The Comparative Analysis of Ideographic Cause Maps
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
In cases where individual maps have been elicited using ideographic techniques, the
process of deriving ‘shared maps’ in an attempt to capture collective cognition is
infinitely more complex. The aforementioned aggregation procedure, for example, by
necessity must involve an additional, preliminary stage, in which the various constructs
upon which causal judgments are to be combined are first pooled, prior to summation or
averaging (e.g., Nelson et al., 2000). The procedure known as composite mapping (Eden
et al., 1983) requires individuals to first describe their own (idiosyncratic) causal beliefs.
Next, they are presented with the causal maps elicited from other participants, following
which a (single) composite map, one that contains all the concepts and relations found
within the individuals’ maps is compiled. Finally, through a process of negotiation
between the researcher(s) and participants, there is an attempt to build a ‘team map,’ that
is, a map that reflects the views of the participants as a collective. In practice, the ability
to derive maps that are acceptable to participants on a group basis has proven far from
straightforward, to the extent that Huff and Fletcher (1990, p.405) find it necessary to
advocate “decision rules for handling inevitable inconsistencies.” However, as was well
illustrated in Langfield-Smith’s (1992) study, even gaining consensus with as few as six
group members can prove to be impossible.
An alternative approach to the analysis of collective beliefs entails the identification of
common elements among diverse causal maps that serve to link participants’ beliefs (Hall,
1984). Laukkanen (1994), for example, operationalized collective cognition using a variant
of this technique by first deriving separate causal maps for each individual, in similar vein
to the earliest stages of the composite mapping procedure outlined above. Next, he
assessed the overall level of commonality, i.e., agreement among the individual maps by
identifying synonymous terms, which he then standardized prior to incorporating these
within a higher-level map, depicting the collective view of his participants.
Recent Advances in the Large-Scale Comparative
Analysis of Cause Maps
A potential criticism of nomothetic elicitation methods in the context of explorations of
collective cognition is that, by constraining choice, they might potentially lead to a
greater convergence of responses than free response methods, by virtue of the standardized
variables employed in the elicitation process (Daniels et al., 2002). Conversely, as
noted earlier, ideographic methods may increase the divergence among cognitive maps,
this being an artifact of the demand characteristics of the elicitation processes, which
tend to accentuate surface-level triviality in the resulting maps (Hodgkinson, 1997b,
2002), although it is by no means inevitable that they will do so (cf., Daniels & Johnson,
2002).
Within the past decade or so, a number of researchers have sought to capitalize on the
strengths of ideographic and nomothetic elicitation procedures, while dispensing with
some of their associated weaknesses, through the development and use of hybrid
techniques (e.g., Hodgkinson, et al., 1999; Hodgkinson & Maule, 2002; Markóczy &
Goldberg, 1995). These techniques require participants to select from a comprehensive
pool a subset of constructs to be mapped, ones that are personally salient, thereby
satisfying the twin imperatives of meaningfulness of the research task and data comparability.
By far the most comprehensive of such hybrid procedures to date is that devised
by Markóczy and Goldberg (1995), which totally obviates the need for subjective
researcher judgment in making such comparisons:
1. Develop a pool of constructs by conducting and analyzing interviews with
[representative participants] and a review of relevant literature. This is done prior
to the study so that each [participant] selects constructs from the same pool.
2. Have each [participant] select a fixed number of constructs by identifying items
from a constant pool of constructs.
3. Construct the causal map of each individual [participant] by having her/him assess
the influence of each of her/his selected constructs on her/his other selected
constructs.
4. Calculate distance ratios between causal maps using a generalized version of
Langfield-Smith and Wirth’s (1992) formula.
5. Perform a variety of statistical tests on the distance ratios to identify what
characteristics account for similarities in thinking.
The distance ratios derived from this procedure can be meaningfully employed in order
to investigate patterns of similarity and difference among subgroups of participants, in
addition to conducting correlational analyses (for substantive applications, see Markóczy,
1995, 1997, 2001). As discussed in the Appendix to this chapter, recent advances and
ongoing developments in computerized systems for the elicitation and analysis of causal
maps are placing this relatively sophisticated approach within easy reach of virtually any
potential user.
Psychometric Issues
As with cognitive mapping techniques in general, users of causal mapping procedures
have tended to downplay reliability and validity issues (Huff, 1990), a fundamental
prerequisite for the advancement of any social scientific field. Hodgkinson (2001) has
discussed the psychometric proprieties required of cognitive maps more generally
(including causal maps), both those elicited directly from participants and those elicited
from secondary data sources and interview transcripts. The material presented in this
section develops and extends the arguments and recommendations put forward in that
earlier publication. Our discussion in this section is necessarily technical, focusing on
the statistical requirements for ascertaining the reliability and validity of causal maps.
It is convenient to introduce this material at this juncture because it is highly relevant
to all stages of the mapping process, not only the elicitation of raw data, but also the
construction, analysis and comparison of causal maps.
As noted earlier, one of the major strengths of direct elicitation procedures is that they
obviate the need for a two-stage approach to map construction, the maps emerging
directly from the elicitation process. In contrast, indirect methods require a considerable
amount of additional effort on the part of the researcher, in that the causal maps first have
to be identified through elaborate coding procedures, prior to the computation of basic
structural indicators and other metrics for capturing the characteristic features of the
maps. In turn, this further complicates the process of establishing the reliability and
validity of the resulting outputs. Accordingly, we begin this discussion of reliability and
validity issues with a consideration of the more straightforward case of maps elicited by
direct means.