Personal Attribute Judgments - Modeling Users' Experiences with Interactive Systems - page 26

Scaling (MDS). These techniques aim at modeling relations between stimuli (e.g.
interactive products), attributes (e.g. “professional - unprofessional”) and overall
judgments (e.g. preference). More specifically, MDS looks for a K-dimensional con-
figuration for the stimuli such that the coordinates of the stimuli in the configuration
space along different axes can be monotonically related to the observed attribute
ratings of the participants (Martens, 2003).
Figure 2.3 illustrates a two-dimensional MDS configuration with two stimuli and
two attributes. The relative positions of the stimuli on a given attribute axis reflect
participants' ratings for the stimuli on this attribute. For instance, website j can be
perceived as being both more legible and colorful than websites i and k .
An important motivation for MDS is the principle of homogeneity of perception
which states that attribute judgments from different participants are related and thus
can be represented in a common configuration space (Green et al., 1989; Martens,
2003). This view, although it often holds in perceptual judgments, has recently been
challenged in more cognitive judgments where the quality dimensions of interactive
products are assessed.
Fig. 2.3 A two-dimensional MDS configuration of three websites using ratings from two
attributes. Website j is perceived as more legible and colorful than websites i and k .
To our knowledge, all RGT approaches up to date have been employing an aver-
age two-dimensional configuration for the quantitative analysis of personal attribute
judgments from all participants. As it will become evident in Chapter 3, such aver-
aging analysis procedures in low dimensional spaces, which constitute the common
practice in the analysis of Repertory Grid data, fail in providing insight into the
majority of elicited attributes.
This chapter proposes an initial Multi-Dimensional Scaling procedure that aims
at inquiring into the diverse views that different individuals might have on a set of
products. The procedure consists of two steps: a) identifying homogeneous groups
of individuals, and b) eliciting MDS configurations for each homogeneous group to
understand how they perceive and evaluate the products being studied.