Deduction and abduction in computer simulation:
Comparing logics in theory development
Prof. Edoardo Mollona
Department of Computer Science and Engineering
Universita degli Studi di Bologna
Mura Anteo Zamboni, 7
40126 Bologna (Italy)
PAPER PRESENTED TO THE 31° SYSTEM DYNAMICS
INTERNATIONAL CONFERENCE
21-25 July 2013
Boston
Abstract
The article presents a review of a sample of simulation studies in the management and
organization body of literature. The proposed approach to the review hinges upon the analysis
of the logic of inference that underpins the selected studies. In particular, I suggest that the two
recurring type of inference that, deliberately or unintentionally, inform the use of simulation
analysis are: deduction and abduction. In addition, the paper proposes an historical journey
into process of diffusion of computer simulation studies within management and organisation
literature. The presented review aims at two goals. First, the paper strives to contribute a point
of view to help researchers to approach the design of a simulation-based study with increased
awareness. Second, the reported analysis ought to help researchers who are not familiar with
simulation to study to appreciate the possible contribution of simulation studies to theory
development.
KEYWORDS: Computer Simulation; Research Methodology; Theory Development.
INTRODUCTION
With different fortunes and oscillating enthusiasm, computer simulation has supported
theoretical investigation in managerial disciplines since the 60's. In the attempt to further
corroborate the role of computer simulation in the repertoire of research strategies available to
social scientists, the aim of the present essay is twofold.
First, I present an historical journey into a selection of contributions to portray the motivations
that fostered the diffusion of computer simulation studies in it and organisation
literature.
Second, differently from received approaches to the review of computer simulation studies,
which mainly aggregate studies on the basis of the adopted simulation technique, my review of
a sample of simulation studies proposes a more subtle point of view to bring forth different
logics of inference that underpin the studies.
To begin with, it is important to set up in the front a definition for computer simulation.
Computer simulation has to do with the manipulation of symbols using a computer code; more
specifically, it uses algorithms to derive propositions from the assumptions that come together
in a computer model. A computer model is a formal model in which ‘[...] the implications of
the assumptions, that is, the conclusions, are derived by allowing an electronic digital computer
to simulate the processes embodied in the assumptions’ (Cohen and Cyert 1961: 115).
In this respect, computer models can be regarded as special cases of mathematical models
(Cohen and Cyert 1961) in which conclusions are derived from assumptions by using a
computer simulation rather than a process of analytical solution. On the other hand, however,
computer models not necessarily have to be stated in mathematical and numerical form
(Clarkson and Simon 1960) since they allow manipulation of symbols that can be words,
phrases and sentences. Therefore, computer models make up the subset of mathematical models
that are solved numerically rather than analytically but not all the computer models are stated in
mathematical terms since they may incorporate not-mathematical symbols. In this respect,
Troitzsch suggests that comp imulation is a third system beside natural language and
mathematics (1998: 27).
In principle, computer simulation is just a technologically-aided process of deduction. Y et, the
crude technology can vary strongly from different approaches and, more importantly, the
difference in the adopted technology often unveils profound differences in the philosophy that
lies beneath modelling.
For example, computer simulations based on systems of difference equations are inspired by a
structuralist stance that sees the behaviour of the individuals that are embedded within a social
system as determined by the feedback nature of the causal relationships that characterize the
system (Forrester 1958, 1961). Agent-Based models or cellular automata, on the other hand,
simulate actions and interactions of autonomous individual entities and build on the idea that the
behaviour of social systems can be modelled and understood as evolving out of interacting
autonomous learning agents (Epstein and Axtell 1996; Axelrod 1997). Thus, a crucial feature of
these models is the emergence of ordered structures independently of top-down planning.
While Agent-Based models and cellular automata show how interaction among individual
decision-making and learning may generate complex aggregate behaviour, differential equation
modelling aims at reducing aggregate and often puzzling behaviours into underlying feedback
causal structures. As a consequence, these latter models typically aggregate agents into a
relatively small number of states, assuming their perfect mixing and homogeneity (Rahmandad
and Sterman 2004) while cellular automata and, especially, Agent-Based models preserve
heterogeneity and individual attributes thereby sacrificing parsimony. The reader looking for an
overview of approaches and techniques may refer to the texts edited by, for example, Liebrand,
Nowak and Troitzsch (1998) or Gilbert and Troitzsch (2005).
However, independently of the approach adopted and the inspiring philosophy, research
employing computer simulation has frequently been regarded, in social sciences, as influenced
by an autonomous logic in respect to mainstream research. Simulation studies, however, have a
long tradition in organizational research. Going back to seminal work in the area of the
behavioural theory of the firm and organizational decision theory, some of the most important
theoretical pieces are based on a simulation approach. This is true, for example, for the well
known Garbage Can model (Cohen, March and Olsen, 1972) and for the work leading to the
development of The Behavioral Theory of the Firm (Cyert, Feigenbaum and March 1959; Cyert
and March 1963). Recent organisation science, as well, has used computer simulation to
advance theory development (Carley and Prietula, 1994; Prietula, Carley and Gasser, 1998).
In recent times, computer simulations, gradually and regularly, have recuperated terrain in
mainstream management and organization journals (Lant and Mezias, 1990; Harrison and
Carroll, 1991; Carley, 1992; Mezias and Glynn, 1993; Carroll and Harrison, 1994, 1998; Lomi
and Larsen, 1996; Sastry, 1997; Adner, 2002; Zott, 2003; Gary, 2005; Lomi, Larsen and Wezel,
2010; Aggarwal, Siggelkow and Singh, 2010).
To push further legitimization of computer simulation in the strategy and organization research,
this essay aims at capturing logical underpinnings of successful simulation work.
The paper is organized as follows; in the next section I briefly pinpoint key milestones in the
history of computer simulation applied to strategy and organization research and, in the
following section, the reasons that motivated early adopters to use computer simulation are
summarized. In section fourth, I consider a sample of recent works that use simulation and I
muse on the differences in the underlying logic of enquiry. In the last section of the chapter I
draw some conclusions.
