The 29th International Conference of the System Dynamics Society
July 24 — 28, 2011 - Washington, DC
Major Issues in Mixed Use of
Grounded Theory and System Dynamics A pproaches
in Qualitative Secondary Data
Bahadir K. Akcam!
Senem Guney?
Anthony M. Cresswell?
Abstract: This article explores the methodological issues in mixed use of Grounded Theory and
System Dynamics approaches in a research project. We discuss how analysts dealing with
qualitative secondary data in the conduct of System Dynamics Modeling work through the
questions of the role of existing literature and generic structures in system dynamics, directions
of research (inductive, deductive, or abductive), the mixed method’s ability to extract system
dynamics modeling information, and potential outputs of such research. These discussions are
based on an analysis from an empirical research project. The article describes a research design
and suggests next steps to develop a coherent analytical technique for researchers.
Keywords: Grounded Theory, System Dynamics, Abductive Approach, Mixed Methods,
Generic Structures, Archetypes
' Westem New England University, Springfield, MA
Corresponding Author:
Bahadir K. Akcam, Assistant Professor of Business Information Systems, Western New England University,
College of Business, 1215 Wilbraham Road, Springfield MA, 01119
? Department of Informatics and Center for Technology in Government, University at Albany, Albany, NY
3 Center for Technology in Government, University at Albany, Albany, NY
1. Introduction
A longtime dispute between qualitative and quantitative researchers has produced a large body of
material debating the merits and disadvantages of these approaches (Weber, 2004). As these
discussions contributed common understanding, some researchers integrated qualitative and
quantitative methods under the name of mixed methods since 1980s (Brewer & Hunter, 1989;
Johnson & Onwuegbuzie, 2004). But System Dynamics as a methodology that uses qualitative
data to develop quantitative simulation models has not been a significant part of these method
oriented discussions. Since its introduction in mid-1950s, the System Dynamics approach has
used qualitative data to study complex social systems. Most of the time, the end product of this
approach is a mathematical model that describes the structure and behavior of the complex
system quantitatively. Despite the central role of qualitative data in system dynamics model
development process, the System Dynamics field does not have detailed protocols to describe the
use of qualitative data or qualitative research methods in the modeling process (Luna-Reyes &
Andersen, 2003). In a response to this gap, this article explores qualitative data and research
methodology issues in System Dynamics modeling based on an analysis from a research project
that adopted the Grounded Theory approach.
The mixed use of Grounded Theory and System Dynamics approaches raises a number of
methodological issues. One set of issues involves how to combine the use of existing literature
and generic structures in a research project. This issue is closely related with the direction of
research (inductive, deductive, or abductive). Direction of research is important to understand the
role of existing literature (theories and preexisting concepts) in the Grounded Theory approach.
While there are conflicting views about introducing fundamental concepts early in Grounded
Theory development (Glaser, 1978; Kelle, 2005), System Dynamics researchers are very
enthusiastic about the role of generic structures in exploring complex systems (Forrester, 1998;
Wolstenholme, 2003). Another issue concerns how to go about extracting necessary information
from qualitative datasets to build quantitative models. System Dynamics researchers look for
causal relationships to understand system structure and reference modes (behavior of variables
overtime) to understand system behavior. This information is extracted from research subjects or
gathered from other resources. System Dynamics models are being grounded on these findings,
which are most of the time qualitative data. The possibility of improved methods for collecting
system related information from qualitative datasets is an important challenge for the
researchers. The final issue is about the output of such a research project. The output
expectations from mixing Grounded Theory and System Dynamics are focused on System
Dynamics artifacts. Fully developed simulation models and causal loop diagrams are such
artifacts that explain the phenomenon in question and articulate different hypotheses all
grounded on qualitative data.
Based on the discussion of these issues, a research approach will be described to address the
need of protocols to combine these approaches. Our intention is to provide a detailed protocol in
the mixed use of Grounded Theory and System Dynamics that has long been missing (Luna-
Reyes & Andersen, 2003).
This article is based in large part on an earlier study of responses to the attack on the World
Trade Center on September 11, 2001. This earlier study will be referred as “the WTC research”.
The WTC research (Akcam, 2009) extends and elaborates a generic dynamic theory (Luna-
Reyes et al., 2004) by using Grounded Theory approach. It explores the socio-technical
processes in an interorganizational collaboration by exploring a generic dynamic theory in more
depth. The generic dynamic theory offers dynamic hypotheses about causal relationships
between socio-technical processes and social accumulations derived from a study of interagency
information integration initiative among New York state agencies—focusing in the WTC
research on the response and recovery process following the World Trade Center (WTC) attack,
which involved an enormous amount of interagency collaboration in response to a very tragic
event.
The data from this case are used to model the socio-technical processes of collaboration in IT
development and use by using the theoretical lenses of the generic dynamic theory described by
Luna-Reyes et al. (2004). Researchers at the Center for Technology in Government (CTG) at the
University at Albany-SUNY interviewed 29 responders over a ten-month period in 2002-2003.
These responders held positions at critical decision-making points in the response and recovery
process. This interview dataset covers rich stories of interagency collaboration in the context of
information, technology, and coordination.
All the issues identified in this article emerged from the interaction between parts of the WTC
research. In summary, the WTC research aims to extend a generic dynamic theory. Having a rich
qualitative dataset encouraged researchers to use it for this purpose. But clear-cut data analysis
protocols were not available to apply to this dataset to gather necessary system dynamics
information to extend and elaborate the generic dynamic theory. The System Dynamics
community developed methods to elicit modeling information from mental database through
several interactive techniques with subjects, but other methods were limited with the written
database, specifically with the qualitative datasets such as interview transcripts as secondary
data. Grounded Theory approach among qualitative methods was selected in the WTC research,
because the interviews were based on semi-structured questions.
The issues identified in this article are from the perspective of secondary data analysis of a
qualitative dataset (interview transcripts in this case). This point is very important, since other
qualitative methods such as interviewing (Luna-Reyes, Diker, & Andersen, 2005) with subjects
can give researchers ability to elicit necessary modeling information directly from them. But in a
secondary dataset, researcher interacts with transcripts of interviews. Most of the time
interviewers from other backgrounds do not do these interviews with the system dynamics
modeling information in mind. So extracting necessary information to gain systems insight and
producing System Dynamics artifacts such as reference modes, causal loop diagrams, and fully
developed models are becoming more challenged. System Dynamics researchers faced similar
challenges by using grounded theory approach to analyze qualitative data and theories to
generate new theories (Black, Carlile, & Repenning, 2004; Rudolph & Repenning, 2002).
‘Qualitative Data and Analysis in System Dynamics’ section of this article explores data types,
qualitative data, and qualitative data analysis methods in System Dynamics field. The following
three sections discuss main issues in mixing System Dynamics and Grounded Theory in
qualitative secondary data. After exploring the main issues, the article ends with a research
design suggestion.
2. Qualitative Data and Analysis in System Dynamics
In his discussion of information sources in modeling, Forrester (1980, 1991) defines three
categories; mental data base, written data base and numerical data base. Forrester uses data bases
in an extended sense. In his description of data, he references its dictionary meaning and
indicates that data is “something that is given from being experientially encountered”, “material
serving as a basis for discussion, inference, or determination of policy” and “detailed
information of any kind” (Forrester, 1991, p. 23). This description of data is far broader than in
its common usage as numerical data.
In Figure 1, Forrester compares the sizes of information content of different information sources.
The mental data base is far more extensive than the other information sources. While he indicates
the importance of mental data base, he also discusses that the significance of information
residing in mental data base is not adequately appreciated in the social sciences (Forrester, 1980).
For modeling purposes, mental data base’s content consists of observations about structure and
policies, expectations about system behavior, and actual observed system behavior (Figure 2).
Mental
data base
Written
data base
Numerical
data base
Figure 1- Decreasing information content in moving from mental to written to numerical data bases
(FORRESTER, 1980, p. 556, 1991, p.23)
Observations
about structure
and policies
Expectations
about |
system behavior
Actual observed
system behavior
Figure 2 - Content of the mental data base as related to components and to behavior of a social system.
