DECISIONAL GUIDANCE - Human-Computer Interaction and Management Information Systems: Foundations - page 84

Montazemi et al. (1996) compared suggestive guidance, informative guidance, and no computer-
based support for two tasks, one with low complexity and the other with high complexity. For the
task with low complexity, users given suggestive guidance outperformed those with informative
guidance, who outperformed those without DSS. For the high complexity task, those with inform-
ative guidance still outperformed those without DSS, but users who received suggestive guidance
performed comparably to those without DSS. The provocative findings concerning suggestive
guidance are discussed in the next section on task differences. The findings with respect to inform-
ative guidance are less surprising, but analyzing the paper raises questions about defining and oper-
ationalizing informative guidance. Determining the value-added content of the informative
guidance is a bit difficult without seeing the system, especially since the study did not have a DSS
without any guidance. But as described in the paper, the informative guidance seems to be not so
much guidance, as defined here, as a decision aid in the classical sense. Since many decision aids
do provide people with information to enlighten their decisions, drawing the line between a pow-
erful decision aid without guidance and an aid with informative guidance may be difficult.
Consider the hypothetical case of a simple decision aid that supports an elimination by aspects
strategy for renting an apartment or buying a television set. The aid provides a matrix of alternatives
and their attributes. Users are prompted to identify an attribute and cutoff value, which the aid uses to
eliminate some of the alternatives. The information provided by the system would not be considered
guidance. The matrix contains the basic information associated with the task. And the information is
not focused on the discretionary choices the user makes (selecting an attribute and cutoff). Indeed, if
the data matrix qualifies as informative guidance, then virtually every decision aid in the world would
be considered informative guidance. Yet, some users might use this information in selecting a cutoff
value, scanning the values of a particular attribute to choose an appropriate cutoff. So this information
might well enlighten or sway the user in executing the process. And what if the decision aid were
enhanced to provide statistical summaries of the values for each attribute (maximum, minimum,
mean, median, and quartiles) to facilitate arriving at cutoffs? It would be easy to view this derived
data, too, as part of the aid's basic functionality. But this feature has also been characterized as inform-
ative guidance (Silver, 1991b, p. 177). Where does one draw the line between basic functionality (and
its information base) and informative guidance? The typology requires clarification of this issue.
Jiang and Klein (2000) studied decisional guidance for choosing a forecasting method where the
choice among methods was structured as a multi-attribute choice problem. Informative guidance
provided ratings for each of the four forecasting methods with respect to six attributes such as con-
sistency and simplicity. Suggestive guidance allowed users to provide weights for each attribute and
generated weighted scores for each forecasting method so users could see which scored highest. The
study found that the form of guidance had a significant effect on which multi-attribute strategy users
followed to select a forecasting method. Moreover, shifts in which multi-attribute selection strategy
was employed often were accompanied by shifts in which forecasting method was selected. As with
Montazemi et al. (1996), this study also highlights the possible blurriness in distinguishing between
basic functionality and decisional guidance. Since the immediate task was a multi-attribute problem,
the guidance provided—attribute scores and a weighting mechanism—might be seen as part of the
decision aid rather than as additional guidance. But from the perspective of the ultimate task of per-
forming a forecast, support for choosing a forecasting method can be seen as guidance.
Task and Individual Differences
Only two of the studies address individual or task differences that may affect the consequences of
guidance. The limited attention is not surprising given the relative newness of guidance studies