COMPUTER SIMULATION AND THEORY BUILDING
An increasing number of scholars in social sciences propose that computer simulation proves
useful in supporting theory building.
First, in general, computer simulation may generate inputs in the form of time-series. This may
result of some help when time-series can be compared directly with real-world quantitative
figures, for example demographic data. In this case, the availability of real and simulated time
series that are accessible in a similar quantitative format facilitates pattern-matching thereby
allowing researchers to visually assess resemblance between simulated series, which follows
from the quantitative simulation of a theoretical hypothesis, and an empirically observed
behavior. In this respect, it is possible to generate measures of how simulated events match
empirical instances of those events (Sterman 1984). This is a not trivial opportunity to facilitate
dialogue between micro and aggregate data: in a computer model simulated aggregated data are
rigorously consistent with assumptions describing microbehavior (Bergmann, 1990) and, as a
consequence, they become hypothetical explanations of really observed aggregate time series.
Second, computer simulation allows for a rigorous longitudinal articulation of theorical
behaviors. In other words, the computer-aided process of deduction goes far beyond the human
capability to appreciate the long-term features of the behavior of selected variables. Thus,
computer simulation can support researcher to find plausible sufficient conditions for complex
pattems of behavior to happen such as peaks and lowest point, oscillations with different
characteristics and changes in rates of growth or decline.
Third, researchers, by simulating a formal model, can articulate their predictions by
contemporaneously producing behavior of different variables and the interactions of these latter.
In particular, researchers can simulate the interaction of independent and dependent variables in
each time step, along a given time horizon. This cross-sectional articulation of patterns of
behavior increases the points of contacts between a set of behavioral hypotheses and the
empirical context of a case study. As Kaplan suggests ‘What counts in the validation of a
theory, so far as fitting the facts are concemed, is the convergence of the data brought to bear
upon it [... ]’ (1964: 314). Thus, a computer simulation expands the terrain where comparison
between theory and empirical setting takes place by generating a rich longitudinal and cross-
sectional articulation of behavior under study. In this light, the convergence of data and the
concatenation of events that is necessary to obtain to use a case study to confirm a theory is
increasing demanding.
In this respect, computer simulation aids researchers to design field studies to produce difficult
experiments where the falsifiability of a theory is easier because fitting the facts becomes
increasingly hard. Of course, on the other hand, had empirically collected facts to fit, at least
qualitatively, into a complex web of interweaved simulated behaviors, the experiment would
lead to stronger evidence to confirm propositions contained in the theory.
Having a detailed (often formalized) description of a causal structure, thus, and a description of
a repertoire of plausible histories, a field researcher will have a variety of points in which the
theoretical hypothesis, which crystallized in to the model, can be falsified (Bell and Bell,1980).
The described avenues through which computer simulations aid theory building, however, prove
effective if they are informed by a rigorous logic of inference. How simulation experiments are
designed and described seems often the result of lucky intuitions and ex-post justification more
than the product of a rigorous ex-ante articulation of a research strategy.
THE LOGIC UNDERPINNING THE USE OF COMPUTER SIMULATION
As Cohen and Cyert suggest (1961), computer models are of two types: synthetic and analytic.
In synthetic models, the modeller knows with a high degree of accuracy the behaviour of the
component units of the phenomenon under scrutiny. On the other hand, in analytic models, the
behaviour of the phenomenon is known and the problem is to capture the mechanisms that
produce the behaviour. In this classification, synthetic and analytic models reveal different
underpinning logics of enquiry. While synthetic computer models are informed by a pure
deductive logic, analytic models are characterized by an inductive logic (Cohen 1961).
To start with, however, a word has to be said to better define what we mean by inductive or
deductive inferences. More specifically, the associations synthetic/deductive and
analytic/induction may sound not necessarily intuitive.
Deductive process has been acknowledged as a key component of scientific reasoning since
Aristotle. A deductive inference moves from general assumptions to specific consequences; in
this respect, consequences drawn from assumptions have an inferior degree of universality than
their premises. Deductive inferences have two properties; first, the information embodied in the
deducted consequences is more or less explicitly, included in the assumptions; second, deducted
consequences originate necessarily from assumptions. In other words, if assumptions are
correct, deducted consequences must be correct as well.
On the other hand, inductive processes move from particular instances to general conclusions.
In this respect, in inductive inferences, derived conclusions are not entirely included in the
premises. In other words, the information content in inducted conclusions is greater than the one
crystallized into the premises. Thus, inductive inferences say something new, or different, in
respect to premises; thus, they add information. This property conceals an hazard because
correctness of premises does not necessarily imply that conclusions are correct as well.
As for the distinction between analytic and synthetic, starting from Kant’s Critique of Pure
Reason (firstly published in 1781), an analytic statement is purely explanatory of an existing
concept and it does not add more information than that already contained into the concept itself.
A classic example reported by Kant regards the statement that affirms that an entity of matter is
extended in the space. The fact that an entity of matter is extended in the space is already
implicit in the definition of entity of matter. It does not add information regarding the concept
entity of matter; rather it provides an extension, or further explanation, of the concept. On the
contrary, a synthetic statement is extensive because it adds more information than that contained
originally in a concept. For example, the fact that an entity of matter has a weight, explains
Kant, is not included necessarily in the concept of entity of matter (it suffices to think of a state
of absence of gravity) and rather it stems from a synthesis between an original concept and a
quality external to the concept.
Given this distinction between synthetic and analytic statements, Peirce, for example, put
forward a dichotomy between deductive/analytic and inductive/synthetic inferences (Harshorne
and Weiss 1931/1935).
Thus, we have to be very careful in interpreting the distinction proposed by Cohen and Cyert
between analytic/inductive and synthetic/deductive, since in their framework the concept of
synthesis pertains to the use of simulation to aggregate local, or partial, components of a
phenomenon, into a global emerging behaviour. On the other, analysis concerns the dissection
of behaviour of interest into its components, or determinants.