(Forrester, 1980, p. 556)
While the size of a written data base is less than mental data base, some part of the written data
base is a recording of mental data base. Other part of the written data base contains concepts and
abstractions that interpret other information sources. Forrester (Forrester, 1980) finds daily and
weekly, public and business press more important than textbooks, journals and professional
literature, because public and business press have more capability to reflect current pressures
surrounding decisions. For a system dynamics modeler, it is important to understand those daily
pressures in order reveal the behavior of systems more accurately. Forrester (1980) also indicates
the importance of abstractions about system structure. He exemplifies Cobb-Dougles function to
discuss the contribution of such abstractions into the structure of system dynamics models.
In his description of shortcomings of written records, Forrester indicates that an author filters
information from his or her perspective and purposes, while he or she is transforming mental
information into written information .Another shortcoming is, “unlike the mental database, the
written record is not responsive to probing by the analyst as he or she searches for a fit between
structure, policy, and behavior” (Forrester, 1980, p. 557).
A numerical data base has the narrowest scope in information sources. The structure and policies
that created the data are missing in numerical data. The cause and effect directions among
variables cannot be extracted from the numerical data. Numerical data base’s contribution to
system dynamics models can be categorized as parameter values necessary for variables,
summarized characteristics of system behaviors in professional literature, and time series
information for comparing model output rather than determining model parameters.
Although numeric information may be seen as very important to build such models, most of the
time information available to modelers is in qualitative nature. Actually Forrester (1991) finds
qualitative data residing in people’s heads more important than the quantitative data. He (1991,
p. 5) discusses that despite qualitative information’s importance, management and social
scientists have long been neglected this “far richer and more informative body of information
that exist in the knowledge and experience of those in the active, working world.”
It is not easy to elicit the wealth of information that people carry in their heads. System
Dynamics researchers acknowledge this challenge. Forrester (1994) indicates that the strength of
system dynamics comes from the fit between “the level-rate-feedback structure” and “the
fundamental and universal structure of real social and physical systems” which is necessary for
an information flow from real-world into a model.
System dynamics researchers developed a series of guidelines for the model building process to
ensure (Richardson & Pugh, 1981; Wolstenholme & Coyle, 1983; Sterman, 2000, 3). Group
Model Building approach became effective in terms of eliciting the mental data (Vennix, 1996;
Andersen & Richardson, 1997). But it is not always possible to directly access mental data base.
As in the Classic Maya Collapse model (Hosler, Sabloff, & Runge, 1977), researchers didn’t
have a chance to gather old Maya people for a group model building session. Even in some
cases, people as problem subjects may be alive, but it may be hard and cost ineffective to reach
them due to their location or availability. This was the case in the WTC research process. It was
very hard for the researcher to access the WTC responders to gather necessary information to
avoid the shortcomings of written records described by Forrester (1980, p. 587). That
information was already collected by other researchers through interviews before and they
indicated that these interviews had rich descriptions of the events. They also agreed that
analyzing this interview dataset can help the researcher to extend and elaborate the generic
dynamic theory.
Luna-Reyes and Andersen (2003) discuss collecting and analyzing qualitative data for system
dynamics. Despite the widely accepted importance of qualitative data in the system dynamics
model development process, there is not a clear description about how or when to use it. This
creates further discussions in system dynamics field about incorporating qualitative data into
quantified model variables, behaviors, and structures. There are well known research approaches
in social sciences that may guide system dynamics researchers in these discussions. Luna-Reyes
and Andersen (2003) indicate that data-gathering techniques such as interviews and focus
groups, and qualitative data analysis techniques such as grounded theory methodology and
ethnographic decision models could have a critical role in rigorous system dynamics efforts.
Scholars see potential in mixing System Dynamics with Grounded Theory and case study
research (Kapmeier, 2006; Kopainsky & Luna-Reyes, 2008; Laws & McLeod, 2004). Several
examples demonstrate successful results in research projects by mixing System Dynamics and
Grounded Theory (Black et al., 2004; Morrison, Rudolph, & Carroll, 2008; Rudolph &
Repenning, 2002).
3. Issue I; Existing Theories, Literature, Generic Structures
(Archetypes) and Heuristic C oncepts
The very first issue of mixing Grounded Theory and System Dynamics is the use of existing
theories, literature, and generic structures in a research project. In the early phases of a research
project, researcher decides when to use existing theories, literature and generic structures. As an
inductive approach, Grounded Theory specifically emphasizes theories emerging from data and
sensitive use of existing literature before data analysis. On the other hand in System Dynamics,
generic structures can be used upfront “as a means of using their isomorphic properties as a way
of starting the model conceptualisation activity by transferring insights from other models”
(Wolstenholme, 2003, p. 8).
The critical question is ‘Do System Dynamic archetypes or generic dynamic models pose a risk
of developing preconceived ideas in researcher's mind that colors qualitative data and force
data into a Procrustean bed?’ Both System Dynamics and Grounded Theory researchers are
interested in the answer of this question and current discussions in the Grounded Theory
enlighten this important issue for System Dynamics researchers too. Following subsections
summarize these discussions.
3.1 Existing Theories and Literature in Early Phases of Grounded Theory
Research
Grounded Theory is one of the most widely used approaches among the qualitative research
methods. With their Grounded Theory approach, Glaser and Strauss (1967) challenged the
hypothetico-deductive approach that “enforces the development of precise and clear cut theories
or hypotheses before the data collection takes place” (Kelle, 2005, para. 2). They indicated that
this approach led overemphasis on the verification of theory and “de-emphasis on the prior step
of discovering what concepts and hypotheses are relevant for the area that one wishes to
research” (Glaser & Strauss, 1967, p. 1f). As an alternative to the hypothetico-deductive
approach in social research, the Grounded Theory approach responded to these pitfalls by
allowing categories emerge from the data.
One of the main advantages of the grounded theory approach is that researchers have the
opportunity to work open minded with the possibilities of the data and the perspectives of the
subjects instead starting with a theory and let that theory to ‘color the data’ (Hyde, 2000). Being
open minded here means that “literally to ignore the literature of theory and fact on the area
under study, in order to assure that the emergence of categories will not be contaminated ...”
(Glaser & Strauss, 1967, p. 37). All these efforts are intended to prevent forcing of data into a
procrustean bed. As a result Grounded Theory became one of the most popular research methods
used by qualitative researchers in the social sciences.
This popularity led some adoption and changes of the Grounded Theory and different versions of
Grounded Theory were evolved over time (Morse, Stern, & Corbin, 2008). As one of the original
theorists, Glaser (2004, para. 5) criticizes mixing of Grounded Theory and qualitative data
analysis methodologies by indicating “the effect of downgrading and eroding the GT goal of
conceptual theory.” Having different versions of Grounded Theory Methodology are important
from mixing it with System Dynamics perspective, because these versions have different
approaches to existing theories and literature. While Classic (Glasarian) Grounded Theory
selected more sensitive approach to existing theory and literature, Strauss and Corbin version
(Corbin & Strauss, 2008; Strauss & Corbin, 1998) took more liberal approach by allowing use of
all kinds of literature before a research project.
3.2 Being Free of Any Theoretical Preconceptions
Strauss and Corbin version of Grounded Theory supports the use of literature before a research
project, but they (1998, p. 12) also indicate that ‘a researcher does not begin a project with a
preconceived theory in mind unless his or her purpose is to elaborate and extend existing
theory’. If we revisit the WTC research’s goal of extending and elaborating an existing generic
dynamic theory, the Strauss and Corbin statement solves a theoretical part of using existing
generic structures in such a research project and directs us to methodological protocol issues. But
his article not only explores issues of extending and elaborating a generic dynamic theory as in
the WTC case, but also discusses issues of developing a new dynamic theory with the generic
dynamic structures and archetypes in System Dynamics researchers’ minds. That’s why,
understanding the criticism of Grounded Theory’s approach to existing theories helps us to better
develop the protocols needed to mixed use Grounded Theory and System Dynamics.
