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
Our search for an analysis approach that would assist this reality led us to causal
mapping. The purpose in selecting causal mapping is not oriented toward defining central
tendencies in the data, but rather toward documenting the range of possibilities. In other
words, we believe it is too early to test the relative frequency of three theorized causes
of UML success when we expect there may be four, five, six, or more potential causes that
have not yet been documented. We would consider it too early to even attempt to
conclude that in some precise percentage of situations UML increased positive development
outcomes by some specific amount. Rather, we use causal mapping to illustrate
the range of variables that could influence decisions to implement UML as well as the
outcomes should UML be implemented.
The employment of causal mapping in our research setting would differ from most others.
For example, Nelson et al. (2000) use the method to define a fairly narrow set of constructs,
variables, relationships, and relationship strengths. Although we expected to find
constructs, variables and relationships, it is not our goal to demonstrate that our causal
map analysis outcome would be reliable across research settings. Rather, we were looking
for indications of these as they emerge from the data. Also, we believe that causal
mapping would document missing connections (e.g., sometimes it is valuable to know
what elements should be absent as well as present in an equation) and bring to light
relationships that may not be linear, where a construct at one level may cause negative
results, at another level have no impact, and at still another level cause positive results.
In conclusion, although we expected to find the same elements as in other research
employing causal mapping our objective was not reliability but validity with regard to
mirroring the complexity of organizational approach to UML.
Our use of causal mapping has similarity with Fahey and Narayanan (1989). We seek to
explore a complex phenomenon (UML) in the wider organizational setting. Our use of
causal mapping can be understood in the context of the basic research method challenge
of the personal pronouns of we, I, and you (Bohman, 2000). “We” represents the objective
approach to research — as in a deductive approach. “I” represents the individualized
approach to research — as in an inductive approach. “You” represents the real world
subjects and their views, attitudes or behaviors. Obviously, researchers using the voice
of “we” or “I” attempt to express salient aspects of “you.” Additionally, researchers
employ a method within which aspects of “you” are studied and analyzed.
The particular concern in this research was that causal mapping strongly emphasizes a
clear definition of the two elements of construct/variable and causal relationship
between constructs. Used in a straightforward and narrow manner causal mapping
represents the voice of “we” — a deductive and objective approach to research, as in
Nelson et al. (2000). The present researchers argue that subjects — the “you” in research
— may possess clear views of constructs/variables and causal relationships. However,
the presumption, in Nelson et al.’s (2000) research is that these will be parsimonious on
the individual level and across individual roles. We believe, however, that our domain
of interest, UML within the organizational context, does not have the pre-existing
background of prior investigation to allow for identification of those constructs/
variables and causal relationships before examining data, but rather using the data as the
source for discovery of these.
Therefore, our data collection method was not intended to define a set of constructs/
variables and causal relationships, as in Fahey & Narayanan (1989). Doing so would have
the danger in leading respondents to fill in the suggested topics without considering
whether the set of topics is complete. That is, since very many constructs/variables and
relationships exist in the real world, directing subjects to talk about a pre-selected subset
of these phenomena may lead to definitions that reflect the pre-selection rather than
represent the real world. In this scenario, the outcome might appear clean, but actually
hide the fact of missing elements. Therefore, given the lack of prior research guidance,
in this research subjects were encouraged to talk freely while guided toward the present
research focus. Our approach should and must be closer to the spirit of having “you”
talk as “you” find appropriate. Data collected in this manner would, hopefully, be a good
source for defining constructs/variables and causal relationships with a high degree of
validity.