Approximate Power for Repeated-Measures ANOVA Lacking Sphericity

Abstract Violation of sphericity of covariance across repeated measures inflates Type I error rates in univariate repeated-measures analysis of variance. Hence the use of Geisser—Greenhouse or Huynh—Feldt is now common (to provide improved Type I error rate). With nonsphericity, no method has been available to compute power. A convenient method is suggested for approximating power and test size under nonsphericity. New approximations are suggested for (a) a weighted sum of independent noncentral χ2,s, (b) the trace of a noncentral Wishart (or pseudo-Wishart) matrix, (c) the expected values of and , and (d) the noncentral test statistic, whether corrected or uncorrected. The new approximations are extensions of the work of Box (1954a,b) and Satterthwaite (1941). The method performed well when evaluated against published and new simulations.

[1]  S. Geisser,et al.  On methods in the analysis of profile data , 1959 .

[2]  Jorge L. Mendoza,et al.  Testing the validity conditions of repeated measures F tests. , 1980 .

[3]  G. Box.  Some Theorems on Quadratic Forms Applied in the Study of Analysis of Variance Problems, I. Effect of Inequality of Variance in the One-Way Classification , 1954 .

[4]  R. Davies.  The distribution of a linear combination of 2 random variables , 1980 .

[5]  H. Huynh,et al.  Conditions under Which Mean Square Ratios in Repeated Measurements Designs Have Exact F-Distributions , 1970 .

[6]  Arak M. Mathai,et al.  Moments of the trace of a noncentral wishart matrix , 1980 .

[7]  Michael L. Davidson.  Univariate versus multivariate tests in repeated-measures experiments. , 1972 .

[8]  R. Arvey,et al.  Small sample profile analysis with many variables. , 1982 .

[9]  S. R. Searle.  Linear Models , 1971 .

[10]  G. Box.  Some Theorems on Quadratic Forms Applied in the Study of Analysis of Variance Problems, II. Effects of Inequality of Variance and of Correlation Between Errors in the Two-Way Classification , 1954 .

[11]  D. F. Morrison,et al.  Multivariate Statistical Methods , 1968 .

[12]  W. A. Nicewander,et al.  A MONTE CARLO COMPARISON OF THE UNIVARIATE AND MULTIVARIATE METHODS FOR THE GROUPS BY TRIALS REPEATED MEASURES DESIGN. , 1974, Multivariate behavioral research.

[13]  S. Geisser,et al.  An Extension of Box's Results on the Use of the $F$ Distribution in Multivariate Analysis , 1958 .

[14]  Keith E. Muller,et al.  Practical methods for computing power in testing the multivariate general linear hypothesis , 1984 .

[15]  T. W. Anderson.  An Introduction to Multivariate Statistical Analysis , 1959 .

[16]  F B Baker,et al.  Estimates of test size for several test procedures based on conventional variance ratios in the repeated measures design , 1967, Psychometrika.

[17]  C. M. Wang.  On the analysis of multivariate repeated measures designs , 1983 .

[18]  Necessary and Sufficient Conditions for F Ratios in the L × J × K Factorial Design with Two Repeated Factors , 1976 .

[19]  Y. Fujikoshi.  Asymptotic expansions for the distributions of some functions of the latent roots of matrices in three situations , 1978 .

[20]  F. E. Satterthwaite.  Synthesis of variance , 1941 .