case stud- example, constitute a means to controlled ob-
ies. No theory concerning MIS would be gener- servation - an objective MIS case studies are
alizable on the basis of a single case study, able to achieve through natural controls. Like-
since the single case study would have tested wise, mathematical propositions constitute a
the theory against the empirical circumstances means to controlled or logical deduction - an
of just a single setting. Instead, the theory con- objective MIS case studies are able to achieve
cerning MIS would be generalizable to other set-
through verbal propositions that apply the rules
tings only on the basis of actually being con-
of formal logic, of which the rules of mathemat-
firmed by additional case studies that test it
ics are but a subset. Finally, replicable observa-
against the empirical circumstances of those tions constitute a means to the replication of a
other settings. theory's confirmation or disconfirmations - an

MIS Quarterly/March 1989 41

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

objective MIS case studies are able to achieve isfy the four requirements better than other case
by testing the same theory through new predic- studies. For instance, first, consider the require-
tions, thereby calling for new observations rather ment that the theory of interest be confirmed
than replications of old ones. through empirical testing. Confirming the theory
by successfully testing it through just one of its
The article has also taken the position that sci-
predictions would not be as rigorous as success-
entific methodology does not involve generalizabil-
fully testing it through several of its predictions.
ity based on the result of a single test, whether
Likewise, confirming the theory in just one or-
it is a single test taking place in an MIS case
ganizational setting would not be as rigorous as
study or a single test taking place in a labora-
confirming it in two or more organizational set-
tory experiment of the natural sciences. Instead,
tings. As the number of explicitly derived predic-
generalizability is a product of successive test-
tions or the number of organizational settings
ing across a range of settings, not a single test
is increased, the theory's degree of confirma-
in a single setting.
tion may also be increased.
It is worth emphasizing how the particular sci-
entific framework described in this article allows Second, the requirement that the theory of in-
terest be more predictive than any rival theory
us to identify some case studies as having more
may, of course, be satisfied more rigorously by
analytical rigor than others. There are two ways
increasing the number of rival theories against
in which analytical rigor may be assessed.
which its predictive performance is compared.
First, there is simply the matter of whether a As the number of rival theories considered is
given case study explicitly addresses each of increased, the theory's degree of relative pre-
the four requirements. As a check for falsifiabil- dictive power may also be increased.4
ity, does the case study consider any predic-
Third, the requirement that the theory of interest
tions through which the theory of interest could
be logically consistent may be more rigorously
be proven wrong? As a check for logical consis-
satisfied by increasing the number of predictions
tency, are all the predictions considered consis-
derived from it, then making sure that none of
tent with one another? As a check for empirical
the predicted events contradicts or precludes
validity, does the case study confirm the theory
one another. In contrast, deriving just two pre-
through empirical testing? Finally, as a check
dictions from the theory, and checking that they
for relative predictive power, does the case study
do not contradict each other, would provide mini-
rule out rival theories? These questions pre-
mal rigor. Thus, as the number of explicitly de-
sume, of course, that the theory of interest is
rived predictions is increased, the theory's
stated explicitly in the first place and that predic-
degree of logical consistency may also be
tions following from the theory are also explicitly increased.
stated. MIS case studies that satisfy all four re-
quirements explicitly and successfully are more Finally, the requirement that the theory of inter-
rigorous than MIS case studies that satisfy any est be falsifiable may be more rigorously satis-
of the four requirements implicitly or fied by increasing the number of predictions de-
unsuccessfully. rived from it and through which the theory could
be proven wrong. In other words, as the number
Table 1 compares a number of MIS case stud- of explicitly derived predictions is increased, the
ies with respect to the four requirements. Of theory's degree of falsifiability may also be
course, none of the cited case studies was con- increased.
ducted with the avowed purpose of fitting this
Campbell (1975), too, has referred to the situ-
article's scientific framework, so any value judg-
ation of the case researcher who pursues ana-
ments pertaining to the quality of these studies
lytical rigor in the ways just suggested. "In some
would be inappropriate. Instead, the table is in-
sense, he has tested the theory with degrees
tended simply as a guide for the reader who
of freedom coming from the multiple implications
wishes to pursue additional examples of MIS
case studies that illustrate, to varying extents,
of any one theory" (p. 182). In other words,
the scientific methodology this article has 4 With respect to the research strategy of using a
described. single case study to distinguish among competing
theoretical explanations, Allison (1971) established
The second matter pertaining to analytical rigor the primary scholarly model used by Markus (1983),
is one of degree. Some case studies may sat- Kling and lacono (1984), and others.

