THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 144

Tuttle and Kershaw (1998) conducted a multi-attribute judgment study based on the Umanath and
Vessey study (1994). In this case, participants assessed performance of hypothetical plant managers,
a holistic task, using both graphs and tables. The researchers make similar arguments to Umanath and
Vessey, above, with respect to the effects of graphs and tables on both accuracy and speed of per-
formance on their judgment task. With respect to processing strategy, Umanath and Vessey relied on
an emergent property of the theory of cognitive fit with respect to the type of strategies that decision
makers use when working with graphs and tables, a property that was shown to be valid in Vessey and
Galletta's (1991) paper. Tuttle and Kershaw, on the other hand, manipulated strategy by encouraging
their participants to use either a holistic strategy or an analytical strategy. Following completion of the
judgment tasks, the researchers confirmed the effectiveness of their judgment strategy manipulation.
Participants used both holistic and analytical strategies to solve judgment problems using either
graphs or tables; that is, strategy was a within-subjects variable, while presentation format was a
between-subjects variable. All hypotheses involved between-subject issues; that is, analyses com-
pared just the performance with graphs versus tables using holistic strategies and using analytical
strategies. This approach is peculiar because within-subjects designs are generally used to increase
the level of control (and therefore the statistical power) over the experimental variables of interest—in
this case, processing strategy. No analyses were conducted, however, on the within-subjects factor
of processing strategy.
The dependent variables were judgment accuracy (both consistency and “judgment model qual-
ity”) and time. For the holistic strategy, graphs produced higher judgment accuracy manifested in
both more consistent judgments and higher judgment model quality, than tables. There was no
difference in judgment time between the two formats. The researchers concluded that display
format impacted accuracy through both judgment consistency and judgment model quality, and
not through judgment time.
For the analytical strategy, there was no difference in judgment accuracy, consistency of judg-
ments, or judgment model quality between tables and graphs; tables, however, proved to be quicker
than graphs. The researchers claim that participants achieved similar accuracy with graphs and
tables by exerting more effort for the format that did not match the strategy (graphs).
By manipulating processing strategy across presentation format, the researchers created a match-
mismatch situation of strategy with presentation format. However, they did not take into account the
characteristics of the task they were addressing, one of the primary variables in the theory of cogni-
tive fit. Use of the holistic strategy resulted in a match with the judgment task and the graphical for-
mat, and a mismatch with the table format. Hence, the findings for the holistic strategy are consistent
with cognitive fit as the match between the task and the problem representation. Use of the analyti-
cal strategy resulted in a mismatch with both the judgment task and the graphical format and a match
with the table format. With graphs, decision makers needed to transform spatial information into
symbolic information, process that data analytically, and then make a holistic judgment. With tables,
a similar process applies without the initial need to transform the data in the presentation format.
Hence it is possible that decision makers with graphs expended considerable time to transform the
data to a level comparable to that of the data presented in tables, which then resulted in no differ-
ences in accuracy between the two formats.
Evaluation of Cognitive Fit in Human Judgment Tasks
Given the above explanations of the findings of the study by Tuttle and Kershaw (1998), the find-
ings of the two studies described in this section can be explained quite readily using the theory of
cognitive fit.