COMPUTER SELF-EFFICACY - Human-Computer Interaction and Management Information Systems: Foundations - page 221

The following sections review both our own research and relevant other research in each of
these domains.
Formal Training
Computer self-efficacy is viewed as an important outcome of training. Given its influence on train-
ing performance, discussed above, and its influence on choice behaviors outside of training, recog-
nizing formal training as an opportunity to improve users' judgments of their capability is critical.
In fact, Marcolin et al. (2000b), following Kraiger et al. (1993), define CSE as an affective measure
of a user's competence. This positions CSE as an important learning outcome in its own right.
Each of the four sources of efficacy information plays a role in formal training. Guided mas-
tery is evident in the self-paced, customized experiences of computer-aided learning and in the
practice sessions that are typically included in a training course. Verbal persuasion occurs through
the feedback of the trainer, who advises the trainee about his or her level of capability and helps
to enhance self-efficacy judgments. By reminding trainees of their level of progress and their capa-
bility relative to other novice learners, trainees can be encouraged to reflect more positively on
their experiences. Thus, they develop confidence that, despite their current low performance, they
will ultimately be able to learn the required behaviors. We have already discussed the relationship
between CSE and physiological states. Behavior modeling remains the most investigated aspect
of formal training from a self-efficacy perspective.
Behavior modeling can be part of the formal design of the training program (Compeau and
Higgins, 1995a; Gist et al., 1989; Johnson and Marakas, 2000; Simon and Werner, 1996; Simon et al.,
1996). It also occurs through every action of the trainer and through the behaviors of other training
participants. Compeau and Higgins (1995a) acknowledged that the trainer in their study acted as a
behavior model for all trainees, with a videotaped model as an additional modeling stimulus for their
experimental groups. By demonstrating how to use the software, showing the resolution of her own
mistakes if she made them, and explaining her approach to using the software, the trainer acted as a
behavior model. One key difference between the trainer-as-model and the videotaped model was in
the social comparison. Bandura (1986) and others argue that the most effective behavior models are
those who are viewed as similar by the trainees, since modeling represents a process of social com-
parison. Thus, compared to a novice demonstrating his or her own use of the software (which is how
Compeau and Higgins constructed their modeling intervention), the trainer, a software professional,
was quite different from the novice trainees. Nonetheless, in their exposure to the trainer over the
course of the training session, the trainees observed and modeled her behavior, with the result that
their confidence increased, regardless of the training group. Consistent with social cognitive theory,
however, SCSE increased more for those subjects who were exposed to the novice model.
Modeling is not limited to that provided by the trainer. Lim et al. (1997) found that subjects
engaged in co-discovery learning learned more than those engaged in a learning alone protocol.
While the authors do not assess computer self-efficacy, their results may also reflect the modeling
that takes place between the learners in a co-discovery setting where each individual acts as a
model for the other.
Models differ in the means through which they present information. Some models demons-
trate flawless performance, executed easily. These masterful models demonstrate how it is done
right, by those who are capable of doing it well. This is typical of many video-based training
materials—a carefully rehearsed script is used to provide a demonstration, which is then edited to
ensure that any errors are removed. The problem with these models is that they demonstrate a level
of understanding so far beyond that of the learner that they can actually have a demoralizing effect.