42 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

Table 1. Checking MIS Case Studies Against the Four Requirements

The Four Requirements
Does the case Does the case
study consider study confirm
any predictions Are all the the theory
through which predictions through Does the case
Case study Main theory the theory could consistent with empirical study rule out
author(s) of interest be proven wrong? one another? testing? rival theories?
Markus (1983) "Interaction" Yesa Yesb Yesc Yesd
theory

Kraemer, Theory of Yese Yes No' Nog

Dickhoven, successful model
Tierney and implementation
King (1987) in federal
agencies

Kling and "Organizational Yesh Possibly' Yesj Yesk

lacono (1984) politics
metaphor"

Laudon (1974) Theory of Yesm Yes Non Nog

resistance to
centralized
computing in
state and local
governments

Kling (1978) Theory of interplay YesP Possibly' Yesq Yes'

between technical
features and
social setting

Kling and "Web models" Yess Yes Yes' No"

Scacchi (1982)

Leonard- Theory of factors Yesv Possibly' Partiallyw No9

Barton (1987) influencing user
acceptance

Fulk and Theory of Not Not No' Nox

Dutton (1984) organizational applicablex applicable"
uses of video-
teleconferencing
See explanation for responses in "Notes for Table 1 and Table 2," following Table 2.

Campbell is extending the concept of "degrees which a given theory is tested. As the degrees
of freedom are increased in this category, the
of freedom" beyond its traditional statistical mean-
ing. There are three ways Campbell's extended theory's degree of confirmation can be corre-
notion of "degrees of freedom" can be applied spondingly increased. Increasing the degrees of
to describe the analytical rigor of a case study. freedom in this second category therefore allows
the case study to strengthen the extent to which
First, there are the degrees of freedom in the it satisfies one of the four requirements.6
number of predictions the case study considers.
As the degrees of freedom are increased in this
Since increasing the number of predictions also in-
category, the theory's degree of falsifiability, creases the number of different ways in which the
degree of logical consistency, and degree of con- case study's finding (e.g., that the theory of interest
firmation can all be correspondingly increased. is confirmed) could be subsequently replicated, the
Increasing the degrees of freedom in this cate- degree of replicability can also be increased.
gory therefore allows the case study to
strengthen the extent to which it satisfies three 6 Increasing the number of cases or organizational set-
of the four requirements.5 tings also allows the degree of generalizability to be
correspondingly increased. It should be noted that
Second, there are the degrees of freedom in the increasing the number of cases or organizational set-
number of cases or organizational settings in tings is not necessarily the same thing as increasing

MIS Quarterly/March 1989 43

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

Table 2. Checking the Degrees of Freedom of

MIS Case Studies in Three Categories
Three Categories of Degrees of Freedom
Number of cases Number of
Number of or organizational rival
Case study Main theory predictions settings theories
author(s) of interest considered considered considered
Markus (1983) "Interaction" theory Several 1 2d

Kraemer, Theory of 30e 2Y 0

Dickhoven, successful model
Tierney and implementation
King (1987) in federal
agencies

Kling and "Organizational Severalh 1 3k

lacono (1984) politics
metaphor"

Laudon (1974) Theory of 4m 42 0

resistance to
centralized
computing in
state and local
governments

Kling (1978) Theory of interplay SeveralP 1 1

between
technical
features and
social setting

Kling and "Web models" 5 2aa 1"

Scacchi (1982)

Leonard- Theory of factors Several" 1 0

Barton (1987) influencing user
acceptance

Fulk and Theory of O" 1 0

Dutton (1984) organizational
uses of video-
teleconferencing
See explanation for responses in "Notes for Table 1 and Table 2." following Table 2.

