The Comparative Analysis of Ideographic Cause Maps

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.