approach to augment the current investigations of behavioral IS researchers. We outline the features of such an additional paradigm, interpretivism, and indicate ...
^KCHff,
V
(UBEiOUSSJ tA
o/ -re
,
ALFRED
P.
WORKING PAPER SLOAN SCHOOL OF MANAGEMENT
IS Research Paradigms: Method versus Substance
Wanda
J.
Jack
J.
Orlikowski Baroudi
Sloan Working Paper # 3028-89-MS
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE, MASSACHUSETTS 02139
IS Research Paradigms:
Method versus Substance Wanda
J.
Jack
J.
Orlikowski Baroudi
Sloan Working Paper # 3028-89-MS
.
SEP
rr.
1
989 2 1989 [
ill
DECEIVED
IS
Research Paradigms:
Method versus Substance
Wanda
J.
Orlikowski
School of Management (E53-329)
Massachusetts
Institute of
Technology
Cambridge
MA
02139
(617)253-0443
Jack
J.
Baroudi
Stern School of Business
New
York University
100
New
Trinity
York,
Place
NY 10006
(212)998-4206
January 1989
(An earlier draft of
August 1988.)
this
paper was presented
at
the
Academy
of
Management Meeting
in
Anaheim CA,
IS
Research Paradigms:
Method versus Substance
Abstract
We examined 155 behavioral IS research articles published from 1983-1988 and found that while this research is not rooted in a single overarching substantive paradigm does exhibit a single methodological paradigm, that of positivist methodology. We argue in this paper that this metatheoretic perspective, stemming as does from the natural science tradition, is not always appropriate for investigating the relationship between information technology and human activity. In particular we urge the recognition of it
it
social process as inherent
in
any inquiry
into the
development and use
of
information systems within
organizations.
such an emphasis on social process, the determination of meaning in social life becomes We argue that positivist research methods are not well-equipped to deal with situated meaning formation over time, and propose an additional research approach to augment the current investigations of behavioral IS researchers. We outline the features of such an additional paradigm, interpretivism, and indicate a number o' research methodologies that subscribe to its tenets. We then suggest how such methoaologies can use'„ily be used :n the IS community to enrich its research efforts In
the
light of
important, as do the dimensions of time and context.
1.
INTRODUCTION '
One
disciplines
of the
most pronounced features
contemporary social research
concerned with human phenomena
anthropology and
their applied
industrial sociology
[Astley
of
-
is
and
collective)
and by
we mean
this
those
such as psychology, sociology,
science, education, industrial psychology and
of administrative
the great range of research perspectives or paradigms that operate concurrently
& Van de Ven 1983;
marked by a plethora
fields
(individual
-
Burell
& Morgan 1979; Morgan 1980;
of "schools of thought"
each with
its
own
Pfeffer 1982].
These
disciplines are
metatheoretic assumptions, research
methodologies and adherents. Given the complex and indeterminate nature of the social phenomena studied within these fields, the existence of a plurality of perspectives allows the exploration of diverse
questions and hence adds breadth as well as depth
to the
knowledge generated.
(Information Systems) research however, such a diversity of research schools
In this
to
be an
paper
implicit
investigated and
evidence
for
we suggest
that while
agreement among
no
clear, collective theory
collective methodological tradition,
studying
phenomena can be
tradition
which can supplement and ennch
binds the IS discipline, there appears
In
section two of this paper
we
to
IS research
we propose and
endeavors, and
in
be
provide
and why we believe such a monolithic approach
sections three and four
limiting. In
not evident.
about the underlying nature of the phenomena
IS researchers
what constitute appropriate research methods.
such a
is
behavioral IS
In
to
explore a methodological
section five
we
outline
some
research themes that could benefit from such methodological diversity.
2.
SUBSTANTIVE VERSUS METHODOLOGICAL PARADIGMS Much
recent self-reflection of the IS discipline has involved a discussion of paradigms^ and the
status of IS research vIs-a-vis the
norms
1982; Culnan 1986, 1987; Culnan &
1982;
Weber
1984].
In
the
all
of
what constitutes a
Swanson 1986; Hamilton &
can
inform
and
unite
Ives 1982;
above studies the focus has been on
substantive ^ research paradigms or sub-paradigms.
paradigm
scientific discipline
a
discipline,
We
that
argue here that there is,
through
[Benbasat 1985;
Keen 1980; identifying is
Klein
and
another sense
methodology.
& Weike
articulating in
which a
The concept
methodological paradigm draws on Kuhn's [1970] notion that the method of puzzle solution important to a scientific
'The significance
field
paradigms
as
is
the substance of the puzzle
itself [Blair
Blair
of is
a
as
1982].
influenced mucfi work in tfie fiistory and pfiilosophy of science |Bernstein Ryan 1970, 1973] One definition given by Kuhn [1970 10] is that a paradigm includes 'law, ' and instrumentation together [providing] models from which spnng coherent traditions of scientific research of
for scientific activity fias
1978, 1985, Hacking 1981, Kufin 1970, theory, application,
,
^We wish to make a clear distinction here between substantive and methodological paradigms The former refer to content in the sense of knowledge about phenomena, such as theones about systems implementation or human-computer interaction The sense of methodology refers to form, the strategies and techniques employed to pursue knowledge about phenomena, such as those underlying the execution ol case studies or protocol analysis.
2.1.
Evidence
of a
Methodological Paradigm
a recent assessment of research
in
into
and Organizational Psychology, Webster &
Industrial
much
Starbuck [1988] conclude that I/O psychologists, while not having achieved attained "shared beliefs, values,
and techniques
...
[they
paradigm rather than a substantive one" [1988:111]. Evidence
single substantive paradigm. citation
MIS research
analysis of
disparate) research areas
m
for
have what] would seem
Similarly,
we
is still
sciences,
As
pre-paradigmatic,
is
in
be a methodological
the IS
field
Cuinans [1986]
in
there
no
is
bibliographic
1982), which established nine distinct (and
publications (1972 to
the IS community. Further, Culnan [1987], again employing bibliographic
it
has made progress,
MIS
research, suggesting "that while
one accepts the argument
if
we wish
psychology
I/O
to
argue that while there
that MIS, like
all
social
of the activity constituting the field of behavioral IS research, ^
the behavioral IS literature published from January
sources were: Communications of the
See
figure
are
concerned
one
consideration.
for the distribution of
with
A
research
total of
Insert figure
The research
1
,
excluded
Management Science and MIS ,
articles
any conceptual
were included m
or
framework
bears
we examined
four major IS outlets.
These
Quarterly
articles
.
As we from
this analysis.
different dimensions.