Third and last, there are the degrees of freedom Table 2 compares the same MIS case studies
in the number of rival theories against which the considered in Table 1 with respect to the three
theory of interest is compared. As the degrees categories of degrees of freedom. As the table
of freedom are increased in this category, the shows, the greater a case study's degrees of
theory's degree of relative predictive power can freedom in each category, the greater the case
be correspondingly increased. Increasing the de- study's analytical rigor. A particular case study's
grees of freedom in this third category therefore "analytical strategy" might therefore be de-
allows the case study to strengthen the extent scribed in terms of the number of degrees of
to which it satisfies one of the four requirements. freedom it pursues in each category.
The central concern of this article has been to
the number of data points in a statistical study. The
address certain methodological issues pertain-
latter is often done simply to increase the "level of
ing to MIS case studies. However, the article's
confidence" associated with a single, statistically in-
ferred observation (e.g., the observation that "the true analysis and conclusions may have ramifications
mean is different from zero"), whereas the former that go beyond matters of MIS case studies
involves increasing the total number of observations. alone. These ramifications might prove interest-
In other words, increasing the number of data points ing to scholars and practitioners alike.
will increase the "degrees of freedom" only in the
conventional statistical sense of this term, not in the For MIS scholars, the article's discussion of sci-
additional senses this article is explaining. entific method might prove interesting for its rele-

44 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

Notes for Table 1 and Table 2

a One prediction is explicitly stated (in Table II, p. 437): "Changing individuals and/or fixing technica
features will have little effect on resistance." Other predictions pertaining to the interaction theor
are considered implicitly in the discussion (pp. 437-438).

b This refers to predictions that the case study treats explicity as well as implicity.

c See p. 438.

d The case study rules out the "people-determined" theory and the "system-determined" theory
438).

e The case study states 30 predictions (these are the "propositions" on pp. 256-287). Each prediction
makes it possible to refute the theorized impact of a specific variable. (See Figure A.1, p. 257, for a
list of the 30 variables.)

'The case study's objective is theory formulation, not theory testing. Still, the theory and predictions
are consistent with the facts of the two cases considered (the "TRIM/MATH" computer model and
the "DRI" computer model).

9 Only one theory is formulated and considered.

h All predictions treated are implicit. For example, the case study states (p. 1225): "The CBIS did not
simply evolve along a natural path nor did it drift, rather it was pushed in a specific direction which
would increase the power and control of key actors within the organization." Thus the reader may
infer the prediction that, if the organizational politics metaphor is true, then we should observe
neither evolution along a natural path, nor drift, but observe development in a direction that would
increase (not decrease or keep constant) the power and control of key actors within the
organization.

' The case study provides sufficient material for the reader to infer predictions (as the prediction in
note h was inferred) that may then be compared.
i For example, the prediction mentioned in note h is confirmed.

The case study rules out the technological evolution metaphor, the economic rationality metaphor,
and the organizational drift metaphor (pp. 1222-1223). The reader may infer additional predictions,
pertaining to these three theories, that are implicit in this portion of the case study.

