Tensor Analysis and Multidimensional Scaling: The Influence of Class Structure on Similarity Judgments.
Abstract
E ASSUMPTION IS VIOLATED IN CASES WHERE THE SAMPLE OF STIMULUS OBJECTS HAS AN UNDERLYING CLASS STRUCTURE. It shows that while the prediction of similarity judgments for pairs of stimuli belonging to the same class is consistent with the scaling model, the prediction of between class judgments does not fit the model. In particular, between class similarity depends on the separation between the two stimuli being compared and the centroids or prototypes of the classes as well as on the distance between the stimuli in the multidimensional space. That is, the perceived distance (dissimilarity) between two stimuli depends not only on coordinate differences but on the relative position of the pair in the geometric space. This result is inconsistent with the assumptions underlying the multidimensional scaling model. A pseudo proof is constructed to demonstrate this inconsistency. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1972
- Accession Number
- AD0753889
Entities
People
- Richard M. Fenker
Organizations
- Texas Christian University