Database Reference
In-Depth Information
and Liao and Palvia (2000). Jarvenpaa and Machesky (1989) investigated the effects of
learning by using a within-subjects design and administering four data modeling tasks to
each subject.
Dependent variables.
The dependent variables can be divided into two broad catego-
ries: user performance and user attitudes. As seen earlier, the two main research questions
of this area are related to modeling performance and user satisfaction, and, therefore, the
widespread use of these dependent variables is understandable.
Performance
has been divided into three subcategories:
model correctness
(also referred
to as
procedural or skill knowledge
of the user by Jarvenpaa and Machesky (1989), measured
by the characteristics of the end result of the modeling process),
time
used to create the
solution, and
declarative knowledge
(understanding of the notation (Jarvenpaa & Machesky,
1989)). In most cases, the correctness of the model has been measured with the degree to
which it corresponds to a predefi ned “correct” solution. Batra et al. (1990) were the fi rst to
refi ne the concept of correctness by measuring the correctness of various
facets
or structural
elements of the model (entities, identifi ers, descriptors, categories, and fi ve different types
of relationships: unary, binary one-to-many (1:M), binary many-to-many (M:N), ternary
one-to-many-to-many (1:M:N), and ternary many-to-many-to-many (M:N:O)). The same
facet structure was used later by Bock and Ryan (1993), Shoval and Shiran (1997), Lee
and Choi (1998), and Liao and Palvia (2000). Kim and March (1995) divided the analysis
of model correctness into
syntactic and semantic
categories: Syntactic correctness refers
to users' ability to understand and use the constructs of the modeling formalism, whereas
semantic correctness is the extent to which the data model corresponds to the underlying
semantics of the problem domain. Another widely used measure of performance has been
the time it takes to fi nish a modeling or model comprehension task (Hardgrave & Dalal,
1995; Jarvenpaa & Machesky, 1989; Lee & Choi, 1998; Liao & Palvia, 2000; Palvia et al.,
1992; Shoval & Even-Chaime, 1987; Shoval & Shiran, 1997).
The
user attitudes
measured within this area of research are confi dence (Hoffer, 1982),
preference to use a certain model (Shoval & Even-Chaime, 1987; Shoval & Shiran, 1997),
perceived value of the modeling formalism (Kim & March, 1995), and perceived ease-of-
use (Batra et al., 1990; Hardgrave & Dalal, 1995; Kim & March, 1995).
In a study in which the dependent variable does not belong to either one of the main
categories,
data model characteristics
were the main point of interest for Hoffer (1982). His
study focused on the nature of the data models which the subjects created when they were
able to freely choose the way to describe a structure of a database. The two characteristics
of the model in his study were “image architecture” and “image size”, that is, the modeling
approach chosen and the number of entities.
Identifi ed control variables.
By investigating the nature of the explicitly identifi ed
control variables in previous research, it is possible to fi nd potential independent vari-
ables of interest for future research, as well as summarize the variables that have to be
controlled in future studies.
User characteristics
(
Human
( in the framework in Figure 1)
is the fi rst category of specifi c control variables in the earlier studies. The most common
individual variable in the user characteristics category is experience
.
The most common
types of experience discussed in prior research are general work experience (Batra et al.,
1990; Batra & Kirs, 1993; Jarvenpaa & Machesky, 1989; Juhn & Naumann, 1985; Liao
& Palvia, 2000), general computer/IS experience (Batra et al., 1990; Batra & Kirs, 1993;
Jarvenpaa & Machesky, 1989; Juhn & Naumann, 1985; Liao & Palvia, 2000), and database
experience (Batra et al., 1990; Batra & Kirs, 1993; Jarvenpaa & Machesky, 1989; Juhn &
(
Human