m The theory contains the variables of "homogeneity," "interdependence," and "internal integration"
(pp. 67-75). The case study states four predictions explicity: (1) "We hypothesize that ceteris
paribus, the more organizations are homogeneous with respect to tasks- the production of similar
products or services - the more likely they share similar environmental and internal problems, the
more likely it is that they will interact with each other in dealing with shared problems, and the more
likely they are to pursue collective solutions to those problems [such as sharing and using a
centralized, computerized database system]" (p. 69); (2) "Here we hypothesize that high and
increasing levels of interdependence among social units are conducive to higher levels of social
integration among those units, and supportive of efforts attempting to increase integration [such as
sharing and using a centralized, computerized database system]" (p. 71); (3) "Therefore, we
hypothesize a tradeoff between homogeneity and interdependence in relation to integration of a
social system. If both qualitites are high in a system, increases in integration would be supported. If
both are low, further integration would be most difficult. If of opposite sign, one low and the other
high, the effects should tend to cancel out" (p. 72); and (4) "For these reasons we hypothesize that
under conditions of high internal unit integration, resistance to system integrating efforts will be very
high, and/or the terms under which such units are included into larger systems will be very favorab-
le... Furthermore, we suggest that if resistance remains high, and if the demands of highly inte-
grated units are very high, the integrating effort will cease or force will be resorted to" (p. 73).

MIS Quarterly/March 1989 45

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

Notes for Table 1 and Table 2 - continued

" The study qualifies itself by taking the position that the theory "is not itself proved by the [four
studies [but] is intended to serve as a guide to the cases" (p. 91). However, this qualification m
read as a sign of modesty, since the study (in Table 3, p. 75) offers what appears to be the favor
results of empirically testing the predictions mentioned in note m. Moreover, the study eve
scribes specific sets of empirical conditions pertaining to the bureaucratic reform process
states the level of resistance predicted for each. These sets of empirical conditions are prese
under the headings of the "pluralist model"; the "collegial model"; the "notables model"; and
"reputational elite model" (pp. 80-90).

P All predictions treated are implicit, but the case study provides sufficient discussion for the read
infer them. The theory of interest is that a computerized information system's impacts are a
product of its technical features and its social setting. One prediction the reader may infer is tha
the theory is true, then deficient technical features alone will not bring about a lack of impact
negative impact. Whereas this prediction is clearly refutable, it is confirmed by the facts th
study reports.

q See p. 492.
The case study rules out the theory that either the technical features alone, or the social setting
alone, can determine the impacts of a computerized information system.

s The case study states five predictions (these are the propositions on p. 26).

' See pp..55-63.

u The case study compares "web" models to "discrete-entity" models, but states (p. 70): "We have
not organized this article to test the relative explanatory power of the discrete-entity and web
models."

v One prediction is explicit (p. 10): "It seems reasonable to hypothesize that the first adopters of SSA
might be younger, more highly educated in the computer field, and more skilled in computer
languages than their colleagues who are not yet using, and may never use, SSA." Other predictions
treated are implicit.

w The prediction (mentioned in note v) was refuted (p. 14): "Age, type of education, and skill in Fortran
and PL1 showed no relationship to SSA use." Other predictions were confirmed.

x This case study avows (p. 106): "Our purpose was not to provide a controlled experimental com-
parison..., but rather to gather exploratory and descriptive data." This case study is included as a
reminder that case studies may also be legitimately used for the purposes of exploratory analysis
and theory generation, not just theory testing. With respect to theory generation, this particular case
study may be regarded as a useful contribution toward the development of a theory of the organiza-
tional uses of video-teleconferencing.

y See pp. 21-22.

2 See pp. 12-14.

aa See pp. 40-53.

