Diversity of Approaches Among Users of Causal Mapping

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
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.


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


Phenomena can be


accurately as a

causal map


causal maps

shape their





Primarily for

physical phenomena

or deterministic

social phenomena






Appropriate for

Locating the experts,

and accurately

representing their

casual maps

Establishing the

correct causal map Accuracy of

representation of


causal map




Social Objectivist Expert-Anchored


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



stakeholders and





sampling drawn by



Participants in Experts

the system


As an input to

decision making

Obtaining relevant





concepts and



Applicability of



Can vary from

undeveloped to

fully developed

Both theory and



General theoretical

framework available;

No operationalization

State of Theory Undeveloped

Hypothesis Intervention


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.


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