Constrained Nonmetric Multidimensional Scaling.
MICHIGAN UNIV ANN ARBOR HUMAN PERFORMANCE CENTER
Pagination or Media Count:
One of the most crucial aspects of most multidimensional scaling procedures is that the lowest-stress configuration is the output produced. Unfortunately, little is known about the uniqueness of a configuration generated from fallible data, yet this affects the interpretation of the spatial output. It is possible, however, to examine the uniqueness of a scaled solution by constraining the configuration to conform to a particular psychological model. A multidimensional scaling program, CONSCAL, which will allow the imposition of such constraints, is proposed. The implications of this approach for interpreting scaling outputs and for model testing in general are discussed. Author
- Theoretical Mathematics