A Program for Computing a-Priori and Post-Hoc Multivariate Contrasts

Abstract

A computer program is described which computes any of three sets of multivariate contrasts following a significant outcome in an overall multivariate analysis of variance (MANOVA). Input to the program comprises the required matrices, vectors, and scalars from the MANOVA output of major well known published programs. The program analyzes multivariate contrasts in (a) the a-priori, orthogonal case; (b) the a-priori, nonorthogonal case; and (c) the strictly post-hoc case. Roy-Bose multivariate simultaneous-contrast intervals are employed in the post-hoc case, whereas Bonferonni t's are used in the a-priori cases. The program, which is in FORTRAN IV, requires 7K of memory and uses less than one second of central processing time.

Get full access to this article

View all access options for this article.

References

Harris, R. J. A primer of multivariate statistics . New York: Academic Press , 1975.

Kirk, R. E. Experimental design: Procedures for the behavioral sciences. Monterey, California: Brooks/Cole, 1968.

Morrison, D. F. Multivariate statistical methods . New York: McGraw-Hill , 1967.

Roy, S. N. Some aspects of multivariate analysis. New York: Wiley, 1957.

Roy, S. N. and Bose, R. C. Simultaneous confidence interval estimation. Annals of Mathematical Statistics, 1953, 24, 513-536 .

SAS Institute. Statistical analysis system user's guide. Raleigh, N. C.: SAS Institute, 1979.

Timm, N. H. Multivariate analysis with applications in education and psychology. Monterey, California: Brooks/Cole, 1975.