Causal Mapping in Organizational Sciences

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 

During the last three decades, the use of causal mapping in organization sciences has

increased, owing in no small extent to several developments in the field of managerial

cognition. A comprehensive review of these developments is beyond the scope of this

chapter (for a review, see Walsh, 1995). Instead, I will selectively cull out those

developments that have facilitated the frequent use of causal mapping.

The first set of studies. Arguably the first effort to introduce causal mapping into

organization sciences occurred with Bougon, Weick and Binkhorst’s (1977) examination

of the Utrecht Jazz Orchestra (UJO). Built around 14 variables obtained through naturalistic

observation, discussion, and interviews, Bougon et al. first asked each UJO

participant to indicate which variables influenced other variables, and whether the

influence was positive or negative. Later they developed “etiographs” by unfolding the

maps into content free graphs, which ranked variables into three clusters of givens,

means, and ends. Their method was not a textual analysis of the kind proposed by

Axelrod, but they made effective use of Axelrod’s ideas to build a cybernetic theory of

organizations. As the authors noted in the 1970’s, their study represented a new

approach to organizational analysis.

Next, following their footsteps, Hagerty and Ford (1984) used a modified version of causal

mapping to examine the cause-effect beliefs about structure. In their study, the researchers

presented a set of causes and effects and asked managers and students to create a

causal map. Using metrics from graph theory, they found both agreement and disagreement

between managers and MBA students.

The two studies invoked different methods of causal mapping. Bougon et al. (1977) used

naturalistic observation and interviews to examine natural phenomena. Ford and Hagerty

(1984) were primarily interested in theory testing, and therefore used an experimental

approach in their use of causal mapping.

Influence of industrial dynamics. A second stream of work invoked industrial dynamics

to examine organizational phenomena. Thus, Roos and Hall (1980) derived their inspiration

from the industrial dynamics (system dynamics) tradition to better understand

political processes within organizations. They conducted a case study of a new extended

care facility connected to a hospital to highlight the advantages of influence diagrams

by comparing the level of understanding before and after the technique was used. Roos

and Hall acknowledged Axelrod, noting that the influence diagram represented their

cognitive maps of factors influencing policies and budget levels for the extended care

unit they studied. Thus, the mapping was not as systematic as in Bougon et al.’s study,

since the primary objective of the authors was intervention-focused, or in their terms “to

help integrate knowledge about decision-makers’ values and the cause-effect of their

pursuing these values.”

Special issues. Two special issues gave a further boost to the users of the causal mapping

technique. In 1987, a special conference convened in Boston to advance the cause of

managerial and organizational cognition research. Several of the papers in this well

attended conference were later published in a special issue of Journal of Management

Studies. One of these papers featured causal mapping as a research tool. Building on

earlier conceptual (Walsh & Fahey, 1986), and empirical works (Fahey & Narayanan,

1986), Fahey and Narayanan (1989) explicitly used the causal mapping technique to trace

the evolution of Zenith, one of the then remaining US television manufacturers. They

used annual reports to capture the thinking within Zenith, and used the term “revealed

causal mapping” to distinguish what they did from cognitive mapping. Unlike Axelrod

who had access to interviews, these authors, whose longitudinal study spanned over 20

years, were not able to access many of the players for interviews and, therefore, had to

rely on archival sources of data. Fahey and Narayanan (1989) also noted that in many

competitive situations, public statements represented strategic disclosure and may not

have corresponded to the “true” cognitive maps held by the decision-makers. Unlike

cognitive maps, which represented “true” thinking, these authors were content to study

causal maps or the “assertions of causality.”

A second special issue for Organization Science was organized by Meindl, Stubbart and

Porac (1996), with the specific purpose of advancing the “cognition agenda.” The editors

noted that developments in a wide range of fields — from the sociology of knowledge

to organization science — have called into question a strictly realist view of the world.

In their opinion, even the environment should be viewed as partly contingent upon sense

making by individuals. The causal mapping technique was featured in this collection,

with Mauri Laukkanen (1994) articulating the steps involved in comparative causal

mapping (i.e., comparing causal maps among individuals). According to him, all comparative

projects have to address three critical tasks:

1) a need to acquire comparable natural data of several individuals or groups;

2) the problem of raw data conversion to achieve the necessary comparability and

pragmatic compression; and

3) the need for a rigorous and efficient computerized platform for comparative

analysis.

