Diversity of Approaches Among Users of Causal Mapping
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
136 137 138 139 140 141 142 143 144 145 146
Throughout the evolution of causal mapping, users have adopted diverse approaches,
sometimes with different philosophical assumptions. So that we may appreciate this
diversity, I will summarize the different approaches along three dimensions: a) Perspective,
b) Research Contexts, and c) Focus. A fourth dimension, methodology, will be
extensively dealt with in a later chapter (Chapter III) by Hodgkinson and Clarkson.
Perspective
Over the last three decades, researchers employing causal mapping as a methodological
tool have invoked three different perspectives: (a) social constructionist, (b) objectivist,
and (c) expert-anchored.
Social constructionist. In this perspective, the researcher is interested primarily in
portraying the causal maps of the subjects — individuals or social systems — under
study. The researcher is intrinsically interested in these maps, and expects the maps to
have value in providing a cognitive explanation for the phenomena of his/her interest.
The primary methodological challenge is establishing the accuracy of the researcher’s
representation of the subject’s causal map. Most social constructionists deal with
organizational and social psychological phenomena, where different individuals can
hold different views of the world, and in most cases there is no single correct view. Barr
et al. (1992) and Narayanan and Fahey (1990) exemplify this perspective.
Table 1. Social constructionist, objectivist and expert-anchored perspectives
Expert casual maps
enable the causeeffect
relations
Phenomena can be
represented
accurately as a
causal map
Individual’s
causal maps
shape their
actions
Assumptions
Judgmental
situations
Primarily for
physical phenomena
or deterministic
social phenomena
Social
phenomena,
uncertain
theoretical
contexts
Appropriate for
Locating the experts,
and accurately
representing their
casual maps
Establishing the
correct causal map Accuracy of
representation of
individuals’
causal map
Key
Methodological
Challenge
Social Objectivist Expert-Anchored
Constructionist
Objectivist. Researchers adopting this perspective are typically interested in establishing
the “true” causal representation of some phenomenon. For many, causal mapping is
a simplified way to accomplish what industrial dynamics did for economic systems. A key
methodological challenge is establishing not merely the accuracy of representation, but
also an accurate description of the phenomenon. In the objectivist perspective, an
individual’s causal maps may be of interest largely to establish the degree to which the
individual holds an accurate description of the phenomenon under study. The objectivist
view is most applicable to the study of physical and technical subsystems, and is less
prevalent in organizational sciences.
Expert-anchored. Researchers adopting this perspective are primarily interested in
those phenomena where human judgment plays an important role. They acknowledge the
social construction of many phenomena, but admit that individuals have varying levels
of expertise within different knowledge domains. Thus, experts in their respective
domains set a benchmark against which other individuals can be judged. Nadkarni and
Narayanan (in press) exemplify this approach.
In the contemporary literature on causal mapping, discussions of the underlying
perspective are often glossed over or left implicit. However, I will emphasize that
researchers should be acutely aware of their perspective since it relates to key methodological
challenges they may confront. For example, researchers representing a phenomenon
as accurate — the objectivist perspective — should establish the accuracy of the
causal map with respect to the phenomena, not merely the accuracy of the representation
of an individual’s causal map. Similarly, the expert-anchored perspective requires
researchers to establish the credentials of the experts, and use the map of an expert (either
a specific individual or a group of individuals in the case of complex phenomena) as a
benchmark for evaluating the accuracy of others’ maps.
Research Contexts
One of the great advantages of causal mapping is the versatility of its application. Indeed,
it has been used in four distinct research contexts: (a) discovery, (b) hypothesis testing,
(c) evocative, and (d) intervention.
Discovery. When utilized in ethno methodological inquiries, causal mapping provides a
systematic approach to unearth phenomena. It is expected that two individuals following
the causal mapping coding rules will arrive at congruent representations of the phenomena
under discovery from the same set of interviews or archival materials. In this way,
the use of causal mapping reduces the “subjective” component of data analysis that has
been the Achille’s heel of ethno methodological studies. However, this comes at a price
— causal mapping reduces the role of human imagination in theory building. It also
restricts researcher attention to phenomena that admit causal modeling. To date, causal
mapping has been used predominantly in discovery contexts.
Hypothesis-testing. Increasingly, causal mapping is being used in “normal” science
investigations, or more accurately, in studies that focus on hypothesis testing via
statistical inference using large samples. The introduction of network methods of
representation of causal maps and the derivative variables, which can be measured on
interval or ratio scales, have enabled researchers of qualitative phenomena to operate
in a hypothesis testing mode. Calori et al. (1994) and Marcozy (1997) exemplify this
context. A significant barrier to large sample hypothesis testing studies has been the
labor intensity of the causal mapping procedure. This may change as more sophisticated
softwares enable us to automate the causal mapping procedure.