In addition, to capture the logic underpinning the use of simulation, another type of inference
can be very useful: the abduction.
Abduction, or retroduction, is an inference that goes from the observation of a fact to the
hypothesis of a principle that explains the observed fact (Burks, 1964; Fann, 1970). As Peirce
himself explains (1955), the form of this inference proceeds as follows: “The surprising fact, C,
is observed; But if A were true, C would be a matter of course, Hence, there is reason to suspect
that A is true” (1955: 151).
Suppose that what Peirce calls A is a model. In other words, we have a model as a candidate
theory. This model produces simulated behaviour similar to those empirically observed. Then,
had the world crystallized into the theory to be true, observed patterns of behaviors would be the
reasonable result of an underpinning structure that is isomorphic to the model's structure.
In this perspective, the comparison between an empirical phenomenon and model-generated
behaviours triggers an abductive inference that contributes to the development of an hypothesis
to explain the observed phenomenon.
We suspect that the appropriate logic to address the review of computer modelling in
management and organisation theory would probably focus on the distinction between computer
simulations that adopt a deductive or an abductive logic of inference (Barton and Haslett, 2006).
Within this framework, deductive computer models focus on the specification of a set of
mechanisms or processes and explore unfolding consequences of such specifications whereas
abductive computer models move from the definition of an aggregate behaviour and use
simulation to test whether candidate mechanisms or processes are able to determine in vitro, and
thus explain, the aggregate behaviour.
We agree, however, that simulation studies show a much broader variety of approaches that
blend elements of deduction and abduction. In addition, in computer simulations abduction and
deduction are intertwined in a cyclical process of theoretical investigation. Abduction works
when we introduce in a model a casual mechanism that we deem possibly responsible for an
observed behaviour. In this case, we run history backward to reproduce the conditions for the
behaviour under study to emerge.
On the other hand, when we run a sensitivity analysis, we may be interested in the relationship
between the causal mechanism, or a class of similar causal mechanisms, and a class of
behavioural phenomena. In this case, when we run a computer model and we observe simulated
consequences of changes in f ’ calibration or amendments in the model's structure, we
are embarking into a deductive inference. Thus, deduction and abduction are often tightly
interlaced in a research design based on computer simulation. We thus expect differences
among simulation studies to be detected in the degree of accuracy of the description of the
elements that compose an aggregate phenomenon or of the features that characterize the
aggregate phenomenon itself.
Simulation studies in which a deductive logic of inference prevails will move from the accurate
modelling of those components that are candidate generative mechanisms of behaviours of
interest while simulation studies informed by an abductive logic will set forth from a rich and
detailed description of an aggregate emerging behaviour.
Nonetheless, maintaining two idealtypes of computer models, deductive and abductive, seems a
good strategy, or at least a safe point of departure, to sketch a set of guidelines to analyze and
conceive of a simulation study. An idealtype crystallises what is essential about a phenomenon
(Swedberg, 2005: 119). It helps to interpret hybrid and unclear empirically observed instances
by gauging the relative distance of the observations from the pure form (Thorton, Ocasio and
Lounsbury, 2012: 53). Our approach is justified by the fact that, differently from other typical,
qualitative and quantitative, research strategies that are more legitimized and disciplined,
simulation-based research has been structured in a variety of different guises.
Only recently, Davis, Eisenhardt and Bingham (2007), by developing a roadmap for rigorous
simulation-based research, have convincingly positioned simulation studies among other
methods of enquiry within strategy and organization research. Yet, the field still looks like a
10
rather heterogeneous collection of studies categorised more on the basis of the technique
adopted (System Dynamics or Agent-Based Model, for example) than in respect to
underpinning logic.
Thus, to carry on our avenue, we apply the two idealtypes to capture the often subtle differences
in the logic underlying simulation studies. In the following, we explain how a study may be
classified in one of the two idealtypes. Table 1 reports the qualifying differences among the
studies analyzed.
TABLE 1- HERE
Computer simulation and deductive inference
To address typical features of deductive inference in simulation studies, we begin from the
classic Cohen, March and Olsen’s Garbage Can simulation model (1972). The authors do not
specify in details a reference mode of behaviour to be explained, beyond the broad idea that
they want to address the way in which organized anarchies” undertake decision-making
activity. Rather, the emphasis is on the modelling of the structural features of decision-making
processes in specific types of organizations. The aim is to develop ‘a behavioral theory of
organized anarchy’ (1977: 2). To do so, the authors developed a model that describes decision
making within organized anarchies and examine ‘...the impact of some aspects of
organizational structure on the process of choice...’ (1972: 2). The structure of the research
design encompasses the modelling of organizational decision-making processes and the analysis
of the behavioural consequences of such modelling.
More specifically, the authors adopted a view of an organization as a garbage can in which are
collected ‘... choices looking for problems, issues and feelings looking for decision situations in
which they might be aired, solutions looking for issues to which they might be the answer, and
decision makers looking for work.’ (1972: 2). Along these lines, they modelled problems that
require a specific amount of energy devoted by members of the organization to be solved and
depicted two matrix structures that describe organizational features. The first matrix defines the
access structure that associates choices to problems by determining what choice is accessible to
11
what problem. The second matrix represents the decision structure and associates decision-
makers to choices by establishing what decision maker is eligible to make what choice.
In their experimental design, they portrayed different kind of organizations with different
energy distribution, different problem loads and different organizational structures. Through
simulation experiments, the authors derived emerging decision-making behaviours with typical
features. For example, they observed that, depending of the different assumptions crystallized
into the initial calibration of the model, organizations may show different styles in decision-
making and problem-solving.
We define this type of work deductive since the curiosity that triggers the effort of researchers
regards the deduction of typical emerging patterns of organizational behaviour given the
detailed description of organizational structures and decision-making processes.