Udo Kelle explores theoretical preconceptions issue in depth (Kelle, 1997, 2005). He (2005,
para. 4) indicates that the classic Grounded Theory methodologists’ standpoint represents one of
the roots of positivist epistemology. The earliest empiricist philosophers like Francis Bacon and
John Locke also supported the idea of inductive process of theory building by being open
minded (free of any theoretical preconceptions) before approaching empirical data. But this
approach, often called “naive empiricism” or “naive inductivism”, lost most of its supporters
after “Immanuel Kant’s sophisticated critique of the pitfalls of early empiricism” (Kelle, 2005,
para. 4). The idea of “being free of preconceived ideas” has been heavily criticized.
"Both historical examples and recent philosophical analysis have made it
clear that the world is always perceived through the ‘lenses’ of some conceptual
network or other and that such networks and the languages in which they are
embedded may, for all we know, provide an ineliminable ‘tint’ to what we
perceive" (LAUDAN, 1977, p. 15 from KELLE, 2005, para. 4)
Kelle (Kelle, 2005, para. 5) emphasizes the impossibility of freeing empirical observation from
all theoretical influence. He refers Lakatos’s thoughts (LAKATOS, 1978, p. 15 from KELLE,
2005, 5) as “one of the most crucial and widely accepted insights of epistemology and cognitive
psychology” that “there can be no sensations unimpregnated by expectations”. Kelle (Kelle,
2005, para. 6) criticizes inductive research strategy that neglects existing theories for demanding
an empty head instead of open mind since it is not possible to build a theory without already
accumulated knowledge. Qualitative researchers bring their own lenses and theoretical concepts
with them in their scientific investigations. Dropping them prevents their ability to perceive,
observe and describe meaningful events any longer (Kelle, 2005, para. 5).
Supporting an early review of literature in their article on improving qualitative methods in
public administration research, Brower, Abolafia and Carr (Brower, Abolafia, & Carr, 2000, p.
389) emphasize the importance of having a clear research question and understanding how
present theory bears on the question. They describe it as an iterative process between theory and
data. They recommend that “the researcher should plug theory into the data early and often...”
and “must remain informed, considering and reading new theoretical possibilities even as he or
she codes and analyzes data”. This is important for allowing “the research question to develop
in productive directions and the range of possible interpretations to grow”. As an important
point, they note that “researchers who do theoretically sensitive coding of social, political, and
economic conditions in field notes and interview transcripts will reveal more theoretically
powerful pictures of causality”. But an essential concem is commonly shared by qualitative
researchers that theoretical sensitivity should be exercised with caution to prevent forcing data
forehand and to allow the regularities and anomalies in the data to suggest possible theories.
3.3 Theoretical Codes and Generic Dynamic Structures
Kelle (2005, para. 7 and 8) acknowledges that the classic grounded theorists are aware of the
problem of excluding theoretical concepts early from research project. Glaser and Strauss’
(1967) “theoretical sensitivity” concept originally recognizes the use of theoretical concepts in
advance in grounded theory by addressing researcher’s ability “to reflect upon empirical data
with the help of theoretical terms”. They (1967, p. 46) note that “Sources of theoretical
sensitivity build up in the sociologist an armamentarium of categories and hypotheses on
substantive and formal levels. This theory that exists within a sociologist can be used in
generating his specific theory (...)”. A parallel idea was later developed by one of the original
grounded theorists, Glaser (1978), as ‘theoretical codes’ that theoretical concepts can be at
researcher’s disposal independently from data collection and data analysis. Kelle (Kelle, 1997,
para. 4.3) also indicates Strauss and Corbin’s more liberal position by quoting them that ‘all
kinds of literature can be used before a research study is begun...’ (Strauss & Corbin, 1990, p.
56).
While the grounded theory field discusses how to introduce theoretical concepts into data
analysis early in a research project, System Dynamics field has been more enthusiastic about
introducing them into research projects through concepts like ‘generic structures’ and
‘archetypes’. System Dynamics field used generic structures as a way to store insights gained in
specific cases by generalizing them since the beginning of the field (Lane & Smart, 1996).
Archetypes have been seen as a way to “generate understanding in new application domains and
systems” (Wolstenholme & Corben, 1993, p. 583). “Forrester had always advocated building a
general model or theory first, and then modifying it to fit the particular situation under study as
the preferred method for building any system dynamics mode/” (Lane & Smart, 1996, p. 92).
Key questions at this point are ‘Js it possible to use generic dynamic theories in similar way to
theoretical codes? Is it possible to divide generic dynamic theories into theoretical code parts, so
that researcher can use them in the mixed method?’ Our experience indicates that once divided
into parts generic dynamic theories can become theoretical codes. In order to understand how
these divided parts can be used in the mixed method, ‘empirical content’ and ‘falsifiability of
statements’ are explored in the next section.
3.3.1 Empirical C ontent of Theoretical C odes
A source of confusion of using theoretical codes in Grounded Theory comes from the differences
between qualitative and quantitative understanding of hypothesis. Even within the qualitative
research, the nature and use of theoretical codes is significantly different from the grounded
theory approach.
In order to understand these differences, Kelle (1997, para. 3.9) explores understanding of
‘hypothesis’ from a broad qualitative and quantitative perspectives. In quantitative approach,
“whatever specific claim the successful H(ypothesis) will make, it will nonetheless be an
hypothesis of one kind rather than another” (Hanson, 1971, p. 291). Focus is on attempting to
falsify an empirically contentfull statement. In qualitative research, especially in Grounded
Theory, hypotheses emerge as researcher interacts with data and initially hypotheses are vague
ideas about relations. Kelle (1997, para. 3.9) proposes that “instead of calling them hypotheses
one should rather call them hypotheses about what kind of propositions, descriptions or
explanations will be useful in further analysis” in Grounded Theory.
Based on the above discussion, Kelle (1997, para. 4.5) explains that qualitative researchers
(whether they apply a ‘grounded theory’ approach or not) use the theoretical preconceptions to
structure data material. These theoretical preconceptions play an important role in their abductive
inferences. In qualitative analysis, these theoretical preconceptions do not often represent explicit
propositions about empirical facts and Kelle (1997, para. 4.5) proposes that “they should be
referred to as ‘heuristic concepts’ which can be used to formulate ‘orientation hypotheses’”
(Merton, 1957, p. 88).
Hypothetical inferences enable a creative process to combine new and interesting empirical facts
with existing theoretical knowledge. But this doesn’t mean that “the theoretical knowledge of the
qualitative researcher should form in the beginning a fully coherent network of explicit
propositions from which precisely formulated and empirically testable statements can be
deduced” as in the hypothetico-deductive approach (Kelle, 2005, para. 32). In qualitative
inquiry, “it should constitute (a sometimes only loosely connected) "heuristic framework" of
concepts (or "coding families") which helps the researcher to focus the attention on certain
phenomena in the empirical field” (Kelle, 2005, para. 32). But this notion brings an ambiguity of
theoretically sensible category development process.
3.3.2 Falsifiability of Statements
“Falsifiability” or “empirical content” concepts are commonly used to identify sound scientific
hypotheses in a hypothetico-deductive framework. In that framework, it is important to originate
“clear-cut and precisely formulated propositions with empirical content as adequate
hypotheses”. Any concepts and hypotheses without these qualifications are regarded as highly
problematic, since they lack of empirical content and cannot be falsified. Kelle (2005, para. 33)
indicates the opposite picture in grounded theory generation framework. In that framework,
“Theoretical concepts with low empirical content, however, can play an extremely useful role if
the goal of empirical research is not the testing of predefined hypotheses but the empirically
grounded generation of theories, since they do not force data into a Procrustean bed—their lack
of empirical content gives them flexibility so that a variety of empirical phenomena can be
described with their help”(Kelle, 2005, p. 33). Kelle (2005, p. 33) acknowledges that such
concepts cannot be tested empirically. But they can be used as heuristic devices as theoretical
lenses to approach the phenomena and the data.
