THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 138

Evaluation of Cognitive Fit in Simple Information Acquisition
and Information Evaluation Tasks
The findings of these three studies largely support the theory of cognitive fit. The exceptions are
the findings by Vessey and Galletta that tables are more accurate than graphs on spatial problems,
and those of Chan that graphs and tables are equally effective in both their nominal and high infor-
mation load conditions.
It appears likely that the lack of support for cognitive fit in spatial problem solving in Vessey
and Galletta's (1991) study is due to the fact that the manipulation of spatial tasks was not suffi-
ciently strong. A significant body of knowledge supports the notion that problem solvers are more
familiar with tables than with graphs, including findings from this study itself. Hence, it is likely
that participant performance, in general, is better with tables than graphs. On very simple tasks,
the effects of cognitive fit may not be sufficient to overcome this effect. One would, however,
expect to find the expected effects on somewhat more complex tasks. This situation would lead to
some results supporting cognitive fit, those on the more complex tasks of this type, no effects in
somewhat less complex tasks, and reverse effects on extremely simple tasks. It appears, then, that
the overwhelming majority of studies analyzed by Vessey (1991) in formulating the basic theory
of cognitive fit, which demonstrated that spatial representations were both more effective and
more efficient in their support of spatial tasks than were symbolic representations, were somewhat
more complex spatial tasks.
It also appears likely that Chan found no interaction effect between problem representation and
information load also because the tasks were too simple, notwithstanding the fact that he investi-
gated two levels of information load.
Cognitive Fit in More Complex Tasks
Many of the studies reviewed in this paper address more complex tasks. What distinguishes the
studies presented in this section is that they represent direct tests of the effect of complexity on cog-
nitive fit, hence testing certain aspects of the theory of cognitive fit presented in Vessey (1994). The
studies examined, therefore, also include those that demonstrate the occurrence of strategy shift as
complexity in symbolic tasks increases. We identified four studies that addressed the match of the
problem representation to the task on more complex tests. Table 8.4 presents the details. We first
present the studies and then evaluate the findings.
Studies of Cognitive Fit on More Complex Tasks
In the first experiment in this category, Wilson and Addo (1994) address some of the issues pre-
sented in Vessey (1994). They hypothesized that both task type and task complexity (in the form of
question complexity) would influence performance with graphs and tables. In an experiment that
varied display format (graphs and tables), task type (spatial and symbolic), and task complexity
(low and high), the researchers tested two separate hypotheses related to the display format x ques-
tion type interaction and the display format x question complexity interaction, respectively.
There were no significant differences in accuracy across any of the treatments. The researchers
argue that their participants “set relatively inflexible accuracy standards and used the amount of
time necessary to meet those standards” thereby causing the effects to be manifested in response
time. They report that their findings for response time support both (1) the theory of cognitive fit,
with graphs best supporting spatial tasks and tables best supporting symbolic tasks, overall; and
(2) their task complexity notions; that is, irrespective of task type, low complexity problems are