Laukkanen (1994) introduced the concept of “standard vocabularies” that can be used

to capture concepts with similar meaning, but denoted by different words by different

individuals. Laukkanen also provided a computer software called CMAP2 to mount the

comparative analysis. Unlike Fahey and Narayanan (1989), who relied on manual

techniques to create and compare causal maps, Laukkannen took the first steps in

hypothesis testing studies.

Managerial and organizational cognition group in the academy of management.

During the 1980’s the move to advance a cognitive agenda was gathering strength. This

culminated in the formation of the Managerial and Organization Cognition (MOC)

interest group in 1989 in the Academy of Management, the premier professional

association of management scholars. MOC was broadly based and focused on “how

organizational members model reality and how such models interact with behavior.”2 The

formation of the interest group, and its emergence as a division in 1999 within the

Academy, signaled the arrival of cognition as a major area of inquiry in management

literature, legitimizing this area within scholarly circles. For those individuals using or

intending to use causal mapping as a research tool, this development gave them a big

boost: It provided a forum to present their work, and with the competition for journal

space, their work could no longer be as easily dismissed as inappropriate.

Mapping strategic thought. In 1990, Anne Huff published Mapping Strategic Thought,

which laid the methodological foundations of the managerial cognition field. In retrospect,

no book in recent years has had more influence on the methodological aspects of

research in managerial cognition than this edited volume. Given the influence of this

book, it is worthwhile quoting Huff about (one of) her reasons for putting the book

together: “We are at the point in strategic management and other organization sciences

that significant enthusiasm for cognitive studies is in danger of outreaching its methodological

foundation. While a number of generally useful articles and books in management

fields recommend a cognitive approach… little has been written about the technical

aspects of specifying and studying cognition in organizations.”

Although the book was not limited to causal mapping methods, causal statements were

featured in four empirical studies (Huff & Schwenk, 1990; Bougon & Komocar, 1990;

Boland et al., 1990; Narayanan & Fahey, 1990). Huff and Schwenk used causal mapping

to study the attribution of success and failure by managers, raising two methodological

issues: the validation and modification of causal maps and the constancy and variability

of the maps. Bougon and Komocar drew attention to the importance of loops as the focus

of change, highlighting the “circularity” of effects caused by a set of linear relationships.

Boland et al. focused on the evolution of cognitive maps. Narayanan and Fahey extended

the adaptation metaphor to the cognitive domain, by reexamining the 20-year history of

the television receiver industry, focusing their attention on Zenith, and contrasting the

results to their earlier study of Admiral (1989).

Most importantly, Huff’s volume provided the technical details of causal mapping, and

articulated for the first time the key methodological issues that needed to be tackled by

serious researchers. These included: the purpose of a causal map, the map’s territory,

sources of data, and sampling, reliability, and validity. Huff and Fletcher (1990) concluded

on a very optimistic note:

“Cognitive maps, as artifacts of human reasoning can be used to study virtually

any question raised by those who are interested in human activities… Our view

is that…it is often most attractive as a method for studying topics that are

intrinsically cognitive for explaining variance that is unexplained by other

methods.”

There is no doubt that Huff’s book served to encourage hesitant researchers. It also

became the textbook of choice for training a future generation of doctoral students.

Eden and Spender’s Managerial and Organizational Cognition. Huff’s volume was

dominated by scholars of the U..S. tradition. By 1980, researchers in Europe were

becoming increasingly interested in cognition. To showcase the European works, Eden

and Spender (1998) edited a book based on the works initially presented at a Managerial

and Organization Cognition research workshop held in Brussels in 1994. According to

the authors,“In the past few years we have seen… Organization Science’s special issue

(1994), Mapping Strategic Thought (Huff, 1990) and new JAI series, Advances in

Managerial Cognition and Organizational Information Processing. … The present

volume explores these questions, but unlike the works cited above, reflects a more

European view — even though one European author appears in both places.”

The book featured several chapters on causal mapping, three of which are noteworthy.

First, Laukkanen succinctly summarized his ideas on comparative causal maps. Second,

Jenkins summarized the key methodological challenges in comparing causal maps. Third,

Eden and Ackerman described techniques used to analyze and compare idiographic

causal maps. The book signaled the era of convergence and cross fertilization of ideas

across the Atlantic.