Similarly, in their simulation study of entrepreneurial strategies, Lant and Mezias (1990)
formalized an organizational learning model by which firms collect performances, set aspiration
levels, search alternatives and change organizational features. They designed an experimental
setting with a population of 150 firms, to each firm one out of sixteen different organizational
features was assigned, one out of three entrepreneurial strategies, and one out of two levels of
entrepreneurial activity. The research design involves the generation of a number of different
simulations to explore what kind of firm would successfully survive. Through the simulation
experiments, the authors derived longitudinal implications on firms’ performances, growth and
survival and generated theoretical hypotheses on the relationship between entrepreneurial
strategies, levels of entrepreneurial activity and firm performances. In this case, again, the
research design moves off from the description of firms’ decision-making processes and
investigates the consequences of these latter in terms of unfolding behaviours. The study, thus,
maintains a deductive attitude in its interest for the dynamic consequences of a set of
assumptions concerning entrepreneurial strategies as these are built in the specification of the
simulation model. As the authors explain, the data generated ‘... represent implications for
organizational performance, growth, and survival of the different entrepreneurial strategies and
two levels of entrepreneurship’ (1990: 152).
On similar veins, Gavetti and Levinthal (2000) examined the role and interaction between
search processes that are forward-looking, and are based on a cognitive choice, and those that
are backward-looking, and are the consequence of experiential learning. Gavetti and Levinthal
represented the environment as a fitness landscape and modelled two decision-making processes
that are alternatively informed by a backward-looking experiential leaning mechanism or a
forward-looking cognitive mechanism. Experimental design devises a set of simulations in
which performances of the two mechanisms are compared. The experiments allowed the authors
to ascertain that the two mechanisms may productively interact, with the cognitive mechanism
that seeds the experiential leaning mechanism. More precisely, Gavetti and Levinthal explored
the role of the two mechanisms in different experimental conditions. For example, they found
that the more complex the environment, the more accentuated is the role of the cognitive
mechanism in supporting decision-making. In this study, as well as in those before mentioned, a
computer model serves as a virtual laboratory where researchers deduct consequences from
different initial calibrations. The trait that is shared among these studies is that the value added
from simulation is to elicit complex implications that are already hidden into a set of
assumptions. In this respect, the term deductive maintains its attitude to describe an inference
process in which consequences are already contained in the premises. However, this inference
process is far from being an unimaginative or infertile process; on the contrary, researchers, by
connecting premises with their often counterintuitive or surprising consequences, discover
plausible causal relationships among variables that may contribute to theory development. This
active role that simulation can play in theory building motivated Mezias and Glynn to say that
‘[...] simulation results do not simply reflect suppositions built in the model, but yield
knowledge that adds value beyond its explicit assumptions’ (1993: 95).
Cc i ion and abductive inference
As we assumed in this work, researchers adopt an abductive inference when they proceed from
an aggregate phenomenon, more specifically, from the description of a behaviour that unfolds
longitudinally over time, and use computer simulation to select plausible determinants of the
phenomenon among alternative causal mechanisms.
For example, Adner (2002) studied the emergence of disruptive technologies and set up his
research design by stating at the front the description of the characteristics of the phenomenon
he wanted to investigate. After clarifying that his contribution is to explain the emergence of
disruptive technologies, Adner modelled consumers’ individual preferences and firm
technological strategy to obtain mechanisms that are sufficient to produce the previously
described phenomenon.
A similar logic inspires the work of Lee, Lee and Rho (2002) that conceived of their research
design with the aim at explaining the emergence of strategic groups. They developed a number
of theoretical hypotheses that define causal relationships among four explanatory mechanisms
and strategic groups’ emergence, persistence and differential performances. They modelled a
population of 50 firms and a pay-off function with two peaks (a global maximum and local
maximum). Adopting an evolutionary framework, they built a genetic algorithm that mimics a
process of variation (innovation in strategy), a process of selection (payoff received) and a
process of retention (imitation of successful strategies by new entrants). They run experiments
varying each of the four mechanisms at time and examined under what conditions strategic
groups are likely to emerge and persist.
Another study with similar features is Abrahamson and Rosenkopf’s analysis of the emergence
of bandwagon in innovation adoption (1993). They defined the phenomenon of interest and
used computer simulation to find sufficient conditions for bandwagon to emerge and for
innovations to be retained by adopters after bandwagons have displayed their effects. More
precisely, they modelled bandwagons and derived behaviour with simulation to understand how
causal structures of the model, and the processes that the causal structures represent, contributed
14
to produce the dynamic behaviour observed in the simulation experiments. Grounding on the
observed cause-effect relationship, they derived propositions about bandwagon occurrence,
extent, persistence.
Similar logic of enquiry informs Lant and Mezias’ speculation on modes of organizational
change (1992). They set off their research design from the definition of a dynamic behaviour of
interest: Tushman and Romanelli theory of punctuated model of organizational change (1985).
Afterward, they scrutinized candidate causal mechanisms to ferret out determinants of the
behaviour of interest. In particular, they formalized an organizational learning model by which
firms collect performances, set aspiration levels, search alternatives and change organizational
features. Then, they used computer simulation to build a population of firms whose activities
are governed by this process of experiential learning and demonstrated that an organizational
change process that is informed by this learning mechanism can unfold displaying the typical
pattem of punctuated change. Using computer simulation, they theorized that the same deep
theoretical structure, in this case a learning mechanism, underpins both convergence and
reorientation processes.
The explanation of the punctuated model of organizational change is at the core of Sastry’s
simulation study as well (1997). Sastry analyzed Tushman and Romanelli’s verbal theory of
punctuated change to demonstrate that the verbal theory does not contain the necessary causal
mechanisms to explain the described behaviour. Sastry conducted a textual analysis of the
verbal theory and used qualitative descriptions to produce a formal model that encapsulates the
theory. She identified constructs and causal relationships that provided the basis of the formal
model.
Once a computer model that formalized key traits of the theory was built, Sastry simulated the
model and compared simulated behaviour with the one crystallized into the theory. The
discrepancy between theoretical and simulated behaviours guided Sastry to introduce two new
mechanisms that were not originally included into the verbal theory but that proved necessary to
produce the behaviour purported in the theory.
