ERIC Number: ED348383
Record Type: RIE
Publication Date: 1992-Apr
Reference Count: N/A
Canonical Correlation Analysis as a General Analytical Model.
This paper focuses on three aspects related to the conceptualization and application of canonical correlation analysis as a dominant statistical model: (1) partial canonical correlation analysis and its application in statistical testing; (2) the relation between canonical correlation analysis and discriminant analysis; and (3) the relation between canonical correlation analysis and chi-square contingency table analysis. The paper shows that canonical correlation analysis can be conceptualized as the statistical model that brings together many other statistical techniques in a unified manner, and the power of this overarching model is significantly increased by applying the concept of partial correlation to the canonical case. Two data sets (one with two Y variables, three X variables, and two classification variables; and the other with two mixed variables with three levels for each) are used to illustrate the points covered. Computer program results are presented to augment the discussion. Appendix A presents the SAS program for some tabulated data. Six tables present analysis results, and there is a 29-item list of references. (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers: Contingency Tables
Note: Paper presented at the Annual Meeting of the American Educational Research Association (San Francisco, CA, April 20-24, 1992).