Network studies. By 1990, the study of social networks had reached a level of maturity

in sociology, with attendant analytical tools, software and particularly measures. Most

researchers using causal maps understood that causal maps in the matrix representation

form can benefit from the work done in social networks. They borrowed network measures

because they were available, but initially did not pioneer the development of new

measures. This task was left to scholars working at the intersection of social networks

and computer science. Thus, following the quantitative tradition at Carnegie Mellon

University, Carley and her colleagues developed numerous measures of causal maps at

several levels of analyses, and created computer programs to analyze the maps. Although

many of the network-based measures are underutilized at this time, the availability of

computer software should facilitate the easy adoption of these measures.

In summary, as shown in Figure 1, over the last two decades, we have witnessed

significant developments in the use of causal mapping. Three significant trends have

contributed to this progress. First, there has been a joining of three disciplines — the

quantitative disciplines such as industrial dynamics, organization sciences and social

sciences. Second, there has been greater international convergence, with U.S. and

European researchers coming together under the auspices of the Academy of Management

to push the frontiers of this method. Finally, computer software has proliferated,

making it easier for researchers to use and analyze causal maps.

In some ways, the use of causal mapping in the IS field is not new. Adherents to both

the social science and operations research traditions in the organizational sciences

sketched above have, over the last two decades, employed causal mapping in the IS field.

These traditions respectively focused on two related problems:

• How do we use causal mapping to generate consensus, either in understanding or

developing problem definitions?

• How do we use causal mapping to find solutions to specific technical problems?

Figure 1. Causal mapping

Precursors

Social Network

Methods

Axelrod

Weick

Huff,

Mapping

Strategy

Thought

Eden

Kosko

Theory

testing

Hagerty & Ford

Operations

Research

Information

Systems

Intervention

Boland & Tenkasi

Research

Fuzzy Causal Maps

Fahey &

Narayanan,

JMS

Nelson et al.

2000 onward

Statistically Based

Research

Advances in Software

Development

Expert Intervention

Carley

1980-2000

Organization Science

Early Influences:

Structure of Argument

Industrial Dynamics

In the first tradition, Boland et al. (1994) illustrated an intensive IT augmented approach

to causal mapping to facilitate a hermeneutic process of inquiry. Causal maps then

became a tool by which participants could glean and appreciate the logics in use by others

in their organizations, and through a process of dialogue could develop a consensus of

how to interpret their world. Similarly, Zmud et al. (1993) demonstrated the use of mental

imagery in requirements analysis. Here, the focus was on developing a consensus in

defining the problem — for example, the requirements of an IS system — so that the actual

design would respond to the requirements and thus make the implementation less

chaotic.

In the second tradition, causal mapping is used to arrive at solutions to specific, often

technical, problems. As an example, Irani et al. (2002) used cognitive mapping to model

various IT/IS factors, integrating strategic, tactical, operational, and investment considerations.

The authors demonstrated how the causal mapping technique can capture the

interrelationships between key dimensions identified in investment evaluation — something

other more commonly used justification approaches cannot accomplish. Thus they

claimed that causal mapping can be use as a complementary tool in project evaluations

to highlight interdependencies between justificatory factors. I hasten to add that

although this use of causal mapping has been less frequent in the literature, it offers great

promise in the future.

During the early days, irrespective of tradition, the use of causal mapping in IS was

application focused, i.e., to solve managerial problems in organizations. This began to

change during the new millennium, with a special issue of Management Information

Science Quarterly (MISQ) which dealt with qualitative methods of IS research. The issue

featured causal mapping in a paper by Nelson et al. (2000). The paper not only provided

a tutorial in causal mapping but demonstrated the possibility that in the IS field, causal

mapping can be used in “evocative” research contexts. By “evocative,” these authors

referred to research contexts in which general theories were available to represent the

phenomena under study, but the operationalization of theories to the respective contexts

was not yet developed.

During the last four years, after the publication of the MISQ piece, there has been growing

interest in the use of causal mapping in IS, not merely for qualitative studies, but for

hypothesis testing studies as well. For example, Armstrong (2003) coupled causal

mapping and survey data in a study of IS experts in Object Oriented and Procedural

Programming. Similarly, a SIG-CPR3 workshop on causal mapping organized in 2003 in

Philadelphia drew an audience of over 25 participants.

Above, I have sketched the evolution of causal mapping to highlight both the growing

acceptance of this tool for research among scholars drawn from different disciplines, and

also to segue to the diversity of approaches to using this technique. I now turn to this

diversity.