DECISIONAL GUIDANCE - Human-Computer Interaction and Management Information Systems: Foundations - page 98

type of guidance over the other. They could simply implement the type of guidance that is consis-
tent with their motives and the means available to them. One might conclude, therefore, that inform-
ative guidance should be studied separately from quasi-suggestive and suggestive guidance.
Similarly, one might conclude that making comparisons across these forms of guidance would not
likely be fruitful and that applying conclusions from one form to the other would be risky.
But this need not be the case. At least three factors could cause a designer to implement
informative guidance despite having the motive and means to direct the user. These are (1) con-
cerns that users will reject suggestions if they feel their independence is compromised, (2) con-
cerns that users will have negative perceptions about overly directive systems, and (3) concerns
that users may learn less about the task when handed a solution. The same three concerns might
explain why a designer might choose quasi-suggestive guidance over suggestive guidance. So
comparing informative guidance with the more directive forms of guidance in terms of these con-
cerns is worth pursuing. In short, directivity may be the key dimension to consider when studying
all three types of effects: behavioral, perceptual, and learning.
A related question designers face is whether to guide or restrict users. Studies such as Anthony
et al.'s (2004) that compare suggestive and restrictive approaches to directing users may be espe-
cially valuable.
Invocation Style
This dimension represents a basic design choice and should be relatively straightforward to study.
The main question is how the various effects depend on whether guidance is given automatically
or invoked by users on demand. Whether we should expect the findings to be unique to specific
guidance mechanisms, to vary by cell in the typology, or to generalize across various types of
guidance is unclear a priori. Two related issues of interest are (1) the effects of allowing people to
select the invocation style for themselves and (2) whether people are more likely to follow quasi-
suggestive and suggestive guidance when they request it.
Timing
Of the three options for timing guidance, the most typical case is guidance addressing a discre-
tionary opportunity the user currently confronts. Most studies of guidance will likely focus on this
case. But a sizable portion of the empirical studies did provide retrospective guidance, making
one wonder why a system would correct users after the fact rather than making suggestions when
the issue is initially on the table. Among the possible explanations are (1) that users may learn
more by first grappling with the issue themselves or (2) that the range of acceptable choices may
sometimes be so great that the system cannot advise in advance but can detect errors after the fact.
Retrospective guidance may therefore be an interesting topic to study.
Modes
This is the most engineering-oriented of the dimensions, addressing how the guidance is gener-
ated. Technologically, providing predefined guidance is relatively easy (although the provision of
guidance does not guarantee its success). Creating effective guidance mechanisms that dynami-
cally generate customized guidance for the user or interact with the user to generate guidance is
technologically more challenging. Comparing the effects of the three modes seems less fruitful
than inventing new dynamic or participative guidance mechanisms and studying their effects.