3.3.3 Heuristic Concepts
Kelle (Kelle, 1997, para. 4.5-5.10) describes three levels of ‘heuristic concepts’ from high
empirical content to low. The first level heuristic concepts are derived from ‘grand theories’ and
they are ‘highly abstract concepts about the relations between actors or between actors and
society in general’ (KELLE, 1997, 4.6-4.8; 2005, 35). “Sensitizing concepts” are this type of
concepts that they “/ack precise reference and have no bench marks which allow a clean cut
identification of a specific instance” (Blumer, 1954, p. 7). This type of heuristic concepts has
low empirical content that can be applied to different phenomena. Kelle (2005, para. 35)
indicates that these concepts may be useful in empirically grounded theory building. Abstract
preconceptions can be gathered from different theories to structure the data. Although
application of codes derived from specific theories makes the data structuration easier, it carries
an important risk of neglecting other theoretical concepts that may be more useful to explore the
phenomena in question. Kelle (2005, para. 37) proposes to use different and event competing
theoretical perspectives on the same data to address this risk.
Second level heuristic concepts are ‘theories of the members of the investigated culture’ (Kelle,
1997, para. 4.7). Strauss and Corbin’s “coding paradigm” and Glaser’s “theoretical codes”
concepts are the second type of heuristic concepts (Kelle, 1997, para. 5.4). They can be derived
from “general common sense knowledge” or “specific local knowledge of the investigated field”
(Kelle, 2005, para. 38). At this type of heuristic concepts, a certain code may increase the risk of
neglecting or excluding other relevant phenomena from examination.
Third level heuristic concepts are the ones that are closer to the Hypothetico-Deductive
approach’s ‘theory’. These concepts have high empirical content and they are falsifiable (at least
in principle). They are not as useful as the other kinds of heuristic concepts in an interpretative
research, since they may force the data into a Procrustean bed (Kelle, 2005, para. 39).
“Empirical content” and “falsifiability” concepts can help to identify heuristic concepts that can
help researchers to use their previous theoretical knowledge (whether they apply grounded
theory or not). Kelle (2005, para. 34) indicates that the first and second levels of “heuristic
concepts may be used to define a category scheme useable for the structuration and analysis of
qualitative data which can be supplemented, refined and modified in the ongoing process of
empirical analysis”.
3.3.4 Logic of Discovery: Inductive, Deductive or Abductive
Having discussed the importance of heuristic concepts in the mixed use of Grounded Theory and
System Dynamics, the logic of discovery or the direction of theorizing becomes questionable.
Researcher carries generic dynamic models, archetypes, and other related theoretical terms into a
research project and analyzes data by consulting these concepts. This approach seems more
deductive than inductive at this stage. Despite having more than twenty approaches to qualitative
research, many features of inductive paradigm are widely shared by most of them (Brower et al.,
2000). As a well established inductive approach, applying the grounded theory method in a
deductive way doesn’t seem like a good methodological approach considering the Grounded
Theorists’ rightful worries about not being sensitive about existing theories, and forcing data into
a Procrustean bed. Although “theoretical sensitivity”, “theoretical coding”, “axial coding”, and
“coding paradigms” are important concepts in grounded theory to overcome “naive empiricism”
of the emergence idea, clear cut methodological rules to address the concept of theoretical
sensitivity are still not available (Kelle, 2005, para. 9). At this stage, exploring the phases of
Grounded Theory and System Dynamics is necessary to understand the direction of theorizing in
Grounded Theory to realize the role of theoretical codes in research methodology protocols.
Discussions in System Dynamics field on the direction of theorizing indicate some confusion in
the field. While discussing theory-building processes in System Dynamics, Schwaninger and
Grésser acknowledge that their chosen theory-building processes ‘do not involve deduction or
induction along, but they utilize both in combination” (Schwaninger & Grosser, 2008). In his
commentary paper on the Schwaninger and Grobler’s article, GroRler (2008) focuses this issue
and separates applications in the field by acknowledging that some of the applications are
inductive and some of them are deductive system dynamics modeling. The logic of discovery
depends on “the process of model building, the nature of simulation results to be expected, and
the validity of conclusions to be drawn from the modeling endeavor” (GrofSler & Milling, 2007,
p. 1). Barton and Haslett (2006) interpret of the SD’s events-patterns-structure framework as an
application of abductive inference.
Kelle (1997, para. 4.4) indicates that ‘the application of a coding paradigm or of ‘theoretical
codes’ to empirical data is based on a logic of discovery which is neither inductive nor
deductive’. He calls it ‘Hypothetical Reasoning’ that ‘represents a special kind of logical
reasoning whose premises are a set of empirical phenomena and whose conclusion is a
hypothesis which can account for these phenomena’. Hypothetical reasoning is based on two
forms of logical inference; qualitative induction and abduction. The main difference between
qualitative induction and abduction is ‘with qualitative induction a specific empirical
phenomenon is described by subsuming it under an already existing category or rule’, with
qualitative abduction, ‘unknown concepts or rules on the basis of surprising and anomalous
events’ are sought by researcher.
In Abductive Inference (or reasoning), identifying a particular phenomenon is the starting point.
Then that phenomenon is accounted by relating it to broader concepts. A bductive inferences are
not only seek in data, but also are seek in “explanatory and interpretive frameworks” such as
researcher’s own experience, stock of knowledge of similar or comparable phenomena,
equivalent stock of ideas (theories, frameworks...) within one’s own discipline and other
disciplines. Although existing theories are useful, the researcher does not force the data to fit the
phenomena into existing theories, but his search includes “new, surprising, anomalous
observations”. At its core, there is “a repeated interaction among existing ideas, former findings
and observations, new observations, and new ideas” (Coffey & Atkinson, 1996, p. 156).
Kelle (2005, para. 31) indicates the importance of careful usage of previous knowledge:
“In making abductive inferences, researchers depend on previous knowledge that
provide them with the necessary categorical framework for the interpretation,
description and explanation of the empirical world under study. If an innovative
research process should be successful this framework must not work as a
Procrustean bed into which empirical facts are forced. Instead, the framework
which guides empirical investigations should be modified, rebuilt and reshaped
on the basis of empirical material.”
If codes from generic structures and the principles of system dynamics theory can be considered
among low empirical level theoretical codes, inductive inference emerges as the mode of
inference in the WTC case as a mixed use Grounded Theory and System Dynamics. If the
system dynamics modeling process becomes the basis for decision of inference mode, abductive
inference emerges as the mode of inference from the perspective of Barton and Haslett (2006).
The question remains on theoretical concepts with high empirical content that whether they can
be used in a qualitative research or not. Literature search on this topic resulted in several other
examples of deductive processes in qualitative research method (Hyde, 2000) such as ‘pattern
matching’, ‘referential adequacy’ (Lincoln & Guba, 1985), ‘analytical abduction’ (Kramer,
2007), hypothesis testing (King, Keohane, & Verba, 1994; Yin, 1994) and ‘content analysis’
approaches and techniques. Kelle’s (Kelle, 1997, para. 3.7) discussion of (quantitative) content
analysis is an example of coding within a hypothetico-deductive research strategy. Hyde’s
(2000) application of “pattern matching” approach to a marketing research phenomena also
exemplifies how to use theoretical concepts with high empirical content (clear-cut hypotheses).
Although increasing empirical content increases the risk of forcing data, Kelle (2005, para. 41)
also acknowledges that the use of categories and assertions with high empirical content can be
fruitful in a qualitative study. He (2005, para. 42) refers “Abductive Induction” research strategy
developed by the “Chicago School” of American sociology in the 1930s and it has been used in
qualitative studies since then. Empirical cases (called “crucial cases”) were used to examine and
modify initial hypotheses with high empirical content.
