Q1. Rudner defines a theory as a
systematically related set of statements, including some law like generalizations
that is empirically testable. The purpose of theory is to increase scientific
understanding through a systematized structure capable of both explaining and
predicting phenomena." Agree or disagree, and why? What are the
requisites of a fully formalized theory?
The role of theory in science is
contained in the following consensus definition of 'theory' proposed by Hunt
(1983) following Rudner (1966): A theory is a .systematically related set of
statements, including some law like generalizations, that is empirically
testable. The purpose of theory is to increase scientific understanding through
a systematized structure capable of both explaining and predicting phenomena
(emphases ours).
This definition incorporates the key
elements of the nature of theory as proposed by philosophers of science from
different branches of the social sciences as Kaplan (1964)—sociology, Blalock
(1969)—statistics, Aiderson (1957)—marketing, and Friedman (1953)—economics. We
feel therefore that this is a reasonably complete specification of the
essential criteria and purpose of theory.
MWB make two normative claims in their
Propositions 1 and 2 regarding the preferred method of theory construction,
i,e, 'all theory generation should depend on some past observation' and 'all
observations should be guided and interpreted through some theory,' Both
propositions imply that the reverse can also be true, but are dysfunctional
ways of theorybuilding. MWB seem to argue that theories can be generated in the
absence of any observations and that data can be interpreted in the absence of
any theory, i.e. theorizing is possible independent of past experience and data
are theoretically neutral. However, a number of philosophers of science
(Kaplan, 1964; Churchman, 1971) have shown that these reverse statements cannot
be true. The mere process of deriving law-like generalizations about a
phenomenon involves the experience of the researcher, and all interpretation of
data is conducted within the context of a framework imposed by the researchers.
Therefore, MWB's first two propositions should be expressed as positive rather
than as normative statements, and as such we are in general agreement with
them. However, an interesting issue with regard to the role of observations in theory-building
remains. Churchman (1971) has pointed out that initial observations have
different roles in two methods of theory generation, i.e. the inductive method
and the deductive method. The inductive method of the Lockean Inquiring System
is 'the process of starting with highly warranted (or well agreed upon)
observational statements about specific events and inferring a generalization'
(Churchman, 1971: 94). Thus, observations are the very basis of the theory. On
the other hand, the deductive mode is 'the process of using a set of
assumptions to prove a theorem by some standard set of rules of inference'
(Churchman, 1971: 94). In this method the role of initial observations is to
provide a basis for speculation about the phenomenon, which is then followed by
development of assumptions and the hypothetical model from which
generalizations are deduced.
Now, MWB state that their Proposition 1
is 'the basis of empiricism, or what Churchman (1971) calls the Lockean Inquiry
System' (p. 190), They therefore seem to imply that all theory generation
should be based on the pure inductive method (i.e, the Lockean Inquiring
System).
It is useful to view the pure inductive
method and the pure deductive method of building theory as representing the
extremes of a continuum. In contrast to MWB, we propose that methods falling
along all points of the continuum, including the deductive method, represent
valid ways to generate theory for strategy researchers. While the inductivist
route has had a prime role in the development of strategy theory, there is no
clear reason why this must be normatively true. The research question and the
phenomenon of interest dictate whether 'more inductively-oriented' techniques
with greater emphasis on the role of initial observations, or 'more deductively
oriented techniques' are likely to be useful (see Karnani, 1984, for a good
example of the productive use of deductive techniques). MWB seem to feel that
because operations researchers and economists, who often use deductive
techniques, sometimes sacrifice relevance for mathematical elegance, this
somehow makes deductive techniques deficient. However, it is not the deductive
procedures which underlie the deficiencies in the theories, but rather the
failure of these theories to correspond to the norms underlying 'good'
theories. These norms are discussed in the next section.
MWB's notion of what constitutes 'good
theory' is contained in their Proposition 3 and discussion of Assertion 1.
Proposition 3 states that 'a theory is better, ceterus paribus, (a) if it is refutable
and (b) if it is consistent with a body of existing theories' (p. 190),
Assertion 1 makes the suggestion that 'well-reasoned' theory should underlie
strategy research, and MWB offer a series of examples of 'well-reasoned theory'
in their discussion.
Because of the interdisciplinary and
integrative orientation of the field, most phenomena of interest to strategy
researchers are highly complex in nature. Therefore, the complete specification
of necessary and sufficient conditions, as well as the explicit formulation of
assumptions, are extremely difficult tasks. For the most part we deal in
partial explanations and the predictive power of our theorizing is limited.
There is an intrinsic tension between the
requirement that theoretical statements have 'precise' explanatory and
predictive power and that they be applicable to a 'wide' range of
circumstances. This issue is of particular concern to strategy researchers, who
seek an understanding of strategic behavior at different levels of aggregation
(i.e, at the firm level, at the strategic group level, at the industry level,
and across industries). Often, predictive precision can be obtained for
phenomena at lower levels of aggregation (such as strategic groups) but this
may imply sacrificing generalizability (for the operation of the phenomena
across industries, for instance).
Do these objections make the task of
seeking lawlike generalizations futile? Not at all. We believe that it is a
worthwhile endeavor to continue to build mid-range theories, for this can be a
useful path to the development of more complete theories. Our view is that a
theory is better if it explains a wider range of phenomena while striving to
aim for predictive precision. For example, a theory of the relationship between
market share and profitability would be 'better' if it isolated the particular
circumstances which govern the nature of this relationship at different levels
of aggregation.
If strategic management is to become a
science it must strive towards 'explaining by law' the phenomena of interest.
It is true that the field has traditionally been concerned with generating
normative implications for practicing managers. However, since the goal of
science is to explain and predict phenomena, the role of positive research must
be recognized. Three related issues merit further discussion.
The first issue has to do with the model
of science that is appropriate to follow. In our view it is doubtful whether
physics, where even ultimate applications often do not constitute a primary factor
for directing research efforts, would be the best model. Rather, it must be
recognized that strategic management is more an applied discipline (similar,
perhaps, to engineering), and therefore concerned with ultimate application of
research findings. However, these applications may sometimes take a while to
emerge. Therefore, we concur with MWB's argument that direct practical
applications should not be required of ail papers.
The second issue concerns the
identification of the audiences for whom the applications are being generated.
Strategy researchers should be encouraged to seek knowledge which is
generalizable beyond the confines of their members. To approach the status of a
'science' it is beneficial to examine issues which are valued in the larger
community of scholars and practitioners. Strategy research findings may be
extremely relevant for public policy-makers, researchers from other
disciplines, consultants, the popular press, or the public at large.
The third issue concerns the role of application
in the development of the field. MWB provide an interesting discussion of the
'division of labor' among researchers. However, this division may be even
broader than MWB suggest. Some philosophers of science suggest that there are
two distinct roles necessary for the advancement of knowledge (Manicas and
Secord, 1983). It is important for scientists (either 'pure' or applied) to
conduct theoretical and empirical research in order to uncover causal
structures. Under this model, it is then the role of the technician to apply
these decision rules to the situations faced by particular organizations. For
instance, a manager or consultant may make strategic plans for a corporation
using an analytical framework which is developed from research findings.
It is important to
realize that the theoretical issues raised by MWB are present (though often
only implied or assumed) in every empirical article published in the strategy
literature. We feel that the strategy field is at an important juncture.
Research can continue in a rather unfocused fashion (as well described by MWB),
or attempts can be made to coordinate research efforts to conform with certain
guidelines. The guidelines we suggest are of necessity somewhat broad. It is
important to guard against forming arbitrary rules to guide the research
process, because these may in fact hinder the transition to an organized and
recognized science.
