The ethnographic research tradition in anthropology
is a valuable starting point for a consideration of the
philosophical basis of interpretive case studies, since it
has been widely drawn on by organizational researchers
concerned with interpreting the patterns of symbolic
action that create and maintain a sense of organization
(see, for example, Smircich, 1983). Geertz (1973) gives a concise view of the status of the data which are
collected in an anthropological study:
What we call our data are really our own constructions of
other people's constructions of what they and their
compatriots are up to. (p. 9)
Van Maanen (1979), writing in the tradition of
organizational ethnography, calls the interviewee's
constructions first-order data and the constructions of
the researcher second-order concepts. He warns that
assuming an ethnographic stance is not a guarantee that
reseachers will collect valuable data no matter how long
they stay in the field. Second-order concepts rely on
good theory and insightful analysis, and mere collection
of in-depth case study data does not provide these
concepts in itself. Examples of second-order concepts
in the IS literature, derived from interpretive case
studies, include the 'automate' concept from the work
of Zuboff (1988), and the concept of 'technological
frames' in Orlikowski & Gash (1994).
A second feature of the anthropological tradition is
its concern with 'thick description'. Geertz (1973) gives
a fascinating example of this involving Jews, Berbers
and the French in Morocco in 1912. The incident
recounted involves 'sheep stealing' by one of the Jews,
who Geertz calls Cohen, from some of the Berbers.
However, on closer examination, the 'stealing' turns
out to involve compensation for an earlier incident in
which Cohen was robbed and nearly killed by members
of the same Berber group. The French misunderstand
this and put Cohen in prison for what they take to
be simple theft on his part. Geertz uses the incident
to point out that the ethnographer is faced with a
multiplicity of complex conceptual structures, many of
them superimposed upon or knotted into one another
and which must be first grasped and then rendered
intelligible to others.
The IS researcher entering an organization today is
also faced with complex and intertwined conceptual
structures which it is difficult to grasp and render
intelligible as did Geertz in his anthropological work.
The need for 'thick' description is just as important in
trying to understand what is happening in connection
with a complex computer-based information system,
involving managers, users and designers, as it was in
trying to interpret the interactions of Jews, Berbers and
the French in Geertz's study. As a specific illustration
of this, Boland and Day (1989) describe how a system
designer reinterprets the behaviour of someone who at
first she thought was trying to help with her design
work, but on further reflection over a period of time
she concludes that he was trying to isolate her from
others for his own political interest. An IS researcher
can only access these subtleties of changing interpretation
by the use of approaches based on 'thick'
description.
In discussing the purpose of his studies, Geertz
argues that he is not trying to answer our deepest
questions about other societies, but merely to make his
interpretations of these societies available in the
'consultable record'. His goal is not to generate truth or
social laws, and this interpretive approach can be
clearly distinguished from the positivist tradition. This
should not be taken to imply that interpretive work is
not generalizable, although the nature of such generalizations
is different in the two traditions. This point will
be considered in some detail in the penultimate section
of the paper.
The differences between interpretive and positivist
approaches can be addressed more formally by considering
their epistemological and ontological stances.
With respect to epistemology, concerned with the
nature of knowledge claims, Archer (1988) defines
positivism as the position that facts and values are
distinct, and scientific knowledge consists only of facts.
He contrasts this position with two alternatives: 'nonpositivism'
in which facts and values are intertwined
and hard to disentangle, and both are involved in
scientific knowledge; and 'normativism' which takes
the view that scientific knowledge is ideological and
inevitably conducive to particular sets of social ends.
Either of the latter two positions is open for the
interpretive researcher to adopt.
With respect to ontology, concerned with the nature
of reality, Archer distinguishes between 'external
realism' which considers reality as existing independently
of our construction of it, 'internal realism' which
views reality-for-us as an intersubjective construction of
the shared human cognitive apparatus, and 'subjective
idealism' where each person is considered to construct
his or her own reality. The usual ontological stance for
an interpretive IS researcher would involve one of the
latter two positions, particularly with regard to the
human interpretations and meanings associated with
computer systems.
The above brief discussion of epistemology and
ontology is summarised in Table 1. It clearly distinguishes
the positivist tradition, but it does not provide
any definitive answers as to what precise philosophical
stance should be adopted by the interpretive IS
researcher. In the related field of 'systems', Mingers
(1984) argued that there is considerable value in a
careful examination of the philosophical basis of
different types of interpretive approaches, and he
identified the existence of at least four substantively
different strands of thought: phenomenology, ethnomethodology,
the philosophy of language, and hermeneutics.
He used this analysis to provide a thoughtful
critique of the underlying philosophy of various key
writers in the systems field, including Checkland (1981)
on soft systems methodology. The different strands of
thought identified by Mingers can be seen to underpin