The
first
used Cuinan's
These categories include research foundations, macro approaches
MIS, MIS management, and MIS curriculum.
by Cuinan's categories in that,
as
different
to
Cuinan's foundation and
m
is
presented
in
This data confirms Cuinan's
figure two.
her studies, there does not appear to be one dominant substantive
and
distinct
streams
of research.
two here
To explore the extent
limit
we
were categorized along several
to
paradigm but rather several
"'We specifically
ICIS Proceedings
May 1988 m
assertion
a
were discarded as we were only concerned with empirical research. The
[1986, 1987] conclusions
multiple methodological
to
this
this
is
one here
MIS, micro approaches
Insert figure
,
1983
To explore
and
behavioral IS research published across these four sources.
conduct
[1987] five research categories.
distribution of articles
ACM
55 empirical research
articles
categories
no substantive paradigm, there
is
the form of a methodological rather than a substantive paradigm.
curriculum
m
to
a multiple paradigm discipline" [1987:347].
paradigm underlying much
—
suggest that
such a lack can be found
citation analysis, identified five intellectual subfields within current
MIS
theory consensus, have
our remarks approaches
to
which a methodological paradigm exists
to IS
behavioral research, as
we
believe this
is
in
behavioral IS research,
the sub-discipline that can
most
we
benefit from
Journal
CommunJcatlons
Frequency
of Articles
56
Percent
analysed the sample
and
three different ways.
by time frame
finally
of the study,
Figure three presents the
section 2.2.
designs.
in
The
A discussion first
by research design, the second by epsitemology
first is
consequences
of the
breakdown showing the frequency
The three primary research designs which emerged from
lab experiments (27.1%)
were
clearly the
of the various research
this analysis are
These three designs account
and surveys (49.1%).
studies. Surveys, however,
of our findings is deferred to
dominant research method
m
this
case studies (13.5%),
for
90%
almost
of the
sample.
Insert figure three here
Articles
study
was
were then examined
classified as descriptive, interpretive or postivist. Additional categories
allowed the data
to
phenomena,
some
illustrating
These
generate the classification scheme.
describing the entire sample. the
underlying epistemology which guided the research. Each
for the
In
descriptive studies there
is
three,
were possible as we
however, proved adequate
no theoretical grounding or interpretation of
rather the studies are straightfonward "factual" accounts of events which are viewed as
issue of interest to the researchers.
under such conditions Descriptive
Little
theory building or theory testing
are studied with structured instrumentation.
Such studies serve
natural setting, deliberately not imposing any a priori understanding on of articles
by epsitemology. Postivism
of the studies, followed of the
possible
is
phenomena which
primarily to test theory.
studies explicitly adopt a nondetermimstic perspective, attempting to explore the
breakdown
is
were case studies and used simple frequencies and
articles typically
precentages. Positivist studies are premised on a pr/on fixed relationships within the
its
for
Interpretive
phenomena
of interest in
Figure four
it.
shows
the predominant epsitemology accounting for
by descriptive studies with 23.9%, and
lastly
by interpretive studies
at
the
72.9%
only
3.2%
sample.
Insert figure four here
The data was
finally
analyzed by the time period of the study. Four different categories were
sufficient to classify the data. Studies could
be one shot cross-sectional, cross-sectional over multiple
time periods, longitudinal, or involve process tracing such as protocol analysis. classified separately as they
do not neatly
collection, but are not truly longitudinal
fit
as the trace
such as a meeting or problem-solving exercise
other categories.
into the
-
is
typically
lasting
Protocol analyses were
They employ continuous data
conducted over a single discrete event
a short penod of time such as a few hours.
-
We
distinguished between multiple time period cross-sectional studies and longitudinal ones; the former
employ some measure(s), administered the
phenomenon;
at
several time intervals, providing
the latter are continuous studies
many
where the researcher engages
discrete with the
snap shots
of
phenomenon
over an uninterrupted period of time, such as a few months or years, and typically focussing on issues of
Research Design
process. The breakdown of articles by time period
presented
is
sectional studies are clearly the predominant form of research
and
the articles in our sample. Longitudinal of the
sample
in
figure five. Static,
one shot cross-
These studies account
in IS.
for
90.3%
of
3.9% and 4.5%
multiple time period studies account for only
respectively.
Insert figure five here
The
figures collectively shovi/ that while no
research, there clearly directs
most
IS research
has adopted a
phenomena withm
2.2.
postivist epsitemology,
methodology they are also
phenomena
Consequences
(i.e.
primarily survey or latxjratory oriented investigating
that given the nature of additional
view reflects Galliers
appropriate
and complexity
not clear, however, that researchers realize
basic assumptions about the
studied
for investigating certain
(i.e.
the epistemology).
we suggest
classes of problems,
investigated by behavioral IS researchers, a
methodological paradigms might also serve
& Land's
some
accepting
implicitly
phenomena
of the
is
Paradigm
of a Methodological is
It
how these should be
the ontology) and
While a Single perspective
number
is
a single cross-section or slice of time.
that by adopting a
nature of IS
by Webster & Starbuck [1988]. The majority of behavioral
similar to that found
is
to inform
our research endeavors. This
[1987] as well as Weick's [1984] assessment of IS research activity,
which they encourage IS researchers
examine the
to
role of
a
priori
assumptions
m
their theoretical
might realize
-
prior beliefs or
or prefer.
technology impact that
woven
may
lie
all
into
What people
what they expect
they usually find them.
methods
is
What
"see"
to find.
they
issues of theoretical substance
when
they use various methods
Researchers
fail
to
see
is
Weick
at
a formative stage
IS
...
is
than people
largely a function of their
usually expect to see rational systems,
that additional
to
is
tightly
for
and
processes and variables affecting
briefly
examines 27 in
organizations.
research "especially
much more
different research
likely to
help
when IVIIS
He concludes the
issues
in
researchers see
more accurately" (1984:129].
Through of
MIS
contribute to investigating issues of technology use
by noting that a broader methodological approach
their subject
in
outside their rational combination."
technology and organizations are
much more
in
and
methodological choices. Weick writes [1984:129] "The question of the appropriate methodology studying technology impact
IS
The methodological paradigm which
a dominant methodological paradigm.
is
IS research
one substantive paradigm dominates behavioral
this
paper we wish
to
encourage a greater awareness and understanding
of the diversity
epistemological and methodological issues underlying social research, so that the practice of IS
researchers
may draw on
IS researchers
a rich array of methodological perspectives. As a maturing research community
would benefit by having
insight into "the appropriateness of different kinds of
knowledge
Frequenc y
Percent
140
90.3
Longitudinal
7
4.5
Cross -Sectional: multiple snapshots
6
3.9
Time Period
of Stud y
Cross -Sectional: single snapshot
_2_
Protocol Analysis
1.3
100%
155
by Time Period Figure 5
Articles Classified
of
Study
purposes
problematic
phenomenon under
and
study, the state of knowledge,
of inquiry to
their
own
limited to
among
a choice
wish
make
to
research strategies,
positivistic
based on assumptions other than the
We
and purpose"
skills, style
research approach depend on "the characteristics of the problem being researched is
the nature of the
fit
1981:386]. This echoes Benbasat's (1985:61] recommendation that the choice of a
& Louis
[Evered
....
mode
[so that they can] explicitly select a
for different
a similar
extend
this
While his suggestion
proposal to include strategies
ones.
traditional positivistic
argument
we
"
to that postulated
by Daft & Wiginton [1979]
management
for
research, that positivistic methodologies (which they refer to as "low variety" techniques) are not complex
enough
inherent complexity, ambiguity, and instability of organizational systems. Invoking
to reflect the
the principle of requisite variety, they encourage the use of "high variety" (interpretive) methodologies,
[1979:187]
noting
complex organizational behaviors are modelled as
"If
they
if
simple,
are
understood, deterministic systems, or even as stochastic systems, then the resulting models
be
We
insignificant.
propose
that
languages
of high variety are useful tools for
will
well
tend
to
developing models of
organizations because they have sufficient scope and richness of meaning to describe organizational
processes."
we
Similarly
argue that the methodology dominant
insufficient variety for the nature of the this
3.
nature
The body
IS
IS
research discipline has
investigated by behavioral IS researchers.