Evocative. In between discovery contexts with ill-defined theories and hypothesis
testing contexts with clearly formulated theories, lies a context that Nelson et al. (2000)
called “evocative.” In evocative contexts, general theoretical frameworks are available,
but specific operationalizations of concepts and linkages among them are undeveloped.
In evocative contexts, experts who practice in a specific domain are available, but studies
are needed to unearth their knowledge and examine it through available general theoretical
frameworks to construct domain specific theories. Causal mapping evokes the
concepts and causal linkages among them.
Intervention. Another popular use of causal mapping has been to assist management
groups and organizations to make decisions. When complex IT systems are installed, the
design phase may be enabled by the use of causal mapping to tease out implementation
Table 2. Causal mapping in four contexts
Diverse sources to fully capture the
phenomena
Primary
stakeholders and
convenience
sample
Relevant
population
sampling drawn by
statistical
consideration
Participants in Experts
the system
Source
As an input to
decision making
Obtaining relevant
data
Operationalizing
concepts
Deriving
concepts and
establishing
linkages
Applicability of
Causal
Mapping
Can vary from
undeveloped to
fully developed
Both theory and
operationalization
available
General theoretical
framework available;
No operationalization
State of Theory Undeveloped
Hypothesis Intervention
testing
Discovery Evocative
challenges that could be addressed during the early phases. Alternately causal mapping
can be used to get managers to reflect upon their reasoning processes. Eden (1992) and
Boland et al. (1994) exemplify this research context.
Differing contexts pose different challenges to researchers using causal mapping. In the
discovery and evocative contexts, validation of the derived causal maps by respondents
is a key requirement in generating accurate representation of maps. In the intervention
studies, derived causal maps can be used for further interpretation, or analysis, or even
consensus building by exploring the differences among respondents. In hypothesis
testing studies, reliability and construct validity assume greater importance.
Focus
Finally, researchers using causal maps as a methodological tool differ in terms of their
focus on: (a) content, (b) structure, and (c) behavior.
Content. A focus on content leads the researcher to detail the concepts in the causal
maps of respondents, and the cause-effect linkages among them. For example, Narayanan
and Fahey (1990), in their longitudinal analysis of Admiral Corporation, attributed among
other things, the absence of concepts pertaining to competition in Admiral’s causal maps
to the firm’s eventual failure. Content-focused studies can be descriptive or comparative.
In descriptive studies, the researcher may choose to describe a causal map in the
respondent’s own terms (a social constructionist perspective) or use concepts drawn
from theory or from an expert. For example, in intervention contexts, the researcher will
sometimes highlight the differences in content among individuals. In this case, the
content categories derived from the individuals can be used without alteration. Alternately,
the researchers may want to highlight the absence of significant content in a
specific firm’s causal map as a way of raising its awareness. In this case, they may recast
the causal maps using a theory or an expert causal map. In comparative analyses,
researchers compare concepts and linkages across different individuals. Here, researchers
standardize the content so that comparisons can move forward (Laukkanen, 1994).
The standardization involves the creation of a dictionary (i.e., a set of words to connote
concepts that can be used across individuals).
Structure. Some researchers are interested in the structure of the causal map. For example,
how comprehensive is the map? How focused is the map in terms of the cause-effect
relationships? Are there feedback loops in the map? Indeed network measures are often
used to operationalize the structural characteristics of the maps. For example, Calori et
al. (1994) argued that the more diversified a corporation the more complex the firm’s map
would be and found empirical evidence to support their claim.
A critical consideration for structure-focused researchers is to demonstrate the validity
of the measures they employ. Are they theoretically valid? Can one demonstrate
acceptable construct validity and reliability for the measures? For example, Nadkarni and
Narayanan (in press) demonstrated the criterion-related validity of complexity and
centrality, two network based measures of the structure of causal maps in an educational
setting. However, in causal mapping research, efforts to establish the validity of the
structural measures are still in the embryonic stages.
Behavior. Finally, some researchers are interested in the behavior of causal maps. They
ask questions such as: Can you derive what decisions will flow from a causal map given
a set of contingencies? Can you predict the decisions emanating from a causal map and
check the predictions against actual decisions? Indeed analysis of the behavior of causal
maps remains the Holy Grail for researchers using this tool in their work on managerial
cognition.