Intersession Variability and Stress: A Monte Carlo Study.
Abstract
Two Monte Carlo studies explore the relation of the tau measure of intersession response variability and the stress of the corresponding multidimensional scaling solution, thereby providing a statistical basis for evaluating the goodness-of-fit of a spatial configuration. In the first Monte Carlo study, the stress and tau of 10-, 16-, and 30-point configurations in 1, 2, 3, and 4 dimensions are shown to be linear functions of the internal error level. In the second study, these relations are shown to be relatively invariant with respect to the particular configurations. Three methods are proposed for establishing acceptable levels of stress for heuristic and for constrained multidimensional scaling. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1979
- Accession Number
- ADA071612
Entities
People
- Elliot Noma
Organizations
- University of Michigan