46 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

vance to MIS research in general, not just MIS tween quantitative and qualitative approaches
case studies. Research methodology in the have, unfortunately, become institutionalized into
study of management information systems has opposing camps.9 Some of the methodological
been gaining attention as a problem, in itself, concepts in this article may prove helpful in avoid-
that deserves investigation.7 In this regard, the ing a similar fate in the academic field of MIS.
article's view of scientific method could help
secure the emerging position of qualitative re- For MIS practitioners, the article's discussion of
search in MIS8 and perhaps, at the same time, scientific method might prove interesting for de-
reconcile the perceived differences between quan- mystifying the aura of MIS research that claims
titative and qualitative approaches in MIS re- to pursue scientific rigor, whether it involves the
search. In this larger methodological context, it qualitative study of a single case or the utiliza-
is interesting to observe that this article's schema tion of a sophisticated statistical tool such as
LISREL. The discussion of scientific method -
for assessing analytical rigor (illustrated in Tables
1 and 2) recognizes no differences between quan- especially the four requirements that a scientific
titative and qualitative approaches. (Indeed, the theory must satisfy - may empower MIS practi-
tioners themselves to identify the point at which
case study by Leonard-Barton (1987) involves
statistical inference.) The degrees of freedom in scientific rigor is achieved in an MIS research
each category (which follow from the four re- effort, and beyond which further rigor, especially
quirements of the natural science model) can if pursued at the expense of professional rele-
be greater or smaller, whether the theory of in-
vance, can be called into question.
terest is stated in the form of mathematical propo- Finally, it is important to point out that, in the
sitions or verbal propositions. In other words, a actual formulation of scientific knowledge, nei-
qualitative case study can possess more ana- ther natural scientists nor social scientists nec-
lytical rigor than a statistical study using LISREL, essarily think in terms of the formalized proce-
just as the reverse may be true. In this sense, dures of any model of science, including the
any distinctions between quantitative and quali- natural science model. Lee (1987a) states:
tative approaches are artificial and inconsequen- "These procedures do not address the private,
tial. Neither type of research is inherently more mental process by which a scientist formulates
rigorous than the other. In other fields of aca- scientific knowledge, but rather the public proc-
demic research, the perceived differences be- ess by which the scientist will on occasion retro-
spectively test the truth of the already formu-
7 In the spirit typical of methodological inquiry, MIS aca- lated knowledge for acceptance by his or her
demics are making their own research methodology
9 In the field of organizational studies, the existence
their object of study. (For example, see Ein-Dor
of opposing camps is clearly evident. Morey and Lu-
(1986); Jenkins (1986); Kauber (1986); Klein (1986);
thans (1984), in their review of the organizational lit-
Naumann (1986); Wand and Weber (1986). These
erature, describe the opposition between the two
papers were presented at the Management Informa-
camps as objective versus subjective (Burrell and
tion Systems Researcher's Workshop, held at the No-
Morgan, 1979), nomothetic versus idiographic (Lu-
vember 1986 annual meeting of the Decision Sci-
thans and Davis, 1982), quantitative versus qualita-
ences Institute.) This methodological inquiry
tive (Van Maanen, 1975), outsider versus insider
addresses not only what it means for our research
(Evered and Louis, 1981), and etic versus emic. This
to be "scientific," but also such matters as the role
has generated concern over what Morey and Luthans
of frameworks, epistemology, and paradigms in MIS
research. call the "widening gap between the two major orien-
tations to organizational research" (p. 29) - a gap
so wide that some authors have called for a rap-
8 For a provocative view of the qualitative methods that prochement between the two approaches (Evered
are emerging in MIS research see Bjorn-Andersen and Louis, 1981; Luthans and Davis, 1982; Morey
(1986); Goldstein (1986); Markus (1986); Rosen and Luthans, 1984).
(1986). These papers were presented at the panel
In the field of operations research, the interdiscipli-
on the "Use of Qualitative Methods in MIS Re-
nary quantitative/qualitative approach, which charac-
search," held at the December 1986 annual meeting
terized the field shortly after its founding during World
of the International Conference on Information
War II, eventually gave way to the dominance of quan-
Systems.
titative approaches. Only recently has the non-
For an introduction to the use of qualitative methods mathematical camp re-emerged to challenge, or com-
in general, see Filstead (1970); Kirk and Miller (1986); plement, the mathematical camp. See Ackoff (1979)
Yin (1984), as well as the December 1979 issue of for a historical review of the events in the develop-
Administrative Science Quarterly. ment of the field of operations research.