The two mechanisms are, respectively, a routine for monitoring organization-environment
consistency and a heuristic that suspends change for a trial period following each reorientation.
The work of Sastry provides the opportunity to speculate further on the features of abductive
simulation research. As we said in the foregoing, typically, abductive inferences bring about
additional information that is not necessarily crystallized into the premises.
The abductive nature of the study of Sastry emerges when we appreciate that in the original
premises of the study, which are captured in Tushman and Romanelli’s verbal theory, there was
not mention or any sort of indication that pointed at the causal mechanisms that Sastry included
into the theory ex-post.
To clarify the position taken in this essay, however, when I suggest that abductive simulations
bring in a study information content that is not included in the stated premises, I am suggesting
that, given a set of initial premises, a simulation study has an abductive nature when it facilitates
the enlargement or the modification of this set of premises.
In this vein, Malerba, Nelson, Orsenigo and Winter (1999) propose a class of computer models
that they define history friendly because of the adherence of these latter to the empirical realm
that is the object of exploration. In their simulation study, they focused on an appreciative
theory that describes the pattern of evolution of the computer industry and developed a formal
representation of that theory. Through simulation, they checked the consistency between the
appreciative and the formal version of the theory by examining whether the formal version is
able to reproduce the same stylized facts as described in the appreciative theory.
The empirically observed behaviour is the pedestal to build the computer model and the
contribution of the simulation study is one of generating a repertoire of plausible causal
mechanisms that might explain behaviours as observed in the real world.
What about induction?
In this vein, another example of abduction is provided by the study of Lomi and Larsen (1996)
on population ecology. They focused on the typical model of density-dependent founding and
16
mortality rates and addressed the micro-processes that take place at the level of individual
organizations. The authors modelled micro-processes of local interaction and simulated
emerging competitive dynamics of organizational populations. They designed a protocol of
simulation experiments through which they tested different specifications of local micro-
processes. For example, they varied the strength of the link between founding decision and local
density. Then, they used data generated by the simulations to estimate a model of organizational
founding and compared simulated estimates with existing population-level empirical estimates.
They demonstrated the ecological model of density-dependent founding rates to be consistent
with a number of micro-assumptions about the patterns and the range of local interaction among
individual organizations. Again, in this case, the study maintains an abductive flavour in its
using computer simulation to include plausible premises, a set of behavioural micro-
assumptions, to the repertoire of possible explanations of observed aggregate behaviours.
The interplay of abduction and deduction in si ion studies
A consideration is fundamental in order not to misinterpret the distinction between deductive
and abductive simulation studies. In most of the simulation studies in social sciences, abductive
and deductive inferences are intertwined. However, we cannot avoid noting that the logic by
which they are inspired often differs not marginally. It is in this light that is justified our
strategy of adopting idealtypes as analytical abstractions.
For example, in the mentioned study of Sastry, the logic of enquiry is clearly stated and hinges
upon two elements. First, the author has a clear imagine of the dynamic features of the
behaviour she wants to explain. Second, she uses the comparison between theoretical and
simulated behaviour as a trigger to import in her modelling candidate causal mechanisms.
On the other hand, at the other extreme, consider, for example, Cohen, March and Olsen's
Garbage Can simulation model. The authors described how problems, choices and people met
within an organization but they start their enquiry without a precise idea about the aggregate
decision-making behaviour that follows from the premises they designed. The curiosity was
17
exactly to understand what the consequences are of representing an organization as an organized
anarchy and the contribution of the study is indeed to suggest that organized anarchies maintain
a peculiar style in their decision making behaviour. Many simulation studies, however, blend
the two components.
For example, in his study on the emergence of disruptive technologies (2002), Adner proceeded
from the description of the phenomenon of the emergence of disruptive technologies. He
investigated how the phenomenon had been analyzed before in the literature and noticed that
previous explanations had focused on the limits of incumbent technologies. Taking a different
angle, Adner focused on the impact of market demand on development strategies. This choice
directed his attention on the modelling of the structure of market demand and, more
importantly, led him to introduce two new constructs, preference overlap and preference
symmetry, to capture features of market demand that are connected to the behaviour of interest.
In addition, however, the deduction, through simulation, of the consequences of modelled
premises led him to produce a repertoire of plausible behaviours depending on changes applied
to the calibration of the simulation model. Through this exercise of deduction the author
provided an articulated portray of the phenomenon under study, eliciting different modes of
competition among technologies. In this case, abductive inference, starting from a defined
behaviour, aided the elicitation of sufficient causal mechanisms to observe the behaviour
whereas deductive inference expanded knowledge of the behaviour by producing various
simulated scenarios.
Beside cases in which the abductive or deductive approaches clearly come into view, many
studies incorporate both approaches. A simulation study may incorporate a loosely defined idea
of the features of the behaviour it is aimed to explain and this idea guides the modelling of the
premises. The deduction of consequences from premises through computer simulation aids the
refinement of the description of the behaviour of interest. On the other hand, the materialisation
of surprising or counterintuitive behaviours induces the search for alternative causal
mechanisms to modify the original set of premises.
Deduction generates repertoires of patterns of behaviour that represent near-histories that
proceed from a common deep causal structure (March, Sproull and Tamuz 1991). This exercise
contributes to theory building by making available ex-ante falsifiable hypotheses that connect
casual mechanisms to behaviours. Deduction may also create counterintuitive and surprising
behaviours that bring about marginal amendments in the modelling of the premises or may
trigger revisions of modelled premises. In this case, the discrepancy between expected and
simulated behaviour is the incentive to refine, or deeply modify, the modelled set of premises by
introducing in the model new causal mechanisms.
For example, in their study on population ecology and competition among structurally different
populations of organizations, Carroll and Harrison (1994) built a mathematical model, designed
a structurally superior population and simulated competition between two populations (one
inferior and one superior). Through the simulation study, they demonstrated, in vitro, that the
dominance of structurally superior populations may not emerge depending on their timing of
entry in the industry. The contribution of this theoretical falsification is to delineate the
hypothesis of historical inefficiency, according to which the explanation of an observed
behaviour is history-dependent and the time in which events happens modify their expected
consequences.
