Canonical Correlation Analysis: An Explanation with Comments on Correct Practice.

ABSTRACT This paper briefly explains the logic underlying the basic calculations employed in canonical correlation analy3is. A small hypothetical data set is employed to illustrate that canonical correlation analysis subsumes both univariate and multivariate parametric methods. Several real data sets are employed to illustrate other themes. Three common fallacious interpretation practices that may lead to incorrect conclusions are discussed; these falla6es affect the interpretation of function coefficients, interpretation of redundancy coefficients, and failure to partition using canonical commonality analysis. The use of the factor analytic method of rotation to simplify results is also discussed. It is suggested that canonical correlation analysis is a powerful analytic method that frequently best honors the complex nature of the reality about which the researcher wishes to generalize. Thirty-two tables and two graphs are presented. (Author/TJH)

[1]  A. R. Baggaley.  Multivariate Analysis , 1981 .

[2]  M. Levine.  Canonical Analysis and Factor Comparison , 1977 .

[3]  T. R. Knapp.  Canonical correlation analysis: A general parametric significance-testing system. , 1978 .

[4]  J. Elashoff,et al.  Multiple Regression in Behavioral Research. , 1975 .

[5]  J. Johansson,et al.  An extension of Wollenberg's redundancy analysis , 1981 .

[6]  Stephen A. Zinkgraf,et al.  Performing Factorial Multivariate Analysis of Variance Using Canonical Correlation Analysis , 1983 .

[7]  William L. Hays.  Statistics (3rd ed.). , 1990 .

[8]  Robert M. Thorndike.  Strategies for Rotating Canonical Components. , 1976 .

[9]  W. Alan Nicewander,et al.  Some symmetric, invariant measures of multivariate association , 1979 .

[10]  William Meredith,et al.  Canonical correlations with fallible data , 1964 .

[11]  Maurice M. Tatsuoka,et al.  9: Multivariate Analysis in Educational Research , 1973 .

[12]  ORSIM2: A FORTRAN Program for Symmetric and Asymmetric Orthogonal Rotation of Canonical Variates and Interbattery Factors , 1982 .

[13]  W. DeSarbo.  Canonical/redundancy factoring analysis , 1981 .

[14]  R. Kirk.  Experimental Design: Procedures for the Behavioral Sciences , 1970 .

[15]  Bruce Thompson,et al.  Canonical Correlation: Recent Extensions for Modelling Educational Processes. , 1980 .

[16]  G. Loftus.  Essence Of Statistics , 1982 .

[17]  Fn Kerlinger,et al.  Foundations of behavioral research (2nd ed. , 1973 .

[18]  B. Thompson.  Canonical Correlation Analysis: Uses and Interpretation , 1984 .

[19]  A. L. V. D. Wollenberg.  Redundancy analysis an alternative for canonical correlation analysis , 1977 .

[20]  John K. Miller.  The Sampling Distribution and a Test for the Significance of the Bimultivariate Redundancy Statistic: A Monte Carlo Study. , 1975 .

[21]  L. Goodwin,et al.  Statistical Techniques in AERJ Articles, 1979–1983: The Preparation of Graduate Students to Read the Educational Research Literature , 1985 .

[22]  G. Keppel,et al.  Design and Analysis: A Researcher's Handbook , 1976 .

[23]  D. J. Spurrell,et al.  A Development of Multiple Regression for the Analysis of Routine Data , 1967 .

[24]  T. Gerig.  Multivariate Analysis: Techniques for Educational and Psychological Research , 1975 .

[25]  G. C. Homans,et al.  The nature of social science , 1967 .

[26]  E. Pedhazur.  Multiple Regression in Behavioral Research: Explanation and Prediction , 1982 .

[27]  V. Willson.  Misuses of Regression Approaches to ANOVA and ANCOVA. , 1982 .

[28]  Jacob Cohen.  Multiple regression as a general data-analytic system. , 1968 .

[29]  Howard B. Lee,et al.  Foundations of Behavioral Research , 1973 .

[30]  Bruce Thompson,et al.  The Importance of Structure Coefficients in Regression Research , 1985 .

[31]  P. Horst.  Generalized canonical correlations and their applications to experimental data. , 1961, Journal of clinical psychology.