Database Reference
In-Depth Information
each other with standard statistical techniques. This approach has defi nitely improved our
understanding of the factors that affect subjects' ability to represent a case situation with
graphical tools, and a controlled experiment is a perfectly valid methodology for investigat-
ing specifi c aspects of a cognitively complex task such as conceptual data modeling. Future
research should, however, utilize the opportunities created by a richer set of background
theories and research methodologies.
We present two key ideas that can help with future research efforts:
a) First, we note that because almost all of the research to date has focused on the techni-
cal characteristics of the modeling formalisms, we know very little about the effects
of users' individual characteristics, task characteristics, or the interaction between
the modeling formalism, user, and task. Below, we discuss a new framework that we
hope will provide additional clarity to future research efforts.
b) Second, we observe that we do not have yet a good understanding of why certain
formalisms work well in some situations and not in others; the mechanisms mediat-
ing the relationships between the main research variables are not clear. We provide
several suggestions for research that can be used to strengthen our understanding in
this area.
An Expanded Framework for Human Factors Research in
Data Modeling
Our review of prior literature and additional conceptual analysis of this stream of
research leads us to believe that the traditional framework that has been used to guide hu-
man factors research on data modeling (see Figure 1) can be improved and clarifi ed. In
this section, we present and justify the suggested changes, which have been incorporated
into a new framework presented in Figure 2. We supplement the framework in Figure 2 by
using the fi ndings from the studies reviewed earlier as well as our theoretical understand-
ing of the domain. It is, however, worth noting that the theoretical basis for this expanded
framework as well as the Batra et al. (1990) framework lies in the classical general MIS
task-technology-human research framework, which, in turn, is a derivation of Leavitt's
(1965) organizational system model.
This framework was developed independently from Wand and Weber's general
framework for research on conceptual modeling (Wand & Weber, 2002), and its context and
intended uses are not as broad as those of Wand and Weber's framework. Our framework is
specifi cally intended for guiding human factors/usability research on data modeling, whereas
the Wand and Weber framework provides a comprehensive overall model for conceptual
modeling research. For example, Wand and Weber include Social Agenda Factors as one
of the critical conceptual factors; we have not included it in our framework because of our
more narrow focus on usability. However, we fully acknowledge the potential importance
of Social Agenda Factors as a broader contextual factor. Our Human category corresponds
directly to Wand and Weber's Individual Difference Factors, and our Task category is similar
to Wand and Weber's Task Factors. Most notably, Wand and Weber elegantly separate re-
search issues related to Conceptual-Modeling Grammar and Conceptual-Modeling Method;
in our framework, these are included in the Data Modeling Formalism category. We strongly
encourage readers to consult Wand and Weber (2002) for a more elaborate categorization