In this case, given the modelled premises, the deduction of a behaviour that is incongruous with
the expected one, facilitates an abductive inference leading to the engendering of the concept of
historical inefficiency.
CONCLUSION
The proposed review suggests that simulation studies are gaining legitimization in management
and organization literature. Nevertheless, few are the reviews that address the role played by this
research approach in the mentioned disciplinary fields. More importantly, these reviews are
often focused on the technical differences that characterize alternative simulation approaches.
Times are mature to propose new theoretical lenses to review the field. Of course, I am aware
19
that different simulation techniques are differently capable to tackling specific problems. In this
respect, I am not downsizing the importance of reviews that portray the repertoire of available
simulation techniques. Yet, I am convinced that an intriguing perspective to appreciate the
variety of contributions and to guide future research is to capture the often subtle difference in
the logic of inference that underpin simulation studies.
In so doing, I focused on two forms of inference: deduction and abduction.
These two forms of inference, it is advocated in the paper, are those more frequently used to
structure a simulation study. I expect, however, that it is unlikely that authors build their
research strategy by consciously making reference to a specific form of inference. More often
the reference is made retrospectively.
Moreover, often, deductive and abductive inferences are so tightly interlaced as to make it an
arduous endeavour to disentangle the two processes at work.
More frustrating, the profound interplay between the two kind of inferences may push scholars
to dismiss the attempt to separate the working of the two mechanisms as an unnecessary
subtlety.
I maintain, however, that the ability to discern the difference between deductive and abductive
inferences scores two important goals.
First, by having in sight the difference between the two logics of inference, researchers more
deliberately can design the structure of their simulation studies. Second, endowed with a
sophisticated point of view, colleagues may more consciously read and appreciate the
contribution that is brought about by computer simulation thereby discriminating between
studies grounded upon a solid logic and work in which the aims and the purpose of using
computer simulation remains obscure.
Technically speaking, a computer simulation cannot be anything different than a computer-
aided process of deduction. This deduction process both unveils not necessarily intuitive cause-
effect relationships that are implicitly hidden in the premises and assists rigorous articulation of
appreciative theories. This facilitates researchers in producing testable hypotheses. On the other
20
hand, when deducted behaviours do not match with expectations, this mismatch activates an
abductive inference that amends the original set of premises.
Of course, this article shows a limit in its considering only a portion of the simulation studies
that populate the management and organization literature. The selection from a rich repertoire of
pieces of work was guided by the intention of highlighting the gradual acceptance of simulation
studies in mainstream journals. Thus, I preferred articles written on these latter rather than those
written on journals more prone to accept simulation studies. Also, this kind of articles, given
their need to legitimize their publication on mainstream journals, generally devote more space
and effort to clarify their methodology.
REFERENCES
Abrahamson, E. and Rosenkopf, L. (1993). ‘Institutional and competitive bandwagon: Using
mathematical modelling as a tool to explore innovation diffusion’. Academy of Management
Journal, 18 (3), 487-517.
Adner, R. (2002). ‘When Are Technologies Disruptive? A Demand-Based View of the
Emergence of Competition’. Strategic Management J ournal, 23 (8), 667-88.
Aggarwal, V. A., Siggelkow, N. and Singh, H. (2010). ‘Governing collaborative activity:
Interdipendence and the impact of coordination and exploration’. Strategic Management
Journal, 32(7), 705-730.
Axelrod, R. (1984). The Evolution of Cooperation. New Y ork: Basic Books.
Axelrod, R. (1997). The Complexity of Cooperation: Agent-Based Models of Competition and
Collaboration. Princeton, N.J: Princeton University Press.
21
Barton, J. and Haslett, T. 2006. Fresh Insights into System Dynamics Methodology- Developing
an abductive inference perspective. Paper presented at the 24" International Conference of the
System Dynamics Society, Nijmegen, The Netherlands, July 23" - 27", 2006
Bell, J. A. and J. F. Bell. 1980. System Dynamics and Scientific Method. In Randers, J. (Ed.)
Elements of the System Dynamics Method. Cambridge, MA and Norwalk, CT: Productivity
Press.
Bergmann, B. R. 1990. Micro-to-Macro Simulation: A Primer With a Labor Market Example.
Journal of Economic Perspectives, 4(1): 99-116.
Bowman, H.R., Fetter, R.B. (1957). Analysis for Production Management. Homewood, Ill.
Burks, A. W. (1964). ‘Peirce’s Theory of Abduction’. Philosophy of Science, 13, 301-306.
Carley, K. M. (1992). ‘Organizational learning and personnel turnover’. Organization Science,
3(1), 20-46.
Carley, K. M. and Prietula, M. J. (Eds.). (1994). Computational Organization Theory. Hillsdale,
NJ: Lawrence Erlbaum Associates.
Carroll, G.R. and Harrison, J.R. (1994). ‘On the historical efficiency of competition between
organizational populations’. The American Journal of Sociology, 100 (3) pp. 720-49.
Churchman, C.W., Ackoff, R.L. and Amoff, E.L. (1957). Introduction to Operations Research,
New Y ork: John Wiley.
22
Clarkson, G.P.E. and Simon, H.A. (1960). ‘Simulation of Individual and Group Behavior’. The
American Economic Review, 50 (5), 920-32.
Coe, R.M. (1964). ‘Conflict, Interference, and Aggression: Computer Simulation of a Social
Process’. Behavioral Science, 9 (2), 186-97.
Cohen, KJ. (1960a). Simulation of the Firm. The American Economic Review. 50 (2), Papers
and Proceedings of the Seventy-second Annual Meeting of the American Economic
Association, pp. 534 -40.
Cohen, K.J. (1960b). Computer Models of the Shoe, Leather, Hide Sequence. Prentice-Hall:
Englewood Cliffs, N.J.
Cohen, K J. (1961). ‘Two Approaches to Computer Simulation’. The Journal of the Academy of
Management, 4 (1), 43-9.
