AESTHETICS IN INFORMATION TECHNOLOGY - Human-Computer Interaction and Management Information Systems: Foundations - page 321

of low-level mechanisms (i.e., biological, physiological) of those processes, and more in abstract,
higher-level aesthetic evaluations. Such processes are more accessible to behavioral research
methods and can convey a decent degree of information about the IT antecedents of aesthetic per-
ceptions and evaluations and about the consequences of such perceptions. To date, most studies
of aesthetics in IT have only measured a general aesthetic evaluation of the IT artifact. Such
a measure can be a one- or a multiple-item scale assessing the system's aesthetics. For example,
Kurosu and Kashimura (1995), Tractinsky (1997), Schenkman and Jonsson (2000), and Hassenzahl
(2004a) have all used a single item asking about the beauty of the IT artifact, while Van der
Heijden (2003) employed a three-item scale measuring perceived attractiveness. These broad
measures were sufficient to demonstrate the relations between overall aesthetic perceptions and
their IT antecedents and consequences. For example, Kurosu and Kashimura (1995) and Tractinsky
(1997) found that manipulating the layout of objects on an ATM machine affects evaluations of
the machine's beauty. In turn, the aesthetic perceptions influenced the evaluation of other system
attributes, such as its ease of use (Tractinsky et al., 2000). Hassenzahl (2004a) found that per-
ceived beauty was related to the perceived goodness of a system. Van der Heijden's (2003) com-
posite measure of visual attractiveness was associated with perceived ease of use, usefulness, and
enjoyment of a Web site, and in Schenkman and Jonsson (2000) perceived beauty was related to
overall preferences of Web sites.
Perhaps of even greater interest is the ability to identify various sub-dimensions or nuances of
aesthetic evaluations that are relevant to the IT context. This can help in finer-grain analyses of the
associations between certain design characteristics of interactive systems and their behavioral
consequences, and in providing richer accounts of HCI processes. There are already a few examples
of research involving such higher-resolution measures. Kim et al. (2003) have identified specific ele-
ments of aesthetic design and were able to link them to various emotional dimensions experienced
while browsing the Web. Similarly, specific (as opposed to general) aesthetic evaluations can
improve our understanding of how design influences various outcome variables (e.g., Lavie and
Tractinsky, 2004; Hassenzahl, 2004). I discuss this issue in more detail in the next subsection.
Outcome Variables
In the proposed framework, the range of potential outcomes can span virtually the whole gamut
of outcome variables employed by behavioral IS researchers. For example, all of the categories of
dependent variables identified by DeLone and McLean (1992) can serve as outcome variables in
research on aesthetics: system quality, information quality, information use, user satisfaction, and
individual and organizational impacts. In addition, aesthetics may influence outcomes that are
more affective in nature and that were not considered at the time as highly pertinent to IT research,
such as online consumer behavior (e.g., Tractinsky and Rao, 2001) and related constructs such as
trust (Karvonen, 2000), pleasure (Jordan, 1998), or flow (Csikszentmihalyi, 1990; Csikszentmihalyi
and Robinson, 1990). All of these and similar variables may mediate the effects of IT aesthetics
on the type of variables proposed by DeLone and McLean.
Moderating Variables
Obviously, the effects of aesthetics are moderated by various factors. A partial list of potential
moderators include the type of system used (e.g., a handheld entertainment system vs. a Web-based
banking system); the task(s) to be performed with the system; the social context in which the sys-
tem is used; cultural (organizational, societal, national) differences; motivational factors; and the