4. Issue II: Application of Grounded Theory Data Analysis to
Extract System Dynamics M odeling Information
Second important issue is the capability of the Grounded Theory approach to extract necessary
system dynamics modeling information from a qualitative dataset as a result of a secondary data
analysis. At the very core of the data analysis, researcher’s goal is to understand the phenomenon
in question. System Dynamics researchers specifically study causal relationships and dynamic
behaviors and they describe their understanding in causal maps, stock and flow diagrams, and
simulation models. Major information resources and parts of a system dynamics model are
depicted in Figure 3.
POLICY
CONCEPTS FROM
WRITTEN LITERATURE SVALUATION
7
PRINCIPLES OF = / \
FEEDBACK POLICY _ | ALTERNATIVE
PURPOSE ues CHANGES ~ BEHAVIOR
4
STRUCTURE l
™~
MENTAL ano) _> MODEL
WRITTEN |<“"_i PARAMETERS
INFORMATION ) 4
a \
RY BEHAVIOR
MISCELLANEOUS ~~~‘ piscREPANCIES
NUMERICAL IN BEHAVIOR
MODEL BEHAVIOR
Y MN SS
AND REAL-WORLD
TIME - SERIES ————————" gewavior
COMPARISON OF
Figure 3 - Creating a System Dynamics Model (Forrester, 1980, p. 559)
Data resources for system dynamics model development (Figure 3) are (1) Mental and Written
Information, (2) Concepts from Written Literature, and (3) Miscellaneous Numerical Data. In
secondary analysis of a qualitative dataset, researcher is restricted to written information
(interview transcripts in the WTC case) to understand mental data base of subjects. But the
analysis process with literature and related numerical data can help researcher to extract
structure, parameters and reference modes from qualitative dataset. This process is also highly
depended on the richness of a dataset from system dynamics modeling perspective.
Important questions are ‘What happen if qualitative dataset is not rich enough to extract enough
modeling information? What should researcher do under uncertainty?’ There are two answers in
System Dynamics field for these questions: One school of thought supports developing
qualitative maps (causal loop diagrams) to explain the phenomenon (Coyle, 2000; Wolstenholme
& Coyle, 1983), another school of thought supports developing fully developed simulation
models even under uncertainty (Homer & Oliva, 2001; Richardson, 1996, 1999; Sterman, 2002).
Later group indicates that simulation adds significant value on top of qualitative models, because
they are testable, they enable to draw behavioral and policy inferences reliably, and even under
uncertainties, simulations can indicate the missing value required for reaching firm conclusions
(Homer & Oliva, 2001).
System Dynamics researchers successfully adopted the Grounded Theory approach in an
inductive way to analyze qualitative data and theories to generate new theories (Black et al.,
2004; Rudolph & Repenning, 2002). These implementations demonstrate the capability of mixed
use of Grounded Theory and System Dynamics to retrieve enough modeling information from
written information and literature.
5. Issue III: Output of the Research Process
Outputs of a research project based on mixed use of Grounded Theory and System Dynamics can
be listed as:
o Substantive Theories to Explain Specific Cases
o Causal Loop Diagrams
o Reference Modes
o Fully developed System Dynamics Models with Dynamic Hypotheses
o Extended and Elaborated Generic Structures
5.1 Substantive Theories to Explain Specific Cases
Since the main goal of researcher is to understand the phenomenon in question, researcher is
subject to similar limitations that other Grounded Theory researchers face in analyzing their
qualitative data. Although a researcher initially relies on some theoretical codes based on generic
dynamic structures and principles of System Dynamics, these codes are empty abstractions
without substantive codes (Glaser, 1998, p. 164). A researcher who adopt the mixed method can
at least reach the same results of a Grounded Theory researcher, which is a substantive theory
explaining a specific case.
5.2 Causal Loop Diagrams
A researcher can describe her causal understanding of case in causal loop diagrams. These
diagrams are referred to as qualitative system dynamics models in the System Dynamics field. It
is important to remember that the System Dynamics field has been discussing merits and
disadvantages of qualitative and quantitative modeling since 1980s. Merits of causal loop
diagrams were acknowledged in these discussions. Researcher adopting the mixed method may
develop causal loop diagrams to explain case in question.
5.3 Reference Modes
Reference Modes are important part of problem definition and model testing stages of system
dynamics modeling. A reference mode is a graphical description of historical behavior and
inferred future trend (Saeed, 1998). Saeed (Saeed, 1998, pp. 2-6) explains the more complex
nature of reference modes. A reference mode is an abstract concept that represents a pattern of
behavior in a qualitative, intuitive, organized, integrated, and noise-free way to describe problem
behavior (Saeed, 1998, p. 4). From this perspective, researcher’s attempts to construct reference
modes based on qualitative data result in rich descriptions of problem understanding. Reference
modes by themselves can become an important medium to communicate the understanding from
research.
Although researchers frequently emphasized dynamic changes in their studies, scholars (other
than systems school) rarely presents these dynamic changes in graphs over time for key
variables. In one of those rare presentations, Levina (2005, fig. 1) presented a degree of project
involvement using different actors to describe the practical change in IT development (Figure 4).
=
e
2 ~
a Strategists
=
=
S
= 7 =
2 y
a =<
‘6 Graphic designers JL \
3 Py J \
o © 33 - A) =
a JY _—— es
Technologists Time
_
Planning Prototyping Development
phase phase phase
Figure 4 — The “Waves” Service Delivery Model (Levina, 2005, fig. 1)
In addition, Fichman and Kemerer (1999, fig. 2) introduced the assimilation gap concept by
describing the behavioral change over time and modeling assimilation gaps for software process
innovations (Figure 5).
100% — — — — — — — — —
Cumulative
"Adoption"
Time
Figure 5 — Assimilation Gap (Fichman & Kemerer, 1999, fig. 2)
Researchers constructed these reference modes by adopting different research methodologies. A
researcher using the mixed method can put the characteristics of reference modes into his
theoretical codes and search for evidence during Grounded Theory analysis of qualitative data.
This process can deliver reference modes with important behavioral insights about key variables.
5.4 Fully developed System Dynamics Models with Dynamic Hypotheses
Many System Dynamics researchers agreed that a research project reaches its maximum value
with fully developed system dynamics models with dynamic hypotheses in System Dynamics
field . The critical question in the mixed method is the ability to collect enough system dynamics
modeling information from secondary qualitative data. This question is discussed in the ‘Issue II:
Application of Grounded Theory Data Analysis to Extract System Dynamics Modeling
Information’ section. As a result, the mixed method can deliver enough modeling information
from a qualitative dataset. Depending on the richness of the dataset, some of the modeling
information may not be available from it. In this case, the researcher either chooses to stop at this
stage and delivers his understanding of the case through descriptions, causal loop diagrams, and
reference modes, or researcher continues to build a fully developer system dynamics model by
indicating missing information and discussing potential findings under such uncertainties.
5.5 Extended and Elaborated Generic Structures
The main goal of the WTC research was to extend and elaborate a generic dynamic theory. The
theoretical codes in this project contained the concepts from the generic dynamic theory from the
beginning. The analysis process in the mixed method delivered enough information about the
structure, behavior and variables in the generic dynamic theory. At the end, the main goal of the
research project was reached by extending and elaborating generic structures based on
substantive theory in the WTC response and recovery case.
Archetypes and generic dynamic structures store insights gained in specific cases by generalizing
them. This is an important research project goal in the System Dynamics field and the mixed
method can deliver enough material to researcher to make changes on these generic structures.
6. Suggested Research Design for the Mixed Use of Grounded
Theory and System Dynamics Modeling
The purpose of this section is to describe the WTC research design and discuss some protocol
related issues in depth. The methodological challenges arose from the interaction between
components of the WTC research: existing dynamic theory, interview dataset, secondary data
analysis, and the Grounded Theory approach. Although all of these components have their own
consistent and widely accepted ways of uses, it is challenging to bring all of them together in a
single research project. A research design was developed to address these methodological
challenges in the WTC research (Figure 6). Given the iterative nature of qualitative research and
system dynamics approach, we acknowledge the limitations of explaining and presenting the
research design on a diagram. Understanding the fundamentals of qualitative research and
system dynamics approach and then exploring the research design here is important to fill the
potential gaps in this graphical presentation.