We
explore
of IS
RESEARCH
research that
employ
that
label
concerned with studying the interaction
is
micro and macro levels of analysis to
the
the following section.
in
BEHAVIORAL
continue
phenomena
in
commonly
is
of IS
and humans
at
and we
referred to as "behavioral IS research",
both shall
Behavioral IS research involves the exploration of numerous
here.
substantive issues, and reflects various levels of analysis. For example, behavioral IS studies explore
human
cognitive
factors,
and
communication
modeling,
co-ordination,
structures
decision-making,
decentralization,
and
ehcitation
project
of
representation
responsibility
management,
of
and
systems
knowledge,
standards, procedures, and strategy.
A number
of different
units
of
centralization
authority,
development,
implementation, job and task design, control of work and workers, changes
in
interpersonal
and
training
skills,
vs
productivity,
analysis are implicated
in
this
research, such as those involving individuals, work groups, organizations, institutions, organizational
networks (national and multi-national), occupations, and societies.
By and understanding
large, the
the
fundamental theme underlying these diverse areas
practical
implications,
technology (as information systems)
into
potentialities,
human
life.
and
Galliers
limitations
of interest of
is
a concern with
integrating
& Land [1987:900] express
this
information
when
they
note that "our
field of
study
...
is
concerned
witfi IS
they serve." Given such a domain of interest, significant •
and common
and
their relations with the organization
we can make
and the people
a few observations about
some
of
its
characteristics:''
Information technology is a medium, and shares with all media the inherent feature of nonIt is, as Turkle [1984:15] points out, a constructive as well as a projective medium. Utilising the evocative analogy of the computer as a Rorschach, Turkle [1984:320] shows neutrality.
how
information technology serves as "a cultural object which different people and groups of apprehend with very different descriptions and invest with very different
people can
attributes". IS thus •
assume
different
meanings
for different
is a human artifact. IS humans in predominantly
Information technology modified and used by
people [Kaplan 1983].
are designed, developed, implemented, social settings (administrations, schools,
businesses, laboratories). IS are not "naturally occurring phenomena"; they do not exist outside of deliberate, ongoing human action. •
The "human life" dimension of our research questions is in contemporary society - complex, dynamic and intrinsically social and political. Ours is overwhelmingly an "organization age" where much of human interaction occurs in and is mediated through the conditions, procedures and locales of modern institutions. An understanding of human behavior must reflect its essentially contextual features, that is, that occurs within social and political -
it
milieux, •
and
that these milieux are constantly changing.
As a consequence of such a characterization of technology and human affairs, it is clear that the relationship between IS and human behavior similarly does not exist in a vacuum, but is situated
The two elements in the relationship are not independent, in ways that are contingent on various temporal, and social factors. Hence to understand the integration of IS into human affairs we grapple with the intrinsically social and dynamic nature of the relationship, and this in
both time and space.
but rather mutually influence each other political
need
to
In the light of such a characterization, we suggest that simple, um-directional, cause-effect relationships do not adequately capture the complex nature of the phenomena studied by behavioral IS researchers, nor can they reflect the reciprocal interdependence posited among its dimensions.
requires an investigation of social process.
Much
of
relationship
the
existing
behavioral
between
Implementation studies
for
IS
research being conducted today technology
information
example,
and
individual
into organizations.
and
also increase users' job satisfaction.
we
Ivancevich et
al.
realities.
successfully introduce information
with
There
how we is
build
systems
that are efficient
a large and growing interest
in
Suchman
1983].
and
effective
and
et al.
1985;
Hiltz
&
and
stress.
Turoff
IS personnel researchers [Bartol 1983; Baroudi 1985;
1983; Weiss 1983] are concerned with understanding the processes which result
dissatisfaction, turnover,
that
computer-mediated support
communication, collaborative work, and group decision-making [Donnellon
1985; Sproull & Kiesler 1986;
organizational
Systems development researchers [Bostrom and Heinen 1977a, 1977b;
Mumford & Weir 1979] are concerned
of
concerned with the ongoing
& Henderson 1981; Franz & Robey 1984; Gmzberg 1981;
[Alavi
Lucas 1981; Markus 1983], are concerned with how, over time, technology
is
Other researchers have focused on the power
shifts
in
job
generated by
technology and technological dependence [Lucas 1984; Markus & Bjorn-Anderson 1987; Saunders &
*This charactenzation relies heavily on the view of the
vi^orld
explicated by Berger
& Luckmeinn
(1966).
Many
Scamell 1986]
employment
studies have been conducted into the effects of computerization on job
levels [see the review of studies by Atteweil
& Rule
And
1984],
skills
the "impacts school" of IS
research examines the implications (individual, organizational, and societal) of widespread use
[Bjorn-Andersen & Pederson 1980; Danziger et
& lacono 1984; Laudon
1982; Kling 1978; Kling
al.
Olson & Primps 1984; Turner 1984; Zuboft 1988]. These are only a sampling under investigation yet
all
share a
common
thread.
All
and
of topics that
of
IT
1974;
one can
find
are concerned with the social processes
surrounding the introduction, creation, use/disuse of information technology, as portrayed by Kling & Scacchi's [1982] metaphor of the ongoing "web of computing".
To
date, as evidenced by the analysis
orientation, a research tradition that
assumptions natural •
in
has
its
roots
IS research about the reality to
phenomena
[Lincoln
& Guba
in
most
IS research reflects a positivistic
the natural sciences.
be studied,
reflect the
As a consequence, underlying
precepts informing the study of
1985:36]:^
The phenomenon
of interest
description of any
chosen aspect
is
section 2.1,
in
single, tangible of the
and fragmentable, and there
is
a unique, best
phenomenon;
•
The researcher and the object of inquiry are independent, and there between observation reports and theory statements;
•
Nomothetic statements (law-like generalizations that are independent of time and context) are possible, which implies that scientific concepts are precise, having fixed and invariant
is
a sharp demarcation
meanings, •
There exist real, uni-directional cause-effect relationships and tested via hypothetic-deductive logic and analysis;
that are
capaoie
of
being identified
• Inquiry is value-free.
The
application of these premises to research on social process
review of the prior characterization of behavioral [1983:385], the
more we recognize
most commonly adopted
tradition,
1987],
not entirely compatible with
is
is
research
that information processing
social world, the less appropriate natural science
research
IS
in
is
will
is
somewhat
reveal.
problematic, as a
To paraphrase Morgan
a social practice that impacts on a
approaches become.