Cohen, K. and Cyert, R.M. (1961). ‘Computer Models in Dynamic Economics’. The Quarterly
Journal of Economics, 75 (1), 112-27.
Cohen, M.D., March, J.G. and Olsen, H.P. (1972). ‘A garbage can model of organizational
choice’. Administrative Science Quarterly, 17 (1), 1-25.
Cyert, R.M., Feigenbaum, E.A. and March, J.G. (1959). ‘Models in a Behavioral theory of the
firm’. Behavioral Science, 4 (2), 1-95.
Cyert, R.M. and March, J.G. (1963). A Behavioral Theory of the Firm. Englewood Cliffs: NJ
Prentice-Hall.
23
Davis, J.P., Eisenhardt, K.M. and Bingham, C.B. (2007). ‘Developing theory through
simulation methods’. Academy of Management Review, 32 (2), 480-99.
Dorfman, R. (1960). ‘Operations Research’. American Economic Review, 50 (4) 575-623.
Edmonds, B., Hemandez, C. and Troitzsch, K. (eds) (2007). Social Simulation Technologies,
Advances and New Discoveries. IGI Publishing.
Eguiluz, V., Zimmermann, M.G., Cela-Conde, C.J. and San Miguel, M. (2005). ‘Cooperation
and the e of role dif iation in the dynamics of social networks’. American Journal
of Sociology, 110 (4), 977-1008.
Epstein, J.M., Axtell, R. (1996). Growing Artificial Societies: Social Science From the Bottom
Up. Washington D.C: Brookings Institution.
Fann, K. T. 1970. Peirce’s Theory of Abduction. Nijhoff: The Hague.
Forrester, J.W. (1958). ‘Industrial Dynamics-A Major Breakthrough for Decision Makers’.
Harvard Business Review, 36 (4), 37-66.
Forrester, J.W. (1961). Industrial Dynamics. Cambridge, MA: The MIT Press.
Forrester, J.W. (1968). ‘Market growth as influenced by capital investments’. Industrial
Management Review, 9 (2), 83-105.
24
Gary, M. S. (2005). ‘Implementation strategy and performance outcomes in related
diversification’. Strategic Management Journal, 26(7), 643-664.
Gavetti, G., Levinthal, D. (2000). ‘Looking forward and looking backward: Cognitive and
experiential learning’. Administrative Science Quarterly, 45 (1), 113-37.
Gavetti, G., Levinthal, D.A. and Rivkin, J.W. (2005). ‘Strategy making in novel and complex
worlds: the power of analogy’. Strategic Management J ournal, 26 (8), 691-712.
Gilbert, G.N. and Doran, J.E. (Eds.) (1994). Simulating societies: The computer simulation of
social phenomena. London: UCL Press.
Gilbert, G.N., Conte, R. (Eds.) (1995). Artificial societies: The computer simulation of social
life. London: UCL Press.
Gilbert, N., Troitzch, K.G. (eds.) (2005). Simulation for the Social Scientist. London: Open
University Press.
Gullahorn, J.T., Gullahom, J.E. (1963). A Computer Model of Elementary Social Behavior. In
Feigenbaum, E. A. and Feldman, J. (eds.), Computer and Thought, New York: McGraw-Hill,
pp. 375-86.
Hanaki, N., Peterhansl, A., Dodds, P.S. and Watts, DJ. (2007). ‘Cooperating in evolving social
networks’. Management Science, 53 (7), 1243-48.
25
Hanneman, R.A., Collins, R. and Mordt, G. (1995). ‘Discovering theory dynamics by computer:
experiments on state legitimacy and imperialist capitalism’. Sociological Methodology, 25, 1-
46.
Harrison, J. R. and Carroll, G. R. (1991). ‘Keeping the faith: A model of cultural transmission in
formal organizations’. Administrative Science Quarterly, 36(4), 552-582.
Harshorne, C., Weiss, P. (1931/1935). Collected papers of Charles Sanders Peirce. Vol. II: 374,
Cambridge, MA: Harvard University Press.
Hoggatt, A.C. (1957). Simulation of the Firm. Research Paper RC-16, IBM Research Center,
Poughkeepsie, N.Y .
Homan, G.C. (1950). The Human Group, New Y ork: Harcourt.
Huff, J.0., Huff, A.S. and Thomas, H. (1992). ‘Strategic renewal and interaction of cumulative
stress and inertia’. Strategic Management Journal, 13 (S1), 55-75.
Kaplan, A. 1964. The Conduct of Inquiry, San Francisco, CA: Chandler Publishing Company.
Klein, L.R. (1953). A Textbook of Econometrics, Evanston, Ill: Row, Peterson & Co..
Lant, T.K. and Mezias, S.J. (1990). ‘Managing discontinuous change: a simulation study of
organizational learning and entrepreneurship’. Strategic Management Journal, 11 (2), 147-79.
26
Lant, T.K. and Mezias, SJ. (1992). ‘An Organizational Leaning Model of Convergence and
Reorientation’. Organization Science, 3 (1), 47-71.
Lee, J., Lee, K. and Rho, S. (2002). ‘An Evolutionary Perspective on Strategic Group
Emergence: A Genetic Algorithm-Based Model’. Strategic Management Journal, 23 (8), 727-
46.
Liebrand, W.B.G., Nowak, A. and Hegselmann, R. (eds.) (2007). Computer Modeling of Social
Processes. London, UK: Sage Publications.
Lomborg, B. (1996). ‘Nucleus and shield: the evolution of social structure in the iterated
prisoner's dilemma’. American Sociological Review, 61 (2), 278-307.
Lomi, A., Larsen, E. R. (1996). ‘Interacting locally and evolving globally: A computational
approach to the dynamics of organizational populations’. The Academy of Management Journal,
39(5), 1287-1321.
Lomi, A., Larsen, E. R., and Wezel, F. C. ‘Getting there: Exploring the role of expectations and
preproduction delays in processes of organizational founding’. Organization Science, 21(1),
132-149.
