THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 142

that, irrespective of task type, tables best support simple tasks and graphs best support complex
tasks, is not borne out in the findings of their study. To fully test this statement, and therefore the
theory, the researchers needed to conduct further analyses.
Based on analyses reported in the prior section, the results that Speier and Morris (2003) report
support cognitive fit for accuracy on limiting tasks, that is, very complex symbolic tasks that can-
not be solved using analytical processes.
The studies by Speier et al. (2003) and Wheeler and Jones (2003) both demonstrate strategy
change, that is, a crossover effect, as more complex symbolic problems are better solved with spa-
tial rather than with symbolic problem representations, foregoing, therefore, the more demanding
analytical processes for the more parsimonious perceptual processes. In Speier et al., the effect
was in accuracy and not in time. In Wheeler and Jones (2003), the effects were in both accuracy
and time. On the other hand, in both of these studies graphs performed better than tables on com-
plex spatial tasks on both time and accuracy.
Hence, the findings of these studies support the theory of cognitive fit for more complex tasks.
EXTENDING THE THEORY OF COGNITIVE FIT TO NEW DOMAINS
In this section, we discuss cognitive fit in two somewhat different types of tasks that have been
examined in two well-defined domains: multi-attribute judgment tasks in the accounting domain
and multi-criteria decision-making tasks in the map-related domains of maps, geographic infor-
mation systems, and spatial decision support systems.
Cognitive Fit in Human Judgment Tasks
We identified two studies that were conducted on multi-attribute judgment tasks, both in the account-
ing domain, in our analysis of articles that used the theory of cognitive fit. Table 8.5 presents the
details. We first present the studies and then evaluate the findings.
Studies of Cognitive Fit in Human Judgment Tasks
The first such study is by Umanath and Vessey (1994). These researchers examined information
load and the ability of the theory of cognitive fit to explain the performance of certain display for-
mats on the multi-attribute judgment task of bankruptcy prediction. They hypothesized that pre-
dicting bankruptcy required both holistic processes that aided in integrating large amounts of data
(a number of financial indicators over a number of years), as well the ability to reference ranges
and/or levels of individual financial indicators. Hence they hypothesized that graphs, which pos-
sess integrating capabilities as well as permitting access to the underlying data, are a more accurate
display format than schematic (Chernoff) faces, which provide integrating capabilities but not
access to the underlying data. They also predicted that there would be no differences in accuracy
of performance between faces and tables because each provides only one of the two necessary
types of data: Tables provide no integrating capabilities while faces do not preserve the underlying
data. Similarly, they expected schematic faces to result in faster problem solving compared with
tables and graphs, for which no differences were expected.
All hypotheses were supported, although the finding for the speed of faces over tables was
marginal.