ADVANCING THE THEORY OF DSS DESIGN FOR USER CALIBRATION - Human-Computer Interaction and Management Information Systems: Foundations - page 65

user calibration as a result of visibility and expressiveness treatment levels (F (2,37) 0.635).

The Bonferroni minimum significant difference of 0.0661 exceeds the 0.016 difference in means
(0.116-0.100). In this case, H 0 cannot be rejected. The data indicate that when problem novelty was
lower and problems were more familiar and less novel, there was no difference in user calibration
between subjects using visibility diagrams and those using expressive text.
Analyses also showed no significant difference in user calibration due to VVIQ subject differ-
ences (questions 1-4, F (1,37)

.229, p

0.71) or deci-
sion time. These results add to the generalizability of the main finding that visibility improves
user calibration when problems are new and somewhat novel.

2.57, p 0.12, questions 5-10, F (1,37) .14, p

SUMMARY AND CONCLUSIONS

To frame the theory of DSS design for user calibration as a dialog design theory, this paper began by
reviewing the general concepts and components of DSS. Next, we drew extensively upon Kasper
(1996) to describe the theory. We began by describing the kernel theories. Next we present the prop-
erties needed to achieve the goal of perfect calibration, including meta-design and critiquing.
Based on these ideas a hypothesis was posited. A test of this hypothesis provided a partial test
of the theory of DSS design for user calibration. Specifically, the laboratory study compared the
effects of expressiveness and visibility on user calibration at two levels of problem novelty. The
results of this study supported the theory. When problems were new and novel, visibility diagrams
significantly improved user calibration compared to expressive text. Later, when problems became
more familiar, less novel, there was no difference in user calibration between visibility and expres-
siveness treatment levels.

IMPLICATIONS FOR FUTURE DSS RESEARCH AND PRACTICE

Although the research reported here is a beginning, focused on relatively simple forms of expres-
siveness and visibility, the long-term goal of this research program is to identify and investigate
the effect of forms of expressiveness, visibility, and inquirability on user calibration under differ-
ing levels of problem novelty. At a minimum, this finding of a significant result encourages more
empirical research into these effects.
At this first-study stage in this research program, recommendations and generalizations are a
stretch. However, it can be said that developers of DSS must be increasingly conscious of visual
imagery and visual computing as both influence decision quality (Gonzalez and Kasper, 1997;
Speier and Morris, 2003) and user calibration, and that these effects have been demonstrated when
the problems being addressed are episodic, new and novel to the decision maker. These results help
move the visual computing paradigm from one of engaging and entertaining the user to one of deci-
sion supporting and improving user calibration in which the decision maker gains a better apprecia-
tion for the quality of the decision he or she is making using the DSS (Shneiderman, 1992).
How the DSS itself should shift among forms and depictions of expressiveness, visibility, and
inquirability in consideration of problem novelty to achieve perfect calibration is an essential
research stream. A DSS dialog could be a static architecture with fixed forms of the three modes
corresponding to a static level of problem novelty, or the dialog could consist of a dynamic archi-
tecture that would shift among the three modes varying its presentation with respect to the level of
problem novelty involved. Wood, reported by Johnson-Laird (1993), confirmed experimentally
that subjects attack problems through imagery in the initial phase of problem solving, and as a con-
sequence of increasing familiarity with the problems, they subsequently switch over to linguistic