Graphics Reference
In-Depth Information
Figure
.
.
Unidimensional scaling of World War II political leaders
INDSCAL
3.7
Suppose data consist of several sets of dissimilarities between objects, the same ob-
jects in each case. For example, several panellists assess the dissimilarities between
all pairs of a set of products. MDS could be applied to each panelist's data resulting
in many configurations. A better approach might be to combine the dissimilarities
in some manner.
CarrollandChang(
)proposedametricmodelcomprisingtwospaces:agroup
stimulus space and a subject's (or individual's) space, both chosen to be of the same
dimension. Points in the group stimulus space represent the objects or stimuli and
form an “underlying” configuration. he individuals are represented as points in the
subject's space. he coordinates of each individual are the weights required to give
theweightedEuclideandistances between thepointsinthestimulusspace,thevalues