The research design has several iterative stages. The first stage is problem definition and
research question formulation. This stage leads to the development of research questions and
heuristic concepts. The second stage is data analysis where the WTC dataset was analyzed using
a grounded theory approach (Strauss & Corbin, 1998). The data coding process is part of this
stage. The second stage ends when theoretical saturation is reached. The products of the data
analysis stage are (1) categories and properties, (2) relations among categories, (3) memos and
diagrams, (4) system dynamics modeling information, and (5) causal understanding of the
phenomenon. While a substantive theory was built along with the theoretical saturation, the
theoretical concepts were densified to complete the theory at the third stage. The results of the
analysis were compared to the generic dynamic theory‘s propositions to increase understanding
of the phenomenon and to extend and elaborate the generic dynamic theory at the fourth stage.
EXISTING THEORY
RESEARCH
QUESTION >
> SD LITERATURE/GENERIC
A TTERATURE MODELS/ ARCHETYPES
HEURISTIC CONCEPTS
>» I
: OPEN CODING
I\ \
(iicorstea! DATASET AXIAL CODING
J Theoretical
‘Saturation?
Yes
([bensiving\
| Theoretical
\Concepts /
~e’
{ Testing
\Propositions |
\ SELECTIVE CODING /
CATEGORIES AND PROPERTIES
RELATIONS BETWEEN
CATEGORIES
MEMOS AND DIAGRAMS
SYSTEM DYNAMICS
MODEL INFORMATION
CAUSAL UNDERSTANDING OF
THE PHENOMENON
EXTENDED AND ELABORATED
SUBSTANTIVE DYNAMIC THEORY
GENERIC
MODELS/
ARCHETYPES
Figure 6 - Research Design
6.1 Problem Definition and Research Question F ormulation
The first stage of the research design is problem definition and research question formulation. In
this stage, research questions are shaped by the existing theory and literature. Following the
fundamentals of research question evaluation in grounded theory (Strauss & Corbin, 1998),
research questions start broadly to give necessary flexibility and freedom to explore a
phenomenon in depth. In the WTC research case, the research problem and questions are based
on extending and elaborating a generic dynamic theory. The research questions are defined based
on the existing theory’s concepts, hypotheses and claims, and themes emerging from the dataset.
The intensity of a literature review generally depends on the approach the researcher is adopting.
Agreeing with the principles of classic grounded theory, a researcher may skip an extensive
literature review initially and may come back after indentifying categories from the data. Given
the notion that “it is important for qualitative studies to articulate and answer a specific research
question” (Brower et al., 2000, p. 386). In order to answer a specific research question, a
researcher may make it explicit as early as possible, so relevant data sources can be accordingly
chosen. Otherwise there is a risk of “wasting time gathering unusable data and arriving at
various research dead ends” (ibid).
This first step of this research design corresponds to the problem articulation step of the system
dynamics modeling process. Having a similar iterative approach to grounded theory, system
dynamics models are also grounded on the data. Additional to the research problem and question
definition, system dynamics researchers also search for key variables and concepts, the time
horizon of the problem, and reference modes (historical and predicted future behavior of the key
concepts and variables). In the modeling process, “results of any step can yield insights that lead
to revisions in any earlier step” (Sterman, 2000, p. 87). Similar to grounded theory, these
insights emerge throughout the research process.
The goal of extending and elaborating upon the theory the WTC research requires delicate
handling of the heuristic concepts and propositions of the generic dynamic theory at the
beginning of the research. Following the fundamentals of grounded theory (Strauss & Corbin,
1998), as part of theory extension and elaboration, heuristic concepts of the II] Theory were
carried into the data analysis at the beginning of the research. These heuristic concepts with low
empirical content are grand concepts and abstract theoretical concepts derived from the generic
dynamic theory. The generic dynamic theory explains the interaction of social processes and
accumulations in an interagency information integration initiative. The description of generic
processes creating technical artifacts in a social process (Figure 7) became the base for
developing heuristic concepts and dynamic hypotheses in the WTC research.
— a Social
Accumulation 1
~ Ar)
Ffectiveness . iS ) ’
on uildling social
4s) effectiveness Building.
| case of processing
\ RD \
Z ge \
fe ge Artifact 1 Artifact 2 ate)
Processing A Processing B Processing C |
/ A >) | B) \ (aJ nN
| cy, AE 7 }
\ pressure to | Need For more
Growing moti oj
\ process a
\. |
~p Effort on ww,
Figure 7 - Generic processes creating technical artifacts in a social process (Luna-Reyes et al., 2004)
An interactive process of developing the research question with the existing theory and the
literature leads the development of heuristics concepts (Box 1). Having a generic dynamic theory
prevents forcing the data into a Procrustean bed and enables emergence of theory relevant
concepts from the qualitative data.
The generic dynamic theory‘s dynamic hypotheses with high empirical content (herein, called
‘propositions’) are also available for the analysis. The dynamic hypothesis concept comes from
the system dynamics approach where it describes “a theory about what structure exists that
generates the reference modes” (a pattern of behavior over time). “A dynamics hypothesis can
be stated verbally, as a causal loop diagram, or as a stock and flow diagram” (VENSIM, 2009).
In the generic dynamic theory, dynamic hypotheses are derived from the generic processes
causal loop diagram (Figure 7). In order to avoid confusion, these dynamic hypotheses are called
‘proposition’ in this research consistent with the qualitative approach to emphasize their
difference to the hypothesis concept in the quantitative approach (Kelle, 1997).
Box 1- Grand Theory and Abstract Theoretical C oncepts developed based on the existing theory
The III theory has ten propositions:
P1: Social practice causes social accumulation.
P2: A reinforcing feedback loop exists between social practice and social accumulation
through individual/group effectiveness that builds social effectiveness.
P3: A reinforcing feedback loop exists between social practice and social accumulation
through individual/group effort that grows motivation.
P4: A balancing feedback loop exists in that, as social practice changes Artifact 1, these
changes affect social practice back through individual/group effectiveness.
P5: A balancing feedback loop exists in that, as social practice changes Artifact 1, these
changes affect social practice back through individual/group effort.
P6: A balancing feedback loop exists in that, as social practice changes Artifact 2, these
changes affect social practice back through individual/group effort.
P7: A feedback loop exists in that, as social practice changes Artifact 2, these changes
affect social practice back through individual/group effectiveness.
P8: Independently accumulated social accumulations affect social practices through
individual/group effectiveness.
P9: Independently accumulated social accumulations affect social practices through
individual/group effort.
P10: Social accumulations have initial values that accumulate independent of subjected
social practice.
Having propositions with high empirical content creates a risk of forcing data initially into a
Procrustean bed. To avoid this risk, the propositions should not be employed or considered in the
analysis. Once relevant concepts and hypotheses emerged and were validated against data, these
propositions can be introduced in order to compare them to the emerging findings and
hypotheses to increase understanding of the phenomenon (Strauss & Corbin, 1998).
6.2 Data Analysis
Grounded Theory techniques are used in this stage. Theoretical sampling process starts based on
the research question and heuristic concepts. “Theoretical sampling is the process of data
collection for generating theory whereby the analyst jointly collects, codes and analyses the data
and decides what data to collect next and where to find them, in order to develop the theory as it
emerges” (Glaser & Holton, 2004, p. 51).
The available number of samples (interview transcripts in the WTC case) is an important
restriction in secondary data analysis. As noted by the grounded theorists, it is possible to reach
theoretical saturation before analyzing all the interview transcripts. But an opposite scenario is
also possible with a limited size of data that theoretical saturation may not be reached with the
given dataset.
The coding processes are “the analytic processes through which data are fractured,
conceptualized, and integrated to form theory” (Strauss & Corbin, 1998, p. 3). Open, axial and
selective coding procedures are discussed in details in the literature (Strauss & Corbin, 1998).
