POWER Subcommand (MANOVA: Multivariate command)

The following specifications are available for POWER in multivariate analysis. For applications of POWER in univariate analysis, see POWER Subcommand (MANOVA: Univariate command).

APPROXIMATE. Approximate power values. This is the default. Approximate power values for multivariate tests are derived from procedures presented by Muller and Peterson ¹. Approximate values are normally accurate to three decimal places and are much cheaper to compute than exact values.

EXACT. Exact power values. Exact power values for multivariate tests are computed from the noncentral F distribution. Exact multivariate power values will be displayed only if there is one hypothesis degree of freedom, where all the multivariate criteria have identical power.

For information on the multivariate generalizations of power and effect size, see ², ³, and ⁴.

¹ Muller, K. E., and B. L. Peterson. 1984. Practical methods for computing power in testing the multivariate general linear hypothesis. Computational Statistics and Data Analysis, 2, 143-158.

² Muller, K. E., and B. L. Peterson. 1984. Practical methods for computing power in testing the multivariate general linear hypothesis. Computational Statistics and Data Analysis, 2, 143-158.

³ Green, P. E. 1978. Analyzing multivariate data. Hinsdale, Ill.: The Dryden Press.

⁴ Huberty, C. J. 1972. Multivariate indices of strength of association. Multivariate Behavioral Research, 7, 516-523.