MIS Quarterly/March 1989 47

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

peers" (p. 577). Kaplan (1964) refers to the Datta, L. "The Politics of Qualitative Methods,"
former process as the actual "logic in use" by American Behavioral Scientist (26:1), Septem-
a scientist, and the latter as a "reconstructed ber-October 1982, pp. 133-144.
logic" (p. 8). In recognizing the natural science Dukes, W. "N = 1," Psychological Bulletin
model as one among many possible recon- (64:1), 1965, pp. 74-79.
structed logics of science, the article also recog- Ein-Dor, P. "An Epistemological Approach to the
nizes the need for future research to investigate Theory of Information Systems," Proceedings
the ramifications that alternative models of sci- of the 1986 Annual Meeting of the Decision
ence could have for MIS case-study Sciences Institute (1), Honolulu, HI, Novem-
methodology. ber 23-25, 1986, pp. 563-565.
Evered, R. and Louis, M. "Alternative Perspec-
tives in the Organizational Sciences: 'Inquiry
from the Inside' and 'Inquiry from the Out-
References side'," Academy of Management Review
(6:3), July 1981, pp. 385-395.
Ackoff, R. "The Future of Operational Research Filstead, W. Qualitative Methodology, Markham,
Is Past," Journal of the Operational Research Chicago, IL, 1970.
Society (30:2), February 1979, pp. 93-104. Fulk, J. and Dutton, W. "Videoconferencing as
Allison, G. Essence of Decision: Explaining the an Organizational Information System: Assess-
Cuban Missile Crisis, Little, Brown, & Co., ing the Role of Electronic Meetings," Systems,
Boston, MA, 1971. Objectives, Solutions (4:2), April 1984, pp. 105-
Behling, O. "The Case for the Natural Science 118.
Model for Research in Organizational Behav-Garfinkel, H. Studies in Ethnomethodology, Pren-
ior and Organization Theory," Academy of Man- tice-Hall, Englewood Cliffs, NJ, 1967.
agement Review (5:4), October 1980, pp. 483-Geertz, C. The Interpretation of Cultures, Basic
490. Books, New York, NY, 1973.
Benbasat, I., Goldstein, D. and Mead, M. "The George, A. and McKeown, T. "Case Studies and
Case Research Strategy in Studies of Infor- Theories of Organizational Decision Making,"
mation Systems," MIS Quarterly (11:3), Sep- in Advances in Information Processing in Or-
tember 1987, pp. 369-386. ganizations (2), L. Sproull and P. Larkey
Bernstein, R. The Restructuring of Social and (eds.), JAI Press, Greenwich, CT, 1985, pp.
Political Theory, University of Pennsylvania 21-58.
Press, Philadelphia, PA, 1978. Goldstein, D. "The Use of Qualitative Methods
Bernstein, R. Beyond Objectivism and Rela- in MIS Research," abstracted in Proceedings
tivism: Science, Hermeneutics, and Praxis, Uni- of the Seventh International Conference on
versity of Pennsylvania Press, Philadelphia, Information Systems, San Diego, CA, Decem-
PA, 1983. ber 15-17, 1986, p. 338.
Bjorn-Andersen, N. "Action Research on the Herriot, R. "Tension in Research Design and Im-
Impact of IS on Clerical Workers," abstracted plementation: The Rural Experimental Schools
in Proceedings of the Seventh International Study," American Behavioral Scientist (26:1),
Conference on Information Systems, San September-October 1982, pp. 23-44.
Diego, CA, December 15-17, 1986, p. 338. Hersen, M. and Barlow, D. Single Case Experi-
Burrell, G. and Morgan, G. Sociological Para- mental Designs, Pergamon Press, New York,
digms and Organisational Analysis, Heine- NY, 1976.
mann, London, 1979. Huberman, A. and Crandall, D. "Fitting Words
Campbell, D. "'Degrees of Freedom' and the to Numbers: Multisite/Multimethod Research
Case Study," Comparative Political Studies in Educational Dissemination," American Be-
(8:2), July 1975, pp. 178-193. havioral Scientist (26:1), September-October
Campbell, D. and Stanley, J. Experimental and 1982, pp. 62-83.
Quasi-Experimental Designs for Research, Jenkins, A.M. "Management Information Sys-
Houghton Mifflin, Boston, MA, 1963. tems Researcher's Workshop," Proceedings
Copi, I. Introduction to Logic, Macmillan, New of the 1986 Annual Meeting of the Decision
York, NY, 1986. Sciences Institute (1), Honolulu, HI, Novem-
Daft, R. "Learning the Craft of Organizational ber 23-25, 1986, pp. 559-562.
Research," Academy of Management Review Kaplan, A. The Conduct of Inquiry, Chandler,
(8:4), October 1983, pp. 539-546. New York, NY, 1964.