Iterative coding processes lead development of categories and their properties, relationships
between categories, and causal understanding of the phenomenon along with memos, diagrams
and modeling information. This iterative process helps researcher to accumulate a causal
understanding of the phenomenon. Causal understanding grounded on empirical evidences is
critical in the substantive theory development process. The substantive theory answers the
research questions that evolved throughout the research steps based on emerging themes.
6.3 Theory Development
After this step it is possible to complete the research process with a substantive dynamic theory
for qualitative researchers. Some of the system dynamics researchers (Coyle, 2000;
Wolstenholme & Coyle, 1983) also settle at this stage by developing causal loop maps to
describe the phenomenon, if they do not have enough modeling information. Despite the
challenge and risks of quantification process, many system dynamics researchers (Homer &
Oliva, 2001; Richardson, 1999; Sterman, 2002) find the actual value by formulating a simulation
model.
Formulating of a simulation model step follows the decision to quantify the causal understanding
of the phenomenon. Formalization helps “to recognize vague concepts and_ resolve
contradictions that went unnoticed or undiscussed during the conceptual phase” (Sterman, 2000,
p. 103). Recognizing the missing information, researcher can go back to data to gather more data
and develop more accurate dynamic description of the phenomenon.
Given the collected data, researchers begin to develop a dynamic hypothesis to account for the
phenomenon during the dynamic hypothesis formulation stage. The hypothesis is called
dynamic, because it explores and explains the dynamic nature of the phenomenon by
characterizing the underlying feedback, and stock-and-flow structure. Although a dynamic
hypothesis is a working theory of the phenomenon, it is a hypothesis due to its provisional
nature. It is revised or abandoned throughout the research process based on the information
gathered from the modeling process and from the real world (Sterman, 2000, p. 95). Initial or
emerged heuristic concepts, categories and their properties are used to define key variables in the
system dynamics model. The time horizon and reference modes are critical category properties
for building a system dynamics model that aimed to be extracted in the coding process. Causal
relationships between categories and memos help to reflect causal understanding of the
phenomenon in the mapping process. Maps of causal structure are developed “based on initial
hypotheses, key variables, reference modes and other available data using tools such as model
boundary diagrams, subsystem diagrams, causal loop diagrams, stock and flow maps, policy
structure diagrams and other tools” (Sterman, 2000, p. 86).
Based on the knowledge gathered from the substantive theory and substantive dynamic theory,
research can revisit the propositions of generic dynamic theory and offer changes to extend and
elaborate the theory or the generic structure.
7. Conclusion
Since its introduction in mid-1950s, System Dynamics approach has been using qualitative data
to study complex social systems. Despite the central role of qualitative data in system dynamics
model development process, System Dynamics field does not have detailed protocols to describe
the use of qualitative data or qualitative research methods in the modeling process (Luna-Reyes
& Andersen, 2003). The Grounded Theory approach is a popular methodology in qualitative data
analysis and it is being used by researchers in System Dynamics modeling. This article discusses
the methodological issues in mixed use of Grounded Theory and System Dynamics approaches
by referring a research experience based on secondary analysis of qualitative dataset.
The first issue deals with the use of existing literature and generic dynamic structures
(archetypes) in a research project. The Grounded Theory field has been discussing the role of
existing literature or preconceptions in the data analysis phase. Several different versions of
Grounded Theory evolved in time and they have different approaches to operating in this role.
This is a critical issue for System Dynamics field too, because generic structures play important
role during the research phases. Based on the discussions in the Grounded Theory field, we
concluded that generic structures can be used as heuristic concepts. But these heuristic concepts
should be low in empirical content, so that they will not force the data into a Procrustean bed.
Any concept with high empirical content coming from generic theories can be used after
theoretical saturation is reached.
The second issue is focused on the mixed method’s ability to extract the necessary modeling
information from a qualitative dataset. Existing examples from System Dynamics field shows
that the mixed method successfully delivers enough modeling information. But our focus in this
issue is more of secondary use of qualitative data to build system dynamics models. Secondary
data analysis restricts researcher to a predefined number of materials. While the question comes
to the quantity of information needed for model building, qualitative vs. quantitative modeling
discussions in System Dynamics field indicate that many researchers find additional value of
developing simulation models even under uncertainties.
The third issue is potential outputs of a research project using the mixed method. These outputs
are (1) substantive theories to explain specific cases, (2) causal loop diagrams, (3) reference
modes, (4) fully developed system dynamics models with dynamic hypotheses, and (5) extended
and elaborated generic structures. If we put aside the expectations from System Dynamics
perspective, the mixed method can deliver similar result in Grounded Theory research at least.
This result is a substantive theory explaining the problem in question. After gaining causal
insights about a problem, a researcher can describe her understanding in a causal loop diagram.
Again revisiting the qualitative modeling issue in System Dynamics field will show that some
researchers can be satisfied with such a result under some conditions. Another set of important
artifact in the field consists of reference modes. Reference modes are building blocks of system
dynamics models. But they are also very useful tools to describe and discuss dynamic behaviors.
Although examples are comparably very rare, several researchers outside the system school
construct reference modes to explain dynamic behavior in problem domains. The mixed method
can also help a researcher to develop reference modes grounded on qualitative data. Finally a
much sought result by researchers, developing a simulation model, can also be reached with the
mixed method as it is discussed under the second issue. Parallel to the enthusiasm of using
generic structures or archetypes in research, researcher should also revisit relevant generic
structures and offer changes based on his findings, where necessary.
Acknowledgements
The authors would like to thank David F. Andersen and Deborah L. Andersen from University at
Albany for their valuable contribution during the research process.
References
Akcam, B. K. (2009). Socio-technical processes in interorganizational emergency response and
recovery process at the World Trade Center. Albany, New York: State University of
New Y ork at Albany.
Andersen, D. F., & Richardson, G. P. (1997). Scripts for group model building. System
Dynamics Review, 13(2), 107-129.
Barton, J., & Haslett, T. (2006). Fresh insights into system dynamics methodology-developing an
abductive inference perspective. Presented at the The 24th International Conference of
the System Dynamics Society, Nijmegen, The Netherlands.
Black, L. J., Carlile, P. R., & Repenning, N. (2004). A Dynamic Theory of Expertise and
Occupational Boundaries in New Technology. Administrative Science Quarterly, 49,
572-607.
Blumer, H. (1954). What is wrong with social theory? American sociological review, 19(1), 3-
10.
Brewer, J., & Hunter, A. (1989). Multimethod research: A synthesis of styles (volume 175 of the
sage library of social research). Newbury Park, CA and London: Sage Publications, 16-
18.
Brower, R. S., Abolafia, M. Y., & Carr, J. B. (2000). On Improving Qualitative Methods in
Public Administration Research. Administration & Society, 32(4), 363.
Coffey, A., & Atkinson, P. (1996). Making sense of qualitative data: Complementary research
strategies. Sage Publications, Inc.
Corbin, J., & Strauss, A. (2008). Basics of qualitative research: techniques and procedures for
developing grounded theory. SAGE.
Coyle, R. G. (2000). Qualitative and quantitative modelling in system dynamics: some research
questions. System Dynamics Review, 16(3), 225-244.
Fichman, R., & Kemerer, C. (1999). The Illusory Diffusion of Innovation: An Examination of
Assimilation Gaps. Information Systems Research, 10(3), 255.
Forrester, J. (1980). Information sources for modeling the national economy. Journal of the
American Statistical Association, 75(371), 555-566.
Forrester, J. (1991). System Dynamics and the Lessons of 35 Y ears. The Systemic Basis of Policy
Making in the 1990s.
Forrester, J. (1994). System Dynamics, Systems Thinking, and Soft OR. System Dynamics
Review, 10(2).
Forrester, J. (1998). Designing the future. Work presented at Universidad de Sevilla, 15.
Glaser, B. G. (1978). Theoretical sensitivity: advances in the methodology of grounded theory.
Mill Valley, CA: Sociology Press.
