Introduction
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
The main purpose of this chapter is to demonstrate the use of evidential reasoning
approach under Dempster-Shafer (D-S) theory of belief functions (Shafer, 1976; see also,
Srivastava & Datta, 2002; and Srivastava & Mock, 2000, 2002) to analyze revealed causal
maps. The Revealed Causal Mapping (RCM) technique is used to represent the model
of a mental map and to determine the constructs or variables of the model and their
interrelationships from the data. RCM focuses on the cause/effect linkages disclosed by
individuals intimately familiar with a phenomenon under investigation. The researcher
deliberately avoids determining the variables and their associations a priori, allowing
both to emerge during the discourse or from the textual analysis (Narayanan & Fahey,
1990). In contrast, other forms of causal mapping begin with a framework of variables
based on theory, and the associations are provided by the participants in the study (cf.
Bougon, et al., 1977).
While RCM helps determine the significant variables in the model and their associations,
it does not provide a way to integrate uncertainties involved in the variables or to use
the model to predict future behavior. The evidential reasoning approach provides a
technique where one can take the RCM model, convert it into an evidential diagram, and
then use it to predict how a variable of interest would behave under various scenarios.
An evidential diagram is a model showing interrelationships among various variables in
a decision problem along with relevant items of evidence pertaining to those variables
that can be used to evaluate the impact on a given variable of all other variables in the
diagram. In other words, RCM is a good technique to identify the significant constructs
(i.e., variables) and their interrelationships relevant to a model, whereas evidential
approach is good for making if-then analyses once the model is established.
There are two steps required in order to achieve our objective. One is to convert the RCM
model to an evidential diagram with the variables taken from the RCM model and items
of evidence identified for the variables from the problem domain. The second step is to
deal with uncertainties associated with evidence. In general, uncertainties are inherent
in RCM model variables. For example, in our case of IT professionals’ job satisfaction,
the variable “Feedback from Supervisors/Co-Workers” partly determines whether an
individual will have a “high” or “low” level of satisfaction. However, the level of job
satisfaction will depend on the level of confidence we have in our measure of the variable.
The Feedback from Supervisors/Co-Workers may be evaluated through several relevant
items of evidence such as interviews or surveys. In general, such items of evidence
provide less than 100% assurance in support of, or negation of, the pertinent variable.
The uncertainties associated with these variables are better modeled under Dempster-
Shafer theory of belief functions than probabilities as empirically shown by Harrison,
Srivastava and Plumlee (2002) in auditing and by Curley and Golden (1994) in psychology.
We use belief functions to represent uncertainties associated with the model variables
and use evidential reasoning approach to determine the impact of a given variable on
another in the model. This combination of techniques adds the strength of prediction to
the usefulness of descriptive modeling when studying behavioral phenomena. Evidential
reasoning under Dempster-Shafer theory of belief functions thereby extends the
impact of revealed causal mapping.
The chapter is divided into eight sections. Section II provides a brief description of the
Revealed Causal Mapping (RCM) technique. Section III discusses the basic concepts
of belief functions, and provides an illustration of Dempster’s rule of combination of
independent items of evidence. Section IV describes the evidential reasoning approach
under belief functions. Section V describes a causal map developed through interviews
and surveys of IT employees on their job satisfaction. Section VI shows the process of
converting a RCM map to an evidential diagram under belief functions. Section VII
presents the results of the analysis, and Section VIII provides conclusions and directions
for future research.