It
IS research studies [Gailiers
appears
that the positivisitic
& Land 1987;
what we know about the phenomena we are
Weill
investigating.
a dissonance between the elements of our substantive theories (and the assumptions they
the nature of IS certain
in
human
life),
The
development
of interest).
result
make about
and our methodological paradigm (whose epistemological premises
assumptions about the phenomena
affects not only the
& Olson
reflect
Such a dissonance m our research endeavours
of theory, but also the practice of IS work.
The
findings of IS research
^Recent work in the post-empincist philosophy of saence [Bernstein 1985] has begun to question the validity of these assumptions for the practice of natural science To the extent that the positivist dogma lose their currency among mainstream natural scientists we should begin to see a growing interest among social scientists for additional researdi perspectives of the sort discussed in this paper
filter
community and are used as prescriptions
into the practitioner
research thus
We
will
have far-reaching consequences.
suggest
exploring the
and reductionist
for action. Simplistic
utility
research
that
into
and dynamic phenomena^ can benefit from
social
intrinsically
paradigms beyond positivism, and attempt
of additional research
resonance between substantive theory and methodology.
particular
In
we
believe that
to obtain is
it
a better
imperative that
the research strategies adopted by IS researchers be compatible with the underlying ontological premises
Morgan & Smircich [1980] argue
of behavioral IS research.
that different ontological assumptions, "world
views", or conceptions of reality favor different methodological approaches to studying reality. As an illustration,
[Boland
1
consider implementation. Clearly a central aspect of
978; Ginzberg
1
981
],
and hence
makes sense
it
this
phenomenon
employ a methodology
to
processual nature
is its
that examines, rather
than suppresses, this essential component.
Social processes such as
implementation, diffusion of innovation, group decision making,
IS
computer-mediated communication, and the over time and as situated indicate "the
methods
study of the process
orthodox approaches
...
to social
Hence when process
predetermined categories that purport
bound by the
are suited to research strategies that
be what
to
designs permit the exploration
is
evolution within the research setting. For
it
Vitalari
many research
longitudinal study to investigate user involvement
data collection process allowed
validity
for the
that
all
evaluation, vary substantially. is
Svhich
we
of
Essentially
not one of degree (one
believe behavioral IS research
we need
of
not be
[1985:243] suggests that "[L]ongitudinal
issues confronting the Information Systems
is
field,
Franz & Robey [1984:1202] used a
systems design. They found
that including the time
facilitated alternative interpretations of
modes
scholarly
discrepancy between the paradigms
(1987:9011.
through a quantitative measure
phenomena. The
capture of respondents' interpretations of events as they occurred,
However, the nature, depth and range
paradigms
and
in
possibility of rationalization that often
is
is
about the social world." But
longitudinal designs offer great opportunity for scientific progress."
Our premise
& Day [1982:31]
time-dependent phenomenon such as learning, adaptation, and
of
dimension improved measurement
examine them
theory do not lend themselves well to the
studied
"real"
is
approach.
limitations of the orthodox
hence avoiding the
will
not abstracted from, social context. Instead as Boland
in,
of the itself
like,
is
of
occurs with retrospective data collection.
inquiry generate
such knowledge, as
we
understanding of social
well
posit that the difference
as the
criteria
reality.
employed
more than merely methodological, having deeper
is
its
between the various research
being more quantitative, say), but rather one of kind
concerned with This point
in
also stressed by Culnan
1
.
The
roots,
m
1987:347] and Galiiers & Land
10
Hence we
philosophy.
assumptions about social (epistemology);
in
fact,
we
two research approaches
find that the
reality (ontology)
discuss
in
paper
this
reflect different
and about what constitutes legitimate knowledge
of that reality
they evince the fundamentally different goals of individual researchers [Rosen
1986:9], and the assumptions they bring to their research activity.
the
In
rest of
this
paper
we
known as
interpretive
perspective (also
motivate
use within IS research.
its
interpretive
We
in
the
paradigm" or "social constructionism") and
"naturalistic
paradigm consistent with the nature will
of
many
of the
must be made clear
entirely compatible.
that
While
we
we do
research
effort.
do we see them as
in
Our recommendation
and self-consciousness
of interpretive
complex
social
in
phenomenon, and
and incomplete. Our available
and
to
The
is
is
different
knowledge gained through any one approach
intention here
is
to
their inherent
in all
increased reflectiveness
to the
purposes and
ways
of studying the
is
interest
same
necessarily partial, biased
urge a recognition of the plurality of diverse research paradigms
a blind and mechanistic adherence
solely relevant
phenomena,
we encourage
modes, as suited
generate understanding and discussion about
be guarded against necessarily,
the use of the different research
the
essentially interchangeable or
research should not to be understood as an intention
As Morgan [1983:368] notes, there are many
of the researchers.
to
their relative
appropriateness.
any one approach
in
What must
the uncritical belief that
research contexts.
following section briefly explores the underlying tenets of the interpretive perspective,
to highlight the
problematic features of the traditional positivist approach
that interpretivism, while not a
our research endeavors.
panacea, may assuage some
Section
five illustrates
research efforts of behavioral IS researchers. IS
behavioral IS
order not to dilute the powerful insights that each can bring to the
replace the existing orthodoxy with an opposing one. Rather
attempt
to the
this
not believe that the positivist and mterpretivist paradigms for
believe they can be used together to investigate
differences must be respected
social
phenomena examined by
be enriched through being appropriate
social research are mutually exclusive, but neither
to
one research paradigm, the
of
which the research results are applied.
It
It
methods
argue that not only are the metatheoretic assumptions of
researchers, but that our research results
world
tenets and
outline the
community.
We
how
to
science.
We
of the limitations currently frustrating
propose
some
the interpretive perspective can be useful
conclude with
some
an
in
general recommendations
of
in
the
for
the
11
4.
THE INTERPRETIVE RESEARCH PERSPECTIVE •
It
is
not our intention here to discuss the positivist research paradigm, nor to explicitly point out
and
particular strengths
Lincoln
limitations.
Numerous such discussions
& Guba 1985; Morgan 1983; Mumford
crucial issues in the
human science
et
al.
We
1985].
are available [Boland
do however wish
to
draw
its
& Day 1982;
attention to two
research process that are typically overlooked, yet which should
influence scholars' choice of appropriate research methodologies.