Glaser, B. G. (1998). Doing Grounded Theory: Issues & Discussion. Sociology Press.
Glaser, B. G., & Holton, J. (2004). Remodeling Grounded Theory. Forum Qualitative
Sozialforschung/Forum: Qualitative Social Research, 5. Retrieved from
http://www.qualitative-research.net/fqstexte/2-04/2-04glaser-e.htm
Glaser, B. G., & Strauss, A. (1967). The discovery of grounded theory: strategies for qualitative
research. Chicago: Aldine Publishing Company.
GroRler, A. (2008). System dynamics modelling as an inductive and deductive endeavour.
comment on the paper by Schwaninger and Grosser. Systems Research and Behavioral
Science, 25(4), 467-470. doi:10.1002/sres.908
GroRler, A., & Milling, P. (2007). Inductive and Deductive System Dynamics Modeling. In The
2007 International Conference of the System Dynamics Society.
Hanson, N. R. (1971). The Idea of a Logic of Discovery. In S. Toulmin & H. Woolf (Eds.), What
I Do Not Believe, and Other Essays. Springer.
Homer, J., & Oliva, R. (2001). Maps and models in system dynamics: a response to Coyle.
System Dynamics Review, 17(4), 347-355. doi:10.1002/sdr.224
Hosler, D., Sabloff, J. A., & Runge, D. (1977). Simulation model development: a case study of
the Classic Maya collapse. Social Process in Maya Prehistory, 553-90.
Hyde, K. F. (2000). Recognising deductive processes in qualitative research. Qualitative market
research: An international journal, 3(2), 82-90.
Johnson, R. B., & Onwuegbuzie, A. J. (2004). Mixed methods research: A research paradigm
whose time has come. Educational researcher, 33(7), 14-26.
Kapmeier, F. (2006). Research Based on Two Pillars: Combining Qualitative Empirical Social
Research and Simulation in Strategic Management. Presented at the 24th International
Conference of the System Dynamics Society, Nijmegen, The Netherlands.
Kelle, U. (1997). Theory building in qualitative research and computer programs for the
management of textual data. Sociological Research Online, 2(2), 1-17.
Kelle, U. (2005). Emergence" vs." forcing" of empirical data? A crucial problem of" grounded
theory" reconsidered. Forum Qualitative Sozialforschung/Forum: Qualitative Social
Research, 6, 2-05.
King, G., Keohane, R. O., & Verba, S. (1994). Designing social inquiry: Scientific inference in
qualitative research. Princeton Univ Pr.
Kopainsky, B., & Luna-Reyes, L. F. (2008). Closing the loop: promoting synergies with other
theory building approaches to improve system dynamics practice. Systems Research and
Behavioral Science, 25(4), 471-486.
Kramer, E. H. (2007). Organizing doubt: grounded theory, army units and dealing with dynamic
complexity. Copenhagen Business School Pr.
Lakatos, I. (1978). The methodology of scientific research programmes. Cambridge University
Press.
Lane, D. C., & Smart, C. (1996). Reinterpreting ‘generic structure’: evolution, application and
limitations of a concept. System Dynamics Review, 12(2), 87-120.
Laudan, L. (1977). Progress and its problems: Toward a theory of scientific growth. Erkenntnis,
15(1), 91-103.
Laws, K., & McLeod, R. (2004). Case study and grounded theory: Sharing some alternative
qualitative research methodologies with systems professionals. In Proceedings of the
22nd International Conference of the Systems Dynamics Society.
Levina, N. (2005). Collaborating on Multiparty Information Systems Development Projects: A
Collective Reflection-in-Action View. Information Systems Research, 16(2), 109-130.
doi:10.1287/isre.1050.0055
Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic inquiry. Sage Publications, Inc.
Luna-Reyes, L., Diker, V., & Andersen, D. L. (2005). Interviewing as a strategy for the
assessment of system dynamics models. In 23rd international conference of the system
dynamics society.
Luna-Reyes, L. F., & Andersen, D. L. (2003). Collecting and analyzing qualitative data for
system dynamics: methods and models. System Dynamics Review, 19(4), 271-296.
Luna-Reyes, L. F., Mojtahedzadeh, M., Andersen, D. F., Richardson, G. P., Pardo, T. A., Burke,
G. B., Wu, Y., et al. (2004). Emergence of the Govemance Structure for Information
Integration across Govemmental Agencies: A System Dynamics Approach. In
Proceedings of the 22nd International Conference of the System Dynamics Society.
Presented at the International Conference of the System Dynamics Society, Oxford,
England, UK. Retrieved from
http://www.systemdynamics.org/conferences/2004/SDS_2004/PA PERS/382LUNA .pdf
Merton, R. K. (1957). Social theory and social structure. Free Press.
Morrison, J. B., Rudolph, J. W., & Carroll, J. S. (2008). The Dynamics of Diagnosing: Virtuous
and Vicious Cycles in the Operating Room. Presented at the The 26th International
Conference of the System Dynamics Society, Athens, Greece.
Morse, J., Stem, P., & Corbin, J. (2008). Developing grounded theory: The second generation.
Left Coast Press.
Richardson, G. P. (1996). Problems for the future of system dynamics. System Dynamics Review,
12(2), 141-157.
Richardson, G. P. (1999). Reflections for the future of system dynamics. Journal of the
Operational Research Society, 50(4), 440-449.
Richardson, G. P., & Pugh, A. L. (1981). Introduction to system dynamics modeling with
DYNAMO. Cambridge, MA: Productivity Press (Now available from Pegasus
Communications, Waltham, MA).
Rudolph, J. W., & Repenning, N. (2002). Disaster dynamics: Understanding the role of quantity
in organizational collapse. Administrative Science Quarterly, 47(1), 1-30.
Saeed, K. (1998). Defining a problem or constructing a reference mode. Presented at the 16th
Intemational Conference of the System Dynamics Society, Québec City, Canada.
Schwaninger, M., & Grésser, S. (2008). System dynamics as model-based theory building.
Systems Research and Behavioral Science, 25(4), 447-465.
Sterman, J. (2000). Business Dynamics: Systems Thinking and Modeling for a Complex World.
Irwin McGraw-Hill.
Sterman, J. (2002). All models are wrong: reflections on becoming a systems scientist. System
Dynamics Review, 18(4), 501-531.
Strauss, A., & Corbin, J. (1990). Basics of Qualitative Research: Grounded Theory Procedures
and Techniques (2nd ed.). Sage Publications, Inc.
Strauss, A., & Corbin, J. (1998). Basics of Qualitative Research: Techniques and procedures for
developing grounded theory. London: Sage Publications.
Vennix, J. (1996). Group Model Building: Facilitating Team Learning Using System Dynamics.
Wiley.
VENSIM. (2009). The System Dynamics Process. In Vensim® Version 5 Documentation.
Retrieved from http://www.vensim.com/documentation/html/21360.htm
Weber, R. (2004). Editor's comments: The rhetoric of positivism versus interpretivism: A
personal view (Vol. 28). Management Information Systems Research Center, University
of Minnesota.
Wolstenholme, E. F. (2003). Towards the definition and use of a core set of archetypal structures
in system dynamics. System Dynamics Review, 19(1), 7-26. doi:10.1002/sdr.259
Wolstenholme, E. F., & Coyle, R. G. (1983). The development of system dynamics as a
methodology for system description and qualitative analysis. Journal of the Operational
Research Society, 34(7), 569-581.
Wolstenholme, E. F., & Corben, D. A. (1993). Towards a core set of archetypal structures in
system dynamics (p. 583). Presented at the The 11th Intemational Conference of the
System Dynamics Society, Cancun, Mexico.
Wolstenholme, E. F., & Coyle, R. G. (1983). The Development of System Dynamics as a
Methodology for System Description and Qualitative Analysis. The Journal of the
Operational Research Society, 34(7), 569-581. doi:10.2307/2581770
Yin, R. K. (1994). Case Study Research: Design and Methods (2nd ed.). Sage Publications, Inc.
