A variety of problems associated with the interpretation of traditional canonical correlation are discussed. A response surface approach is developed which allows for investigation of changes in the coefficients while maintaining an optimum canonical correlation value. Also, a discrete or constrained canonical correlation method is presented. (JKS)

This paper explains how a major statistical package (BMDP) can be used to produce partial, semipartial, or bipartial set correlation in terms of a procedure outlined by Karpman (1980). (BW)

The canonical redundancy statistic, an estimate of the amount of shared variance between two sets of variables, exhibits an amount of bias similar to that of the first squared canonical correlation coefficient. Two formulae, Wherry and Olkin-Pratt, adequately correct the bias of the redundancy statistic. (Author/BW)

Problems of sampling error and accumulated rounding error in canonical variate analysis are discussed. A new technique is presented which appears to be superior to canonical variate analysis when the ratio of variables to sampling units is greater than one to ten. Examples are presented. (Author/JKS)

Conducted simulations to explore methods for comparing bivariate distributions corresponding to two independent groups, all of which are based on Tukey's "depth," a generalization of the notion of ranks to multivariate data. Discusses steps needed to control Type I error. (SLD)

A revision of the initialization strategy is suggested for OSMOD, a method of nonlinear principal component analysis for multivariate nominal and/or ordinal data. The revision is effective in diminishing the possibility of obtaining locally optimal solutions with categorical data provided by the original version of OSMOD. (SLD)

Measures of multivariate association, based on Wilks'"lambda" or the Bartlett-Nanda-Pillai trace criterion "V", are compared in terms of properties of univariate R-squared, which they generalize. A unified set of derivations of properties is provided, which is self-contained and not restricted to decompositions in canonical…

A computer-assisted, K-fold cross-validation technique is discussed in the framework of canonical correlation analysis of randomly generated data sets. Analysis results suggest that this technique can effectively reduce the contamination of canonical variates and canonical correlations by sample-specific variance components. (Author/SLD)

Reviews methods for testing whether two or more correlation matrices differ from each other. Illustrates methods for evaluating the similarities of two independent correlation matrices, such as those obtained from boys and girls, and two dependent correlation matrices, such as those obtained in longitudinal research that uses multiple measures.…

Although stated objectives of multivariate multiple regression and canonical correlation seem different, aspects of the analyses themselves are mathematically equivalent. A numerical example of the similarities is given, with reference to omnibus significance testing, variable weighting schemes, and dimension reduction analysis. (SLD)

Application of equality constraints on the categories of a variable is a simple and useful extension of multiple correspondence analysis. Equality is an easy way to incorporate prior knowledge. A procedure to deal with unequal category numbers and with subsets of variables is outlined and illustrated. (SLD)

The key ideas of multiple imputation for multivariate missing data problems are reviewed. Software programs available for this analysis are described, and their use is illustrated with data from the Adolescent Alcohol Prevention Trial (W. Hansen and J. Graham, 1991). (SLD)

Study identifies most representative ability and achievement for the Linking sample (N=824) of the Wechsler Intelligence Scale for Children-Third Edition (WISC-III) and the Wechsler Individual Achievement Test (WIAT). Primary benefit of core-profile typology is that it provides contrasts for testing multivariate IQ-achievement discrepancies. Two…

Computer simulations used to compare the Mahalanobis distance procedure with the Andrews-Pregibon statistic for identifying multivariate outliers result in the conclusion that the choice of procedure is not critical, since both identified valid data points as outliers. Results with false outliers as the dependent variable suggest the importance of…

Discusses the propriety and practical advantages of specifying multivariate time series models in the context of structural equation modeling for time series and longitudinal panel data. For time series data, the multiple indicator model specification improves on classical time series analysis. For panel data, the multiple indicator model…