Firstly,
example,
methodologies
particular research
positivistic
and reinforce a particular view
reflect
of reality.
research techniques encourage deterministic explanations of phenomena,
in
For that
these explanations emerge from interactions between the researcher and his/her subjects, where the researcher, by definition, dominates the relationship.
researcher focuses on the
and control
validity
the search for causal relations the positivist
In
research procedures, and hence adopts a predefined
of the
and circumscribed stance towards the phenomenon being investigated. Such a posture to
the
and
discovery
reflection
indeterminate,
of
and
non-deterministic,
Laboratory subjects and survey respondents act and react mechanically
discover one-sided things
if
it
insists
phenomena
of interest
one-dimensionally causal, can the previous section, and as
we
-
utilize
The second issue
phenomenon
it
we have
strong reason
information technology and
such
positivist
human
can
be
argued
same
that
into the
of
elucidate, there
there
exists
is
no way
of
only get answers to
human
life
-
are determinate and
we have
is
no reason
to
indicated
suspect that
for the
of the
everyday human
human
human
phenomena
reality,
in
in
life.
this
is
Unlike the natural
independence between
an
made
nature of these phenomena. As Giddens [1987:19] notes, there
You
life.
assertion cannot be
discourse
...
techniques with confidence. As
do not impinge on and change the nature
science do enter
refer to actors' intentions
suspect that the relationships
to
pertains to the role played by social research
of study, the
natural science
if
Markus & Robey [1988]
the case with information technology and
sciences where
relationships.
As Rowan [1973:210] notes: "Research can only
on setting up one-sided relationships
those questions you are asking." Only
in
reciprocal
and causes (which focus on the behavior produced by antecedent conditions rather
than the intentional action of knowledgeable actors).
underlying our
not conducive
the research stimulus. Further,
to
emphasis on causal relationships sets up an opposition between reasons (which or voluntary motives)
is
and
researcher
and
sciences. While the results of studied, the results of
clearly
human
can and do transform the
the social sciences, unlike
in
natural science,
keeping the concepts, theories, and findings of the researchers "free from appropnation
by lay actors". Clearly behavioral IS research enters
into
the very constitution of the
phenomena
it
12
studies,
manner
a
in
not available to natural
between human science research and human of this, claims of objectivity
4.1.
What
is
and value-neutrality
we
and the
in
positivist
many
and
education
marketing,
interpretive
in
the two are not independent of each other.
the
human sciences become
make sense
but rather as "an
paradigms
emergent
of
The world
of that reality.
social process
continually involved
interpretivist
-
research
is
in
because "man
is
then the interpretive task
The
the
actors
social
are given and
is
human consciousness and
that "any individual
[Winograd & Flores
of a social group,
1
in
understanding
The aim
986:28].
through
subjective
all
their participation
m
As Denzm [1983:132]
and endow them with meaning.
one
of
unraveling and revealing the meanings persons give to their webs."
is
in
of significance, feeling, influence
emphasizes the importance
humans
the processes through which
particular, this tradition
unproblematic,
but
does not presume
attempts to understand
and
participation
in,
that social structure, culture or relations
how and why
models, assumptions, and interests are never
to
through their
individuals, it
meaning. Meaning and of
mind which can
constitutive of those behaviors.
we must uncover
human thought and
social-political
construct and reconstruct their reality
merely because they reveal subjects' states
understand what human beings are, then
he has woven,
that
meanings and
a social world, give
be correlated with external behavior, but because they are
understandings that penetrate
and power
of subjective
his
of
webs
intentional descriptions are important not
implicit
that
in
socialization into, interaction with,
tacit
researchers)
the
caught
as well as symbolic action In
(including
and hence
not conceived of as a fixed constitution of objects,
is
activities of interpretation"
interpretive perspective
[Morgan 1983:396].
receiving
between the
distinction
thereof, are social products
understand how members
to
is
knowledge
as an extension of
social processes, enact their particular realities
notes,
light
the former's primary presumption of social constructionism.
is
experience" [Burrell & f^organ 1979:253]. The basic premise is
the
problematic.
A fundamental
psychology).
social
understood independently
incapable of being
world
In
social science fields (organizational studies, political science,
Interpretivism asserts that reality, as well as our
construct and
a reciprocal and reflexive relationship
focus on the intrinsic premises of the interpretive paradigm, which
increased attention and popularity sociology,
life;
is
an Interpretive Perspective?
section
In this
There
science/
"If
we
are to
those models, interpretative schemes, and
action" [Bernstein 1978:230].
be taken
for
Meanings and
granted (as they are
in positivist
research), but are always to be understood as historically and contextually situated [Morgan 1983:397].
Thus the underlying premise
'This support,
is
particularly evident
and
in
of the interpretive
researcher
is
"that individuals act
areas such as end-user computing, implementation, IS management,
the strategic use of information technology
towards things on the
staffing, task design,
group
13
basis of the meanings that things have for them, that meanings arise out of social interaction and that
meanings are developed and modified through an
In this
sense
of attempting to
understand meaning,
meaning
actions are meaningful, and
interpretive process" [Boland 1979:260].
is
approaches are not
positivist
open
not a category
to
useful;
causal analysis; thus so long as
meaningful actions form the subject matter of social inquiry, the most important category understanding of social
not be that of
will
life
guidedness" [Winch, quoted
cause and
Ryan 1970:131],
in
The primary endeavour rigid
to describe,
is
a
[1973:216] notes that
can only answer
in
in
terms
of
our questions and our categories"; to
draw on
their
in
methodology
captured
satisfactorily
social
in
Rowan
1967].
the sense that they
contrast interpretive techniques allow
own concepts and
experiences.
is
premised on the
hypothetical deductions, covariances
in
process one must get inside the world
of
complexity of social process cannot be
belief that "the
and degrees
those generating
it"
of
dramas. Clearly researchers are never able
to
expectations they bring with them into the
completely suspend
field.
phenomenological and "grounded theory" schools)
expunge any a
Some
interpretive
explicit,
engagement
f^^ost
the
field,
such
It
is
generalization
generalization that
1985:17].
In
the
may be conducted" [Rosen
first
causal is
make make an
relationships
conducted
more global
from a sample
level,
in
the interpretive
important distinction between the positivist sense of to
the population),
view every particular social relation
at a
and the
1986:15].
and a second mode
"the extension from the micro-context to the totality that
latter
mechanisms operating
to
the
foundation from which
[that provide] the
of non-generalizability is often raised against studies
necessary (of
...
to
at the
accompany
that the interviews, archival research, observations
interpretive, social constructionist analysis
tradition.
those of the
methods however,
of the other interpretive
are "accompanied by an overlay of social theoretical ideas
The argument
knowledge and
(particularly
substantive perspective (albeit a flexible and dynamic one) in
the
ideology and sociul
the theory, cultural
methods
is
researchers must deliberately attempt
insist that the
pure data-generated "essence" of a phenomenon,
researchers'
all
rituals,
knowledge, so they can adopt a tabula rasa perspective and hence arrive
priori
propose that an
freedom. To understand
[Rosen 1986:10]. Central here
focus on meaning formation or sense-making via language, symbols, myths,
an
analyze and
Features of Interpretive Methodology interpretive
like,
& Strauss
research "we are talking to "processed people"
own words and
participants to use their
4.2.
positivist
rule-
researcher-imposed
priori
formulations of structure, function, purpose and attribution are resisted [Glaser
our
for
meaningfulness and
effect, but that of
understand the social world from the actors' perspective, and any
"Human
and hence the
is
shaped
it"
of
[Burawoy
the product of generative forces or
interpretive analysis
is
an induction (guided
14
and couched within a theoretical framework) from the concrete individual
case An example from a study
systems development process
of the
[1982:33] interpretivist research where they note "What
be generalized structure]
to
other design settings, even though
be tested by
is to
its ability
Morgan [1983:398] employs 'esearcher
is
more concerned
this
predictions
and
^neralization
of
in
same usage
statistical generalization.
relative.