48 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
 Case Study Methodology

Kauber, P. "What's Wrong with a Science of Lee, A. "Case Studies as Natural Experiments,"
MIS?" Proceedings of the 1986 Annual Meet- Human Relations, forthcoming.
ing of the Decision Sciences Institute (1), Leonard-Barton, D. "Implementing Structured Soft-
Honolulu, HI, November 23-25, 1986, pp. 572- ware Methodologies: A Case of Innovation in
574.
Process Technology," Interfaces (17:3), May-
Kirk, J. and Miller, M. Reliability and Validity in June 1987, pp. 6-17.
Qualitative Research, Sage University Paper Louis, K. "Multisite/Multimethod Studies: An In-
Series on Qualitative Research Methods (1), troduction," American Behavioral Scientist
Sage Publications, Beverly Hills, CA, 1986. (26:1), September-October 1982, pp. 6-22.
Klein, H.K. "The Critical Social Theory Perspec- Luthans, F. and Davis, T. "An Idiographic Ap-
tive on Information Systems Development," Pro- proach to Organizational Behavior Research:
ceedings of the 1986 Annual Meeting of the The Use of Single Case Experimental Designs
Decision Sciences Institute (1), Honolulu, HI, and Direct Measures," Academy of Manage-
November 23-25, 1986, pp. 575-577. ment Review (7:3), July 1982, pp. 380-391.
Kling, R. "Automated Welfare Client-Tracking Markus, M.L "Power, Politics, and MIS Implemen-
and Service Integration: The Political Econ- tation," Communications of the ACM (26:6),
omy of Computing," Communications of the June 1983, pp. 430-444.
ACM (21:6), June 1978, pp. 484-493. Markus, M.L. "Case Study Research and the
Kling, R. and lacono, S. "The Control of Infor- Use of Communication Systems in Organiza-
mation Systems Developments After Implemen- tions," abstracted in Proceedings of the Sev-
tation," Communications of the ACM (27:12), enth International Conference on Information
December 1984, pp. 1218-1226. Systems, San Diego, CA, December 15-17,
Kling, R. and Scacchi, W. "The Web of Comput- 1986, p. 339.
ing: Computer Technology as Social Organi- Miles, M. "Qualitative Analysis as an Attractive
zation," in Advances in Computers (21), M. Nuisance: The Problem of Analysis," Admin-
Yovits (ed.), Academic Press, New York, NY, istrative Science Quarterly (24:4), December
1982, pp. 1-90. 1979, pp. 590-601.
Kraemer, K., Dickhoven, S., Tierney, S. and Miles, M. "A Mini-Cross-Site Analysis," Ameri-
King, J. Datawars: The Politics of Modeling can Behavioral Scientist (26:1), September-
in Federal Policymaking, Columbia University October 1982, pp. 121-131.
Press, New York, NY, 1987. Morey, N. and Luthans, F. "An Emic Perspec-
Kuhn, T. "Reflections on My Critics," in Criticism tive and Ethnoscience Methods for Organiza-
and the Growth of Knowledge, I. Lakatos and tional Research," Academy of Management
A. Musgrave (eds.), Cambridge University Review (9:1), January 1984, pp. 27-36.
Press, New York, NY, 1970, pp. 231-278. Nagel, E. The Structure of Science, Hacket, In-
Laudon, K. Computers and Bureaucratic dianapolis, IN, 1979.
Reform, Wiley, New York, NY, 1974. Naumann, J.D. "The Role of Frameworks in MIS
Lee, A. "The Scientific Basis for Conducting Research," Proceedings of the 1986 Annual
Case Studies of Organizations," Academy of Meeting of the Decision Sciences Institute (1),
Management Proceedings, R. Robinson and Honolulu, HI, November 23-25, 1986, pp. 569-
J. Pearce (eds.), San Diego, CA, August 11- 571.
14, 1985, pp. 320-323. Popper, K. The Logic of Scientific Discovery,
Lee, A. "The Case Study of an Organization as Harper Torchbooks, New York, NY, 1968.
a Scientific Research Strategy," Working Rosen, M. "Ethnographic Study of the Role of
Paper 86-25, College of Business Administra- Information Technology in Organizational Con-
tion, Northeastern University, Boston, MA, trol Systems," abstracted in Proceedings of
1986. the Seventh International Conference on In-
Lee, A. "Quixotic Communication: The Case of formation Systems, San Diego, CA, Decem-
Expert Witness Testimony," Knowledge (8:4), ber 15-17, 1986, p. 339.
June 1987a, pp. 549-585. Sanday, P. "The Ethnographic Paradigms," Ad-
Lee, A. "Integrating Positivist and Interpretive Ap- ministrative Science Quarterly (24:4), Decem-
proaches to Organizational Research," pres- ber 1979, pp. 527-538.
entation at the Annual Meeting of the South- Schon, D. The Reflective Practitioner: How Pro-
ern Management Association, New Orleans, fessionals Think in Action, Basic Books, New
LA, November 4-7, 1987b. York, NY, 1983.

