Database Reference
In-Depth Information
etc.). This category of variables has been utilized widely in earlier research, as discussed in
the review above (Agarwal et al., 1996; Batra & Srinivasan, 1992; Brosey & Shneiderman,
1978; Burton-Jones & Weber, 1999; Hoffer, 1982; Lee & Choi, 1998; Ramesh & Browne,
1999; Weber, 1996). Finally, an individual's
technical skills
in the use of a specifi c data
modeling formalism should be conceptually separated as a factor affecting the user's per-
formance. One of the reasons why it is essential to differentiate technical skills from other
aspects is that this is the only subcategory of individual differences in this framework that
can be affected by
training
(other factors that could be infl uenced by training include confi -
training
dence, self-effi cacy, task motivation, etc). Technical skills have been used as an independent
variable in several studies (Batra & Antony, 2001; Weber, 1996). In general, the division
of the framework elements into components forces us to specify the nature of the relation-
ships of interest at a signifi cantly more detailed level. This, in turn, will lead us closer to
true theoretical models at least in part based on applicable theories from relevant reference
disciplines, such as Anderson's ACT theory with its variants (Anderson, 1993), which was
suggested as an important theoretical basis for research on information modeling (including
conceptual data modeling) by Siau (1999).
Second, the framework should incorporate two different types of dependent variables
to acknowledge the fact that we are not only interested in objective performance but also
users' attitudes towards the tools, the tasks, and their own performance. The most often
used non-performance dependent variables are ease-of-use perceptions (Batra et al., 1990;
Hardgrave & Dalal, 1995; Kim & March, 1995; Lee & Choi, 1998) and modeling formal-
ism preference (Batra & Sein, 1994; Kim & March, 1995; Shoval & Even-Chaime, 1987;
Shoval & Shiran, 1997).
Third, the framework should acknowledge and explicitly incorporate the potentially
complex moderating effects of other variables on the relationship between the data modeling
formalism and user performance and attitudes. The direct effect of task complexity on the
dependent variables, particularly performance, is seldom the main point of interest; in most
cases, we are interested in the way different formalisms support users at various task com-
plexity levels. The same is true with task type: a relevant research question is the suitability
of various modeling formalisms for specifi c task types and thus, we should explicitly express
in our research model that task type moderates the relationship between the data modeling
formalism and the dependent variables. The best examples of this are the experiments by
Kim and March (1995), who studied the use of two formalisms for user (validation) and
analyst (modeling) tasks, and Lee and Choi (1998), who compared four different formalisms
in two task types. The commonly used analysis of performance by facets (Batra et al., 1990;
Bock & Ryan, 1993; Lee & Choi, 1998; Liao & Palvia, 2000; Shoval & Shiran, 1997) is, in
fact, a form of analysis of the moderating effects of task type, because modeling a specifi c
facet can be seen as a subtask. As discussed above in the summary of results, the facet being
modeled often moderates the impact of a specifi c modeling formalism on performance.
Finally, the research framework should explicitly acknowledge that various individual
characteristics have differential effects on user performance and attitudes and that many of
the effects of individual differences moderate the relationship between the data modeling
formalism and the dependent variables. In addition, some of the relationships between the
categories of individual characteristics affect each other in a signifi cant way: Task-related
experience affects an individual's technical data modeling skills (in addition to training), and
the general individual differences (such as intelligence) moderate the relationship between
the training an individual receives and the individual's skills.
training
(other factors that could be infl uenced by training include confi -