'Cich
when he notes
...
[such a
"
that "the interpretive
methodology
-
a
similar
distinction
between
[1983:165] indicates that those
these
-
of studies,
as Weick [1984,
serve to establish a theory's pattern of adequacy.
make
criteria
verified
are understood to be indeterminate, as well as
different kinds of
knowledge vary from one
significant
and
theoretical
interpretivists refer to generalizations,
However, over time and through a series will
fact,
the crucial issue of evaluating the credibility of interpretive research,
to
what counts as
makes
Even though
agnize that "different research perspectives :o
only explored a single instance
of generalization,
[1984]
ng McGuire] notes, empirical confrontation
With respect
a deeper structure that can
is
terms of measured relations between networks of Yin
outcome".
and contextually
we have
to explore
given by Boland & Day's
describe the experience of other system designers
ntrary to those characteristic of positivist -iporally
we hope
is
beyond the
with identifying generalized processes that are not content specific
erefore cannot be characterized
'ough
to
situation to the social totality
to
knowledge claims, and the
another" [Morgan 1983:393].
espoused by the
we must
positivist tradition are not
criteria
In particular,
appropriate for
evaluation of interpretive research: "Networks of shared meanings do not lend themselves to study by
:nods of detachment and objectivity. Instead the researcher ?rting of
data are done systematically with care and discipline."
soundness
of the analysis, that
ond the particular setting ysis to the actors
ortunity 'her,
is
for
in
the
at
is,
"scientific"
m
She suggests
on how the research contributes
to
that the collection
and on
that evaluation rests
our understanding of social
life
hand. Interpretive methodology stresses the importance of submitting the This not only allows for verification and criticism but also provides an
field.
the collection of further information through correction, clarification, and elaboration.
as Nissen [1985:49] notes, the adequacy of interpretive studies as "theoretical reconstructions
~an experience"
will
be subject
to critical
intersubjective scrutiny by the
community
of scholars,
of
and
3Cted or refined as appropriate.
A number inquiry"
of interpretive
and attempt
researchers address the question of the "trustworthiness of interpretive
to allay the
oft-expressed concern that
it
is
somehow
"non-scientific".^
Pettigrew
[1985:246] stresses that primacy should be given to realism of context and to the generation of measures
*See Lincoln & Guba [1985]
tor
an extensive discussion
of
what they term
"naturalistic trustworthiness critena"
15
that are "sensitively linked to the subtleties
and nuances
the use of acontextual and blunt instruments.
between good and bad
distinguish
of a particular context or contexts", in contrast to
He provides some broad
guidelines that can be used to
interpretive research [1985:247-249]:
research should strike a balance between description and analysis, where the role of description is to clarify and establish the context, structure, and process to be explained by
• the
the analysis; •
there must be an empirical and theoretical justification for both the chosen time frame (the horizontal component) and the levels of analysis considered (the vertical component);
•
the research should display an adequacy of data sources,
i.e. the use of multiple sources and multiple methods of data collection over time, the cross-checking of data across sources and among methods, and the examination of competing theoretical interpretations (or rival plausible hypotheses in the parlance of positivism);
and analyses need to be interpreted by theoretical themes that represent an attempt at generalization, "at placing the work within a wider scheme of things theoretically and conceptually" [Petligrew 1985:248];
• the findings
themes and concepts must be
• the theoretical
In their
that the it
will
closely coupled to the data.
formulation of grounded theory, Glaser
grounded theory has
be used. That
realities of the
& Strauss [1967] discuss
practical application. Firstly, the theory
in
the
substantive area.
The
intent
is
if
is
it
to
the theo'y to
for
assumptions, hence sharpening people's
sensitivity to the issues they face,
directions of potential change.
must be
Thirdly,
it
sufficiently
general
that will arise daily within the substantive area, not just to a single,
and concepts
of the theory
general guide
to the
users
in
should not be too abstract as
to
be relevant
to
many
situations.
number
in its
who employs
everyday
to
to
to the
reveal taken-for-granted
and
to indicate
be applicable
images and issues
to multiple
be obtuse, yet they need
What
is
required
is
to
serve as a
diversity
which
general concepts with plausible interrelationships so as
Such a theory
person applying the theory participates [1967:243] note that "A person
of
to the
which
narrow type of concern. The categories
negotiating their ever-changing reality.
allows the construction of a sufficient
be applied
in
make sense and be understandable
substantive area. Secondly, the theory must
people working
must f/nhe substantive area
the theory must correspond closely to the data
is,
four properties that ensure
is
not
validation
seen as
static or "closed", but
as a process: the
and development. By contrast, Glaser & Strauss
quantitatively derived theory
"knows
his
few variables better
than anyone, but these variables are only part of the picture"... This kind of theory typically does not
account
for
enough
variation
Also, such theory usually
in
situations to allow appreciable institution
does not
offer sufficient
means
for predicting
and control
the diverse
of
change
consequences
of
puposeful action on other aspects of the substantive area." Finally, the theory must allow the user control oyer the structure that the theory should
difference
in
and process
of daily situations,
as they change over time.
address issues that "are worth applying the theory
the lives of participants, so that they
will
to", that is,
The
them.
in
any
partial
notion here
issues that
have the motivation and understanding
make to
is
a
gain
16
control of situations.
4.3.
Types
of Interpretive
A number
research methodologies within the social sciences have been developed
of
these
practice
Research
notions
ontology
interpretive
of
and
epistemology,
to
put into
phenomenology,
including
ethnomethodology, hermeneutics, symbolic interactionism, contextualism, and ethnography While there
among
are commonalities
these specific approaches
frame
social world primarily from the
in
that
attempt to understand and explain the
all
what
of reference of the participants, there are differences in
For example, phenomenology
constitutes the focus of study
consciousness
in
is
concerned with methodically studying
The
order to understand the essence of experience [Boland 1985:194].
intent
is
to
provide a self-reflective description of the situation of interest, via the interaction of participants and
A conscious
researcher.
attempt
is
made
to
mutually explore, define and then deliberately suspend the
prejudices and assumptions atxiut the world that the participants and researchers bring to the interaction
m
[Schutz 1967]. For example, Boland & Day [1982] employ a phenomenological approach
year daily
in
the
work
systems designer, mutually exploring with him the various pressures and issues
of a
life
of his
such as interaction with users, constraints on design, moral dilemmas, and personal
life,
biases. Through such a study a that
studying a
number
of
work concerns and issues
had not been previously revealed by inquiry cast
Through such a mutual inten/iewer
-
m
relating to
systems design emerged
the systems rationalist
mould [Weick 1984:128].
interviewee exploration, the social processes mediating technological
development and use became apparent, providing
insight into
some
of the
meanings underlying systems
design.
Ethnomethodology
common methods
for
,
a derivative of phenomenology,
constructing
reality.
