ERIC Number: ED343913
Record Type: RIE
Publication Date: 1992-Jan
Reference Count: N/A
Applying Procrustean Rotation To Evaluate the Generalizability of Canonical Analysis Results.
Tucker, Mary L.; Campbell, Kathleen Taylor
Statistical invariance procedures provide a way of looking at the generalizability of research results from sample to sample when the research has not been validated by replication. This paper discusses the Procrustean Rotation invariance procedure following a canonical correlation analysis. The computer program RELATE is used to gauge the replicability and generalizability of the illustrative substantive research results. A Procrustean rotation forces orthogonal (uncorrelated) functions of factors to a best fit position after setting the factor vectors to unit length (1.0) in order to equalize the contribution of each factor vector to the determination of the amount of rotation necessary. This rotation technique can be used as a cross-validation procedure, splitting data from a single sample and comparing factor vectors for each half. The sample used to illustrate the procedure is from a study of leadership styles conducted by M. L. Tucker (1990) using data from 106 college faculty and administrators, with the university research sample split into two uneven data sets (n=48 and n=58). Two tables of illustrative data and an 11-item list of references are included. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers: Cross Validation; Invariance; Procrustes Rotation; RELATE Computer Program; Research Replication; Rotations (Factor Analysis)
Note: Paper presented at the Annual Meeting of the Southwest Educational Research Association (Houston, TX, January 31-February 2, 1992).