Database Reference
In-Depth Information
Variables of Interest in Empirical Studies
Research framework
. Figure 1 includes a schematic representation of the research
framework that has been used either explicitly (as by Batra et al., 1990) or implicitly in many
of the earlier studies.
Human
refers to the individual level factors related to the character-
istics of the individuals who perform the data modeling tasks,
Data Model
is used in this
context to describe the differences between the data modeling formalisms, and
Task
refers
Task
to the characteristics of the tasks of interest related to data models, such as model creation,
comprehension, or validation. The model indicates a reciprocal relationship between Hu-
man, Data Model, and Task, which all, in turn, have an impact on the quality of the result-
ing data model, that is, (human) Performance in the data modeling task. Variables in the
Human, Data Model, and Task categories have been used in earlier studies as independent
and control variables, as indicated in the discussion below, and Performance is a natural
dependent variable in the studies.
Independent variables.
The most frequently used independent variable in the earlier
studies has been the data modeling approach or
data model,
data model
,
data model
as it is called by, for example,
Batra and Davis (1992) and Navathe (1992) and in the research framework in Figure 1.
In early research, Brosey and Shneiderman (1978) compared hierarchical and relational
data models, whereas several later studies have compared different types of semantic and
relational data models (Amer, 1993; Batra & Antony, 1994; Batra et al., 1990; Jarvenpaa
& Machesky, 1989; Juhn & Naumann, 1985; Liao & Palvia, 2000; Sinha & Vessey, 1999)
and/or two different semantic data models (Kim & March, 1995; Lee & Choi, 1998; Liao
& Palvia, 2000; Nordbotten & Crosby, 1999; Palvia, Liao & To, 1992). Several of the most
recent studies have compared semantic data models to object-oriented data models (Bock
& Ryan, 1993; Hardgrave & Dalal, 1995; Lee & Choi, 1998; Liao & Palvia, 2000; Palvia et
al., 1992; Shoval & Frumermann, 1994; Shoval & Shiran, 1997; Sinha & Vessey, 1999).
The next category of independent variables consists of
user characteristics
(
Human
(
in the research framework in Figure 1). The most commonly used independent variable is
experience: The level of general MIS or programming experience was used as an independent
variable in studies by Brosey and Shneiderman (1978) and Hoffer (1982), whereas Batra
and Davis (1992), Weber (1996), and Lee and Choi (1998) analyzed the differences between
subjects with various levels of data modeling experience. Ramesh and Browne (1999) dif-
ferentiated between “database-knowledgeable” and “database novice” based on the subjects'
understanding of basic ER concepts. Agarwal, Sinha, and Tanniru (1996) investigated the
impact of the type of design experience on modelers' ability to use different formalisms for
different tasks. In addition to programming expertise, Hoffer (1982) studied the effects of
cognitive style, another category of individual differences. Finally, Siau, Wand, and Ben-
basat (1995) and Burton-Jones and Weber (1999) have explored the effects of the subjects'
familiarity with the problem domain or the problem domain expertise.
A set of
task characteristics
(
Task
in the research framework in Figure 1) has also been
used as an independent variable in the studies: Brosey and Shneiderman (1978) manipulated
the
task type
(comprehension, problem solving, memorization), as did Batra and Antony
(2001) (task's compatibility with a support tool). Hoffer (1982) varied the description of
the situation on which the data model was based so that the situation was either related to
a specifi c task or to the entire organization.
Task complexity
was used as an independent
variable in Shoval and Even-Chaime (1987), Hardgrave and Dalal (1995), Weber (1996),
(
Human
