DECISIONAL GUIDANCE - Human-Computer Interaction and Management Information Systems: Foundations - page 86

matching the task outperformed those given only mismatches. Both studies referred to giving
users only displays that fit the task as suggestive guidance, and both found merit in such guidance.
These two studies raise questions about how suggestive guidance should be operationalized.
Providing only displays that fit the task seems more restrictive than guiding, since users do not
have a choice. Including a treatment that recommended the matching display but still allowed
users to choose for themselves would be interesting and more closely match the definition of sug-
gestive guidance.
Wilson and Zigurs (1999) gained insight into the restrictiveness-guidance design issue by ask-
ing their subjects about their preferences. Most subjects indicated they would prefer making their
own choices over being restricted, but most also would prefer suggestive guidance over none.
These preferences are intuitively appealing, yet intriguing, given that Antony et al.'s (2005) sub-
jects perceived the restrictive system as the more usable.
These studies highlight a pair of issues: First, restrictiveness must be sharply distinguished
from guidance, both definitionally and operationally. Second, the design trade-off between the
two demands further empirical study. Studying system restrictiveness and suggestive guidance in
a single study seems highly beneficial.
Invocation and Timing
Seven of the nine studies provided guidance automatically, while two (Parikh et al., 2001; Wang
and Benbasat, 2004) provided it on demand, requiring users to request guidance. How guidance
is invoked has been discussed in the literature (Silver, 1991b), but was not included in the typol-
ogy. When studying the effects of decisional guidance, distinguishing those systems where guid-
ance is always delivered to the user from those where the user may request it seems important.
Moreover, for those systems providing guidance on demand, one would like to distinguish those
users who requested guidance from those who did not. Just as Wilson and Zigurs (1999) asked
users about their preferences for being restricted and being advised, researchers might want to
solicit users' preferences for automatic versus on-demand guidance. Even better than asking users
would be to observe their behavior and its consequences under each condition.
Several researchers have adopted decisional guidance as part of their efforts to classify CASE
tools. Vessey et al. (1992) were the first to do so, distinguishing three approaches to supporting a
given design methodology: a restrictive philosophy that forces users to follow the methodology's
rules, a guiding philosophy that is suggestive, and a flexible approach that does not try to influence
users. Jankowski (1997) further divided the guided approach into four cases based on two dimen-
sions. His first dimension addresses the invocation issue, capturing the difference between unso-
licited (automatic) and solicited (on-demand) guidance. He refers to these as active and passive
feedback guidance, respectively. Scott et al. (2000) make a similar distinction. Jankowski's (1997)
second dimension distinguishes guidance provided while the user is performing a task from guid-
ance provided after the task is complete. This second distinction, which addresses the timing of
guidance, also seems useful outside the realm of CASE tools.
Timing is implicitly or explicitly an issue in four of the nine empirical studies of decisional
guidance. Several of the studies provided users with corrective feedback after they had already
exercised their discretionary power. For instance, much of the suggestive guidance provided by
Montazemi et al. (1996) was in the form of corrections to conclusions the user had drawn.
Similarly, much of the guidance offered by Antony et al. (2005) reflected corrections to rule vio-
lations made by the user when designing the database. The guidance provided by Limayem and
DeSanctis (2000) included several types of cognitive feedback given when various problems were