MIS Quarterly/March 1989 49

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms
Case Study Methodology

Schon, D., Drake W. and Miller, R. "Social Ex- Yin, R. "The Case Study as a Serious Research
perimentation as Reflection-in-Action," Knowl- Strategy," Knowledge (3:1), September
edge (6:1), September 1984, pp. 5-36. 1981b, pp. 97-114.
Susman, G. and Evered, R. "As Assessment of Yin, R. The Case Study Strategy: An Annotated
the Scientific Merits of Action Research," Ad- Bibliography, the Case Study Institute, Wash-
ministrative Science Quarterly (23:4), Decem- ington, D.C., 1982a.
ber 1978, pp. 582-602. Yin, R. "Studying Phenomenon and Context
Taylor, C. "Interpretation and the Sciences of Across Sites," American Behavioral Scientist
Man," in Interpretive Social Science, P. Rabi- (26:1), September-October 1982b, pp. 84-
now and W. Sullivan (eds.), University of Cali- 100.
fornia Press, Berkeley, CA, 1979, pp. 25-71. Yin, R. Case Study Research: Design and Meth-
Tice, T. and Slavens, T. Research Guide to Phi- ods, Sage Publications, Beverly Hills, CA,
losophy, American Library Association, Chi- 1984.
cago, IL, 1983.
Van Maanen, J. "The Fact of Fiction in Organ-
izational Ethnography," Administrative Sci-
ence Quarterly (24:4), December 1979, pp.
539-550. About the Author
Wand, Y. and Weber, R. "On Paradigms in the Allen S. Lee holds a Ph.D. in social science
IS Discipline: The Problem of the Problem," and policy analysis from M.I.T. In addition to
Proceedings of the 1986 Annual Meeting of case-study methodology, he is conducting re-
the Decision Sciences Institute (1), Honolulu, search on the implementation of management
HI, November 23-25, 1986, pp. 566-568. information systems, particularly human resource
Yin, R. "The Case Study Crisis: Some Answers," information systems. He is currently assistant pro-
Administrative Science Quarterly (26:1), fessor of management science at Northeastern
March 1981a, pp. 58-65. University.

50 MIS Quarterly/March 1989

This content downloaded from

102.115.189.225 on Tue, 15 Sep 2020 17:55:03 UTC
All use subject to https://about.jstor.org/terms