THE THEORY OF COGNITIVE FIT - Human-Computer Interaction and Management Information Systems: Foundations - page 160

type of steps involved in acquiring information from a graph; the steps, in increasing order of
complexity, are “identification, scan, comparison, and estimation”; (2) the effort required to iso-
late the relevant image from the rest of a data display; note that this issue is essentially one of
proximity, a variable introduced in this article under multi-criteria decision making in map-
related domains; and (3) polarity, which refers to questions that require yes or no responses. For
their research, they emphasize the importance of proximity.
Note, however, that this approach begs the question of the processing strategy used for task
solution. Applying cost-benefit principles to graph/table decision making shows unequivocally
that problem solvers may use a number of different strategies to solve more complex problems.
Processing strategy will therefore influence significantly the number of steps taken in the prob-
lem solution. 3 Hence any study that uses Addo's approach should also incorporate analyses of
appropriate processing strategies.
• Smelcer and Carmel (1997) suggest applying the approach used by Card, Moran, and Newell
(1980, 1983; keystroke-level model, GOMS) to evaluate the number of knowledge states for
each task, “which helps to explain problem-solving times.” 4 They point out that when prob-
lem solvers use maps, they can reduce the number of knowledge states needed for problem
solution by using visual heuristics. On the other hand, those problem solvers using tables rely
on exhaustive algorithms to find the correct solution from among the knowledge states in the
mental representation. Note the similarity of these arguments to those for the use of percep-
tual and symbolic processes in the theory of cognitive fit. Smelcer and Carmel acknowledge
that to apply Card et al.'s approach to determining the number of knowledge states that need
to be traversed for effective task solution, one must understand the process used—for exam-
ple, by conducting protocol analysis. Their analyses succeed in explaining problem-solving
times quite well.
• Lohse (1993) advocates using a “computational cognitive engineering approach for quanti-
fying the effectiveness, efficiency and quality of graphic designs,” presumably based on eye
movements. Lohse, however, tested his approach using very simple tasks, so simple, in fact,
that he reports that “tables were always faster than graphs regardless of the task.” It is not
known whether Lohse's approach would scale to more complex tasks.
It is interesting to note that Lohse's claim about the speed of problem solving is the reverse
of that made by Dennis and Carte (1998), who claimed that speed is always faster for graphs
than for tables. The difference, of course, lies in the complexity of the tasks under examination.
Second, one of the more interesting directions of the research analyzed here is that involving
new dimensions of fit. Future research could seek to determine new dimensions systematically.
One might approach this task by first identifying different types of tasks or different characteris-
tics of tasks that might be a target for matching. This approach could be difficult, however,
because there is no theory of tasks on which such an investigation might be based. That may mean
that new dimensions of fit will be uncovered by analyses of prior or existing research, as was done
here, a tedious and time-consuming approach.
Third, recall that our extended model of cognitive fit (Figure 8.4) allows the researcher to con-
sider independently the roles and nature of the internal representation of the problem domain, the
external problem representation, and the mental representation for task solution. Hence, this model
opens the way to investigating the nature of both the internal and mental problem-solving represen-
tations in cognitive fit. These representations may be investigated in two ways: (1) by examining the
knowledge structures that problem solvers form during problem solving; and (2) by examining the