Business relevance of propositions
Box 8 Is business reality deterministic or probabilistic? A note on “pragmatic determinism”
In Chapter 4.3 we claim that many causal relations in real life situations in business and management can be formulated as deterministic necessary conditions. This claim is
usually received with scepticism. Most business researchers assume that deterministic conditions and relations do not exist in the actual practice of management and busi-
ness. It is assumed that every causal relation that is of interest to business research is multi-causal or multi-factorial and, thus, must be expressed in probabilistic statements.
Our response to such criticisms consists of three parts:
1. academic theories in business and management in fact express deterministic
conditions and relations; 2.
even if reality is probabilistic, this does not undermine the usefulness of deter- ministic theories;
3. managerial theories-in-use are deterministic.
Many theories are deterministic
Goertz 2003 reviewed the political science literature in search for theories that do not present themselves as deterministic but actually are. He found no less than 150 neces-
sary condition hypotheses covering large areas of political science, sociology, and eco- nomic history 2003: 76–94. On the basis of this finding he formulated Goertz’s First
Law: “For any research area one can find important necessary hypotheses” 2003: 66. We are confident that we would find an equally impressive list of necessary condition
hypotheses in a review of management theories. A prominent example is Porter’s the- ory of the conditions of competitive advantage of nations see Box 12; 9.1. Other
examples are the theories-in-use tested by Sarker and Lee see Box 11; 5.1 and the examples of case studies in Chapter 5 of this book 5.2 and 5.4.
In this book we use the concept of “necessary condition” as formulated in classic mathematical and philosophical logic. The necessary character of A for B is expressed
in this formulation by “if”: “B only if A”. The sufficient character of A for B is expressed by “B if A” meaning “always B if A”. In this logic such expressions are always either
true or false. This leads to the common view in theory that a necessary condition is dichotomous: true or false Figure A.
But conditions and effects can also be continuous. Various authors have shown that it is possible to express necessary conditions for continuous variables using multi-value
logic. Goertz and Starr 2003 present these authors and their ideas. They show how it is possible to express a continuous expression of a necessary condition, as illustrated in
Figure B adapted from Goertz and Starr, 2003: 10. In the upper left part of the graph there are no instances. The basic idea of a necessary condition as depicted in Figure B is
that a specific value of A is necessary for a specific value of B, which is expressed in the graph by the necessity that every instance is situated below a sloping line between the
area with and without instances. This idea was formulated by Ragin 2000 and by Goertz 2003.
Reality is probabilistic
The standard view of a theory with a proposition that expresses a necessary condition is the absence of even one exception of the necessary condition in the entire domain.
Finding one single instance would fatally undermine the correctness of the presumed
Present
Absent
Concept B
Absent Present
Concept A
Figure A
Concept A is a “dichotomous”
necessary condition for concept B
Concept A Concept B
Figure B
Concept A is a “continuous”
necessary condition for B
Continued
necessary condition. This situation is depicted in Figure C for the dichotomous neces- sary condition and in Figure D for the continuous necessary condition.
Figures C and D depict situations in which a large majority of instances “behave” according to the formulated necessary condition statement, but there are a small num-
ber of exceptions. But what would be a better formulation of the reality depicted in these figures? Not a probabilistic one. Despite the exceptions, a continuous necessary
Present
Absent
Concept B
Absent Present
Concept A
? ?
Concept A Concept B
? ?
?
Figure C
Concept A is a “nearly” necessary
condition for B
Figure D
Concept A is a “nearly continuous” necessary
condition for concept B
Continued
condition statement is a more fitting formulation of this reality than a formulation of a regression through the points in Figure D. In practice and in empirical research, excep-
tions to deterministic relations can always be found but the fact that reality probably is probabilistic does not undermine the usefulness of deterministic theories.
Ragin and other authors have formulated the idea of “almost always necessary condi- tions”, i.e. probabilistic statements that express a very high chance e.g. up to 0.99 that
there is a deterministic relation. These authors have developed mathematical expres- sions for such “probabilistic necessary conditions”. Ragin has also developed a statisti-
cal tool by which data as presented in Figure D are analysed in such a way that the proportion of cases on the “wrong” side of a sloping line are calculated and the “signif-
icance” of this proportion is tested against a benchmark.
Managerial theories-in-use are deterministic
Managerial relevance is not dependent on the few exceptions. Even if managers would know that the probabilistic necessary condition hypothesis is true, they would act as if
the condition was completely deterministic and make sure that the necessary condition is in place.
We use the term pragmatic determinism for the view that it is sometimes preferable
to act as if a complete determinism exists, although it is acknowledged that there might be some exceptions to the assumed determinism.
have the form “How can the company or our management team, etc., achieve the success of a project, an investment, etc.?”
For critical decision making e.g. when a decision must be made about whether or not a huge investment should be made, or when a
gono-go decision must be made about a merger a practitioner would prefer deterministic knowledge of the factors that would “guarantee”
success in other words, of “sufficient” conditions for success or of con- ditions that are minimally required in other words, of “necessary” con-
ditions for success. Probabilistic knowledge, such as “If a certain condition is present, then success is more likely” may entail too much
risk for such critical decision making and, therefore, may not be enough for decision making. Obviously, this does not imply that, in this type of
situation, having no knowledge at all would be better than having some probabilistic knowledge, quite the contrary. But it does imply that having
knowledge about a deterministic condition would be even better.
For less critical management decisions e.g. on ways of maximizing the average financial result of projects probabilistic knowledge could
be sufficient. If the manager knows which factors increase the likelihood
of success of projects, he will be able to increase the average project performance or the relative number of successful projects. Hence,
depending on criticality of the management decision, deterministic knowledge may be required, or probabilistic knowledge may be enough.
Although most research articles published in business research jour- nals deal explicitly or implicitly with probabilistic propositions, such
articles often conclude with a discussion of “managerial implications” in deterministic formats such as “This study has shown that managers
must do A in order to be successful”. We believe that much of such “deterministic” advice does reflect the fact that many managerial prob-
lems actually require or, at least, would be helped with knowledge of necessary conditions for success see Box 8. Many research problems
could, therefore, from the outset better be explicitly formulated in terms of necessary conditions than of probabilistic relations.
The question arises whether or not true determinism does exist, or that there is always an exception to the general rule, which makes
reality probabilistic. Our position in this debate is that if the researcher wants to contribute to Van de Ven’s 1989 idea that “Nothing is quite
so practical as a good theory” he could best have a “pragmatic deter- ministic” view. Pragmatic determinism is the view that it is sometimes
preferable to act as if a complete determinism exists, although it is acknowledged that there might be some exceptions to the assumed
determinism in reality see Box 8.