Application of Correspondence Analysis to classical statiscal procedures /
| Main Author: | |
|---|---|
| Other Authors: | , , |
| Format: | Thesis Book |
| Language: | English |
| Published: |
1989.
|
| Subjects: | |
| Online Access: | Link to OAKTrust copy |
| Abstract: | A probability-based decision making methodology for Correspondence Analysis (CA) is introduced. Exact distributional p-values and empirically based p-values are given for various distance measures. CA based procedures for outlier detection (in goodness of fit problems), test of symmetry, and deviations from the underlying assumptions in ANOVA (factorial experiments) are presented. These three procedures are illustrated and compared to the results obtained for the classical procedures. It is demonstrated how, even under the assumptions of the classical techniques, CA not only obtains the same results as the classical techniques, but also provides a diagnostic tool that identifies reasons for the rejection of a null hypothesis. Furthermore, CA continues to provide the diagnostic tools when the underlying assumptions required for the classical procedures are violated. |
|---|---|
| Item Description: | Typescript (photocopy). Vita. "Major subject: Statistics." |
| Physical Description: | x, 102 leaves : illustrations ; 29 cm |
| Bibliography: | Includes bibliographical references. |