THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 141

alternatives increases, decision makers increasingly use perceptual processes that are supported
by spatial representations instead of analytical processes supported by symbolic representations.
Hence, the authors observed a change in processing strategy as complexity on the symbolic task
increased. The type of task investigated here is therefore a limiting task (Johnson and Payne, 1985;
Vessey, 1994). Hence, the hypotheses in this experiment can be explained by the theory of cogni-
tive fit and the findings for accuracy support those notions.
With regard to their findings for time, in the high complexity condition, decision-making time
using the text-based interface may have been lower than that for the visual interface simply
because the participants could do little to respond to the task effectively using the text-based inter-
face. Hence, whatever they did, they did very quickly. In the low complexity condition, tables,
although more accurate than graphs, may have required more time because participants still had
to review a significant amount of data, a task that can be achieved more quickly with a visual than
with a text-based interface.
In the third study in this category, Speier, Vessey, and Valacich (2003) investigated the role of
cognitive fit in the solution of both complex spatial and symbolic tasks in both the presence and
absence of interruptions. In the case of spatial tasks, there were no interaction effects for either accu-
racy or time: As expected, graphs outperformed tables on complex spatial tasks on both accuracy
and time. In the case of symbolic tasks, interruptions effectively increased the cognitive complexity
of the task under investigation and were therefore expected to induce problem solvers to change to
a more parsimonious strategy earlier than an environment without interruptions. In other words,
with an appropriate choice of task complexity, the crossover effect described above can be observed.
The researchers found that the effect was manifested in accuracy and not in time. Problem solvers
were more accurate with tables than with graphs on their complex symbolic tasks when there were
no interruptions. However, with interruptions, there was no difference in performance with graphs
and tables. Hence, graphs permitted problem solvers to handle the increased cognitive complexity
better than did tables.
The final study in this category is that of Wheeler and Jones (2003). In a study of bank loan
decision making, the researchers address the choice, and suitability of the choice, of certain “deci-
sion aid features” in terms of performance on tasks of “high and low predictability.” High and low
predictability tasks were defined by the amount of data to be analyzed, which is, in essence, a
measure of task complexity. The task is symbolic in nature and the data are discrete (debt ratio,
cash flow, revenue trend—number of quarters out of the prior 20 that revenues increased). A data-
base aid presented the history data in point form (symbolic problem representation), while a
regression aid depicted the data visually (spatial problem representation). Hence both “decision
aid features” shared the same history data; only the presentation of that data varied. These “aids”
are referred to, henceforth, as “problem representations.”
Wheeler and Jones (2003) hypothesized that participants would perform better with the data-
base representation (symbolic) in high predictability (low complexity) environments, and better
with the regression representation (spatial) in low predictability (high complexity) environments.
Their findings support their hypotheses. Hence the study by Wheeler and Jones (2003) is a further
example of strategy change on a symbolic task that occurs as complexity increases and the problem
can no longer be solved analytically.
Evaluation of Cognitive Fit in More Complex Tasks
As noted above, the study by Wilson and Addo (1994) goes quite a way towards supporting the
extended notion of cognitive fit based on task complexity (Vessey, 1994). However, their statement