Garfinkel [1967]
those social processes or conventions that people use reality.
The
activities,
interest
reality.
learning about the
and how they account
conventions are
become
is in
the
implicit
unspoken
for
ways
these
and taken-for-granted so
"rules"
and norms
inappropriate.
by
examining
For example,
individuals
that the
Zimmerman
modified the stated rules of the agency not being achieved.
how
come to
everyday
Typically these
common assumptions we make
where the
[1970] studied
when
to refer to
of their
& Morgan 1979:247].
behavior and hence
situations
evolve
negotiate and reach consensus on the nature of
others [Burrell
of social
in
cultures or groups
employs the term ethnomethods
The research methods adopted by ethnomethodologists aim
assumptions
that
which people order and make sense
in
to
to
proposes
to define
about
reality
and sustain our
understand these underlying
codes
"normal"
receptionists
in
of
behavior
are
a public welfare office
the purposes for which these rules had
been
instituted
were
17
Symbolic Interactionism examines the process by which individuals in,
how
ar^d
social
they construct their acts accordingly [Boland 1979:260]. Symbolic interactionists believe that
and
action
interaction
only
is
possible through exchange of shared
establishment of social rules. However
embedded
themselves, but
be seen as existing
meanings the people put on events,
the
in
reality is not to
and shared routines [Tomkins & Groves 1983:368]. Some symtwiic understand the meanings that phenomena have
for
interaction
and
Contextualism
is
to
conform
to or
adapt
concerned with the "event
phenomenon
the historical situatedness of the
the rules and routines
and people
via the rules
interactionists [Biumer 1969] try to
individuals influence their social reality by shaping rules, rituals
and shows how people use symbolic processes
negotiation,
and "impression management"
in
situations,
Goffman [1959] analyses the
the "presentation" of themselves to others.
and the
interpretations
people, and thereby identify significant forms of social
how
behavior. Other symbolic interactionists examine
human
interpret the situation they are
their
circumstances
and routines
of
of "situation defintion"
with their purposes.
in line
[Pettigrew 1985], and the primary focus
in its setting"
is
example, Pettigrew [1973, 1979, 1985]
of interest. For
studies the relationship between organizational structure and process over time and space, via in-depth
immersion
and
in
subjective or interpersonal
change, that
resources
in
processes
one
to
dynamics
British firm.
decision-making
relations of
is
less on
of continuity
and
For example, Pettigrew [1973] studied the
surrounding
Through in-depth analysis over
wrought by such decisions on the
mechanisms
identifying the
of various contexts.
ongoing
the
in
The focus
a particular organizational setting.
meaning construction, as on
the processual
is,
organizational
exposure
longitudinal
the
time, Pettigrew
purchase
computer
of
examines the changes
dependence among occupational groups
within
the
organizational context.
Ethnography
is
ethnographer's method data.
He
or
she
an
anthropological
for collecting
tries to learn
data
it
to others.
to live
for
how and why
Interpreting the
both
data
collection
and
among and observe those who
the subjects' rules for social
duration of time "sufficient" to understand
there to explain
is
method
life,
to interact with
them
analysis.
are generating the for
a frequency and
they construct their social world as
meaning behind behavior
is
The
it
is,
and from
the basis for ethnographic
research.^ Geertz [1973] identifies the cultural form of interpretation as "thick description," a statement of
what
is
observed among a social group based upon action and the meanings that underlie and follow
from such action.
If
the researcher reports observations purely descriptively, primarily focusing on the act
devoid of meaning, the audience
is
^e focus
given a "thin description" of the occurance. Thick description, on the
here on a brand of ethnography pnmarily concenned with 'cultural analysis" [Geertz 1973] However there are many and interested readers should consult Sanday [1979] for an elaboration of the diversity of thought within ethnographic study. styles or schools of ethnographic research,
18
other hand,
is
concerned with
conspirational wink, that
imbued
of description
is,
between an involuntary muscle
the differences
telling
Ethnographic description
with interpretation.
be confused with the recountmgs that would be provided by the actors themselves is,
instead, an explanation cast
recounting [Geertz 1973:15]
in
in that
and language
the theory
the researcher
of the
researcher.
making interpretations
is
twitch
of
in
It
is
and a not to
a social setting, is
it
a second order
data obtained indirectly
''° through participant observation or interview.
Despite the roots of ethnography
early social anthropology,
in
it
was
not
until
the I970's that
ethnography was employed as an epistemology and a methodology within organization theory [Van
Maanen 1979,
The emergence
1988].
"Organizations are not simple systems
machines
like
manifesting complex patterns of cultural activity
[Pondy IS
speaking
specifically
people their
According
et al. 1983:4].
particular
in
to
Van Maanen
settings
come
...
to
[they] are
of organizational structure
The method
of organizational
members, but
individual
-
is to
by
very nature, symbolic entities"
their
is
for,
-
and here he
take action, and othenwise
a methodology
for
and processes as organizational is
human systems
"uncover and explicate the ways
understand, account
ethnography
growing awareness that
[1979:540], a principal aim of ethnography
day-to-day situation." Organizational ethnography
components
reflects the
or adaptive organisms; they are
ethnography
of organizational
work
ethnography
of organizational
in
which
manage
"seeing" and interpreting the
life
is
created and maintained.
not to simply recount the organizational experiences of
rather to use social theory to explain these experiences, which
m
turn,
expands
and refines social theory.
All
of these research
methods can
offer a refreshing
of interest in IS research. In the following section
can make
5.
to
research
into IS
INTERPRETIVE Having outlined
we now
consider
of articles
its
IS
will illustrate
utility
of the
Benbasat, Goldstein &
examples
to follow traditional poitivistic
similar
of
an additional research paradigm,
of behavioral IS research.
approaches.
we
We
In
our sample a majority
do, however,
see an increased
believe the behavioral IS research
a variety of research approaches. Recent discussions [Boland
Mead
1987; Franz & Robey 1984; Galliers
1985; Weick 1984] draw attention
On a
phenomena
the contributions that such approaches
assumptions and methodologies
applicability to specific
of
perspective on the
phenomena.
recognition of the need for additional approaches, and
accepting the
insightful
RESEARCH
some
were found
we
and
to
field
is
& Day 1982;
& Land 1987; Lee 1988; Mumford
et
al.
the value of such additional forms of inquiry for IS research.
theme see Van Maanen 119791 on
first
and second order constructs
19
Particularly interesting
ACM
[November 1988] where
means &
the recent collection of research articles published
is
Maass &
Iscoe, 1988; Rosson,
increase the insights
such as ecological design [Soloway
Such methdological
Kellogg, 1988].
generate about the phenomena of interest
we
the Comnnunications of the
and exigencies
investigators explored the nature
of non-positivistic methodologies
in
of
1988; Curtis, Krasner
et al.
diversity
is
systems design by
welcome and bound
to
research and practitioner
to the IS
communities.
5.1.