Macy, M., and Skvoretz, J. (1998). ‘The evolution of trust and cooperation between strangers: A
computational model’. American Sociological Review, 63 (5), 638-60.
March, J.G., Sproull, L.S. and Tamuz, M. (1991). ‘Learning from a sample of one or fewer’.
Organization Science, 2 (1), 1-13.
27
Malerba, F., Nelson, R., Orsenigo, L. and Winter, S. (1999). ‘History-friendly' models of
industry evolution: the computer industry’. Industrial and Corporate Change, 8 (1), 3-40.
McPhee, W.N. (1961). ‘A Note on a Campaign Simulator’. Public Opinion Quarterly, 25(2),
184-93.
Meinhart, W.A. (1966). ‘Artificial Intelligence, Computer Simulation of Human Cognitive and
Social Processes, and Management Thought’. The Academy of Management Journal, 9 (4),
294-307.
Mezias, S.J. and Glynn, M.A. (1993). ‘The three faces of corporate renewal: Institution,
Revolution, and Evolution’. Strategic Management Journal, 14 (2), 77-101.
Morecroft, J.D.W. (1983). ‘System Dynamics: Portraying Bounded Rationality’. Omega, The
International J ournal of Management Science, 11 (2), 131-42.
Nettle, M. and Dunbar, R.I.M. (1997). ‘Social markers and the evolution of reciprocal
exchange’. Current Anthropology, 38 (1), 93-9.
Orcutt, G.H. (1960). ‘Simulation of Economic Systems’. The American Economic Review, 50
(5), 893-907.
Orcutt, G.H., Greenberger, M. and Rivlin, A.M. (1958). Decision-Unit Models and Simulation
of the United States Economy. Mimeo, Harvard University.
Peirce, C. S. 1955. Abduction and Induction. In Buchler, J. (Ed.) Philosophical Writings of
Peirce, New Y ork, NY: Dover Publications Inc.
28
Prietula, M. J., Carley, K. M. and L. Gasser (Eds.). (1998). Simulating Organizations:
Computational Models of Institutions and Groups. Menlo Park, CA: AAAI Press / The MIT
Press.
Rahmandad, H., Sterman, J. (2004). Heterogeneity and Network Structure in the Dynamics of
Diffusion: Comparing Agent-Based and Differential Equations Models. WP n° ESD-WP-2004-
05, Massachusetts Institute of Technology Engineering Systems Division Working Paper Series,
Cambridge, MA.
Sastry, M. A. (1997). ‘Problems and paradoxes in a model of punctuated organizational
change’. Administrative Science Quarterly, 42(2), 237-275.
Shubik, M. (1960). ‘Simulation of the industry and the firm’. The American Economic Review,
50 (5), 908-19.
Simon, H.A. (1952). ‘A formal theory of interaction in social groups’. American Sociological
Review, 17 (2), 202-211.
Sterman, J.D. (1984). Appropriate summary statistics for evaluating the historical fit of system
dynamics models. Dynamica, 10: 51-66.
Swedberg,R. (2005). The Max Weber Dictionary: Key Words and Central Concepts. Stanford,
CA: Stanford University Press.
Thorton, P. H., Ocasio, W. And Lounsbury, M. (2012). The Institutional Logics Pespective.
Oxford, UK: Oxford University Press.
29
Troitzsch, K.G., Mueller, U., Gilbert, G.N. and Doran, J.E. (Eds.) (1996). Social science
microsimulation. Berlin: Springer.
Troitzsch, K.G. (1998). Multilevel process modelling in the social sciences: Mathematical
analysis and computer simulation. In Liebrand, W.B.G., Nowak, A. and Hegselmann, R. (Eds.),
Computer Modeling of Social Processes, London, UK: Sage Publications.
Tushman, M. L., Romanelli, E. (1985). Organizational Evolution: A Metamorphosis Model of
Convergence and Reorientation. In Cummings, L. L. and Staw, B. M. (eds.), Research in
Organizational Behavior, Vol. 7, Greenwich, CT: JAI Press, pp. 171-222.
Vazsonyi, A. (1958). Scientific Programming in Business and Industry. New Y ork: John Wiley.
Zott, C. (2003). ‘Dynamic capabilities and the emergence of intraindustry differential firm
performance: Insights from a simulation study’. Strategic Management Journal, 24(2), 97-125.
NOTES
(1) Dorfman (1960: 603) recommends that computer simulation is particularly useful in the
area of general systems analysis and in problems that involve inventory and queuing
management. In particular, Dorfman explains that the operation researcher tries to
simplify ’ ... his problems as much as he dares (sometimes more than he should dare),
applies the most powerful analytic tools at his command and, with luck, just squeaks
through. But what if all established methods fail, either because the problem cannot be
forced into one of the standard types or because, after all acceptable simplifications, it is
30
still so large or complicated that the equations describing it cannot be solved? When he
finds himself in this fix, the operations analyst falls back on "simulation" or "gaming."’.
(2) In Cohen, March and Olsen’s Garbage Can model, organized anarchies are
characterized by three general properties: problematic preferences (inconsistent and ill-
defined set of preferences), unclear technology and fluid participation.
TABLE 1
Generative mechanisms Emerging behavior
Cohen, March Organized Anarchy
& Olsen, 1972 Problematic preferences Aggregate behavioral
Unclear technology implications
Fluid participation
Lant & Mezias, 1990
Entrepreneurial strategies
Firms’ performances, growth
&survival
Gavetti & Levinthal, 2000
Backward-looking experiential
Survival in different fitness
learning landscapes
Forward-looking cognitive
Ader, 2002 1) Preference overlap Technology adoption
2) Preference asymmetry
Lee, Lee & Rho, 2002 1) Mobility baniers Emergence and persistence
2). Strategic interactions
3) Rivalry across firms
4) Dynamic capabilities
ofstratesic groups
Sastry, 1997
1) Monitoring Routine
2). Trial Period suspending
heuristic
Panetuated change
31
