Adam Wyse

Abstract

Demographic information collected on surveys is often of certain interest to those who use these surveys to measure customer satisfaction or do market research. This thesis presents a method for imposing a dichotomous incidence variable, possibly a demographic variable, on the weighting of items in a set of successive-categories (rating) data using dual scaling. The idea is to augment the matrix of rating data with the "criterion variable" containing the dichotomous information so that this item determines one of the initial solutions of the dual scaling analysis. In conjunction with the augmentation of the criterion item, the original data are "centered" between two numbers that represent the two criterion groups. The resulting modified data matrix is then subjected to a dual scaling analysis. The procedure is discussed with practical guidelines for its use and interpretation of results. Examples of application involve both fabricated and actual data.

Library of Congress Subject Headings

Discriminant analysis; Multivariate analysis; Incidence functions; Correspondence analysis (Statistics); Demographic surveys--Statistics

Publication Date

8-1-2005

Document Type

Thesis

Department, Program, or Center

Center for Materials Science and Engineering

Advisor

Lawrence, Daniel

Advisor/Committee Member

Voelkel, Joseph

Advisor/Committee Member

LaLonde, Steven

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: QA278.65 .W97 2005

Campus

RIT – Main Campus

Download

Share

COinS