Choosing the Appropriate Research Approach
We technology
see both in
positivist
organizations.
to deliberately exploit the
and
mterpretivist
studies as
What we urge however
is
that these
We recommend
strengths of each.
methodologies be used prudently, so as
that epistemological choices
approach and technique be made consciously and with attention
phenomenon under
investigation.
established effects, and less appropriate
This
same wisdom
commonly
is
It
for
held
that
suggested by a few commentators). On the contrary,
it
to
studying
information
approaches and techniques each
technology
of
organizations
in
to positivist
to
meaning
to
testing
of findings.
only appropriate for exploring
its
research (as has been
own, providing unique and
However we is
nature of the
studies are suited
can stand on
valuable insight into the reciprocal complexity of social processes.
approach
is
about research
particular
exploring relationships or explaining the
as only serving as a precursor
it
the
to
positivist
holds that qualitative or interpretive research
phenomena. However we do not see
information
vehicles for studying
useful
feel that the
employ a whole
most
fruitful
repertoire
which can complement and enhance each other. The caveat
is
of
that
researchers be mindful of the implications that follow the adoption of any one methodology's set of
assumptions.
Positivist
research
assumptions underlying actors.
particularly suited to studying structural
is
this
methodology
Hence researchers adopting
as predictors of social action. area of social relations interpretations,
-
This
positivist
and the construction enactment
human
this interpretive
limits their ability to investigate
will
behavior, as the
prior to the individual
focus on and investigate external forces
researchers realise that they are omitting a whole
and
social reality. Likewise, interpretive research
shared, social
the point of view and through the interpretive
researchers adopting
human
action dimension that revolves around shared meanings,
of a cultural of a
if
of
system as ontologically
methodology
appropriate
is
the interactive,
useful for investigating the
posit the social
causes
schema
perspepective need
reality
of to
is
by understanding human behavior from
the
human
actors themselves.
understand that
their
However
methodological choice
the structural pressures on social action.
Both structural and action-oriented perspectives on social relations are important and worthy of
20
Hence both methodological perspectives can
study [Giddens 1979].
information technology use the
investigating
m
organizations, provided each
phenomena
interest
of
We
relevant insights.
approaches can bring
We
can envision a number
different influences to
of scenarios
bear on the
where
and
interpretive
•^search design allowed for the
complementary and
investigated.
in their
approach:
(i)
investigation of the impact
the inconsistency of results
development
and omissions;
to potential analytical errors
of context-sensitive
measures as the
positivist
the
for
ationships.
exploration
questionnaire
Lee [1987] argues
and
hence
avoided
using positivist and interpretive approaches
for
oical scenario
presented
for
combining these approaches. Weick [1985:575]
/ersing this traditional order
^s
in
is
more
m
a mutually supportive This
the
is
contrast, suggests that
can often shed
useful, noting that qualitative research
it
on
light
suggested, but not explained, by quantitative techniques. He writes "To understand the
::tionships
which a data table
-imarized
in
nuanced, what the data mean, and how people generated the
is
numbers, one must
the
scriptive analysis.
...
treat the
activities
table as a preliminary display that triggers serious
Quantitative studies produce discoveries, the
meaning
of
which
is
known
only
they are verified qualitatively."
While
positivist
jnnot stress
and intepretive research approaches can be
enough how important
ethodological position. This eing investigated.
To
will
illustrate
it
is
for
researchers
be cognizant
to
variable (say, A)
interval \+^
variables
,
but
is
of the variables
implicitly
in
which the variables
are decided upon, the time intervals for is
in their
phenomenon
distort the
t,
of
measurement
employed. Researchers may thus
and negatively
the underlying process or reason for the relationship.
It
find
related at
may be
that
are simply the surface manifestations of more fundamental, even contradictory forces,
and hence should not be the primary focus design
of the limitations inherent
positively related to another (say, B) at interval
know nothing about
A and B
withm and across studies, we
consider a positivist longitudinal research design,
are preselected, and the hypothetic-deductive model of analysis
one
utilized
ensure that researchers do not inadvertently
merest are prespecified, measures
that
m
(iii)
causal
one-directional
simplistic
with the positivist studies testing the hypotheses generated by the interpretivist studies
ay
ei'
interactions
of
the
(ii)
as developed on the basis of the interpretive unstructured interviews and participant observations:
:owed
and
that positivist
data collection and analyses techniques prompted further exploration,
positivist
deeper insights and alerted the researchers
fielding
this
of
creative combination of research
phenomena being
a medical setting. They note a number of benefits of
^e•ween
recommend
thus
Kaplan & Duchon [1988] adopted such an integrative approach of IT in
be the only valid way
to
either within or across studies, so as to yield
approaches be combined,
interpretive
does not claim
our discipline.
to
contribute to our understanding of
presumes
stability of
of interest, but
such insight
will
go undetected.
Further the
the variables and so, by definition, discounts the operation of social
21
processes.
Assume researchers
Consider another example:
between user (after three
a rate
months)
much lower
system.
Initially,
is
Now have
than
initial
will find
much
learn (hence the
initial
its
level of high
functionality they
of queries issued
much about how
it
tell
us
play a central role
measure
the
of
three
initial
began
to
months? Was
it
approach
to
the data
change, multiple time frames and
a viable alternative
5.2. Specific In
to
used once
is
it
we
are
left
is
retrieved?
discussion
IT
filed,
Perhaps there are some other possible with an incomplete understanding of the
The
point to note here
rnariy
dependent and independent variables would be required. This if
by no
we draw on
a
not impossible.
number
means meant
to
An
interpretive approach,
of
streams
of behavioral IS
believe,
research and attempt
Many
Impacts Research:
Attwell
implementation
of
reviewing the literature on the effects of computing
in
of the studies into the "impacts" associated with the
& Rule
[1984],
in
In their
conclusions they state: "The sheer variety of
disparate and seemingly conflicting conclusions that can be derived from the studies noted
warrant despair.
static
we
be prescriptive, but rather suggestive and hopefully provocative.
organizations found very few clear cut answers.
in
is
scenarios where an interpretive perspective might usefully be used. This
illustrative
have been inconclusive.
that the
discarded
Research Scenarios
some is
the data
Is
such cases.
the following discussion
provide
(!)
m
system was
just that the
capture the richness of the users' experience and the social mechanisms
would make the analysis complex and unwieldly, is
the
use inappropriate? Does computer connect time or number
underlying processes that would explain the occurrence of the observed events. that for a positivist
was
accomplish more with less use (lower use but
directing a person's activities?
in
If
use and low satisfaction) and that as users became more
explanations for the observed relationship. Indeed
of
finally stabilizing at
use and satisfaction have an inverse relationship over
that
longer than the anticipated
was
does
use
Contrast this with the scores they obtain on user satisfaction with the
use.
higher satisfaction)? Or
or
find that initial
the researchers demonstrated that use declines as satisfaction increases, or
and discovered
proficient
Imagine that they
relationship
the satisfaction levels are low but over time they increase, stabilizing at a high level.
learning curve just difficult to
of that system.
studying the
in
very high and then tapers off over the next six months with use
the researchers correlate the data they time.
and use
satisfaction with a system,
are interested
Why do
problem may be snapshots
all
these worl