Computer Programs for Robust Analyses in Multifactor Analysis of Variance Designs

Abstract

Robust analyses of multifactor analysis of variance data may be secured even when assumptions are violated, by use of a set of five programs. ANOVR produces means, the summary table, group variances, the variance-covariance matrix on repeated measures, and tests most assumptions. FOLUP tests any set of pairwise contrasts of means, while COMCON tests any set of complex contrasts. Both latter programs permit solutions using heterogeneous group variances, or using a heterogeneous variance-covariance matrix from repeated measures, as well as conventional solutions using mean squares from the summary table. ANOVR produces the Box-Geisser-Greenhouse index when there is just one repeated measure factor, while PVCVRL will produce corresponding indices when there are two or more such factors. BARTKI produces output that generates the robust and flexible Bartlett and Kendall test for homogeneity of variances. Intelligent use of the set of programs frees the user from most assumptions of AOV.

Get full access to this article

View all access options for this article.

References

Box, G. P. E. Some theorems on quadratic forms applied in the study of analysis of variance problems: II. Effect of inequality of variance and correlation of errors in the two-way classification . Annals of Mathematical Statistics, 1954 , 25, 484-498.

Brown, M. B. and Forsythe, A. B. The small sample behavior of some statistics which test the equality of several means. Technometrics, 1974, 16, 129-132.

Collier, R. O. , Baker, F. , Mandeville, G. K. , and Hayes, T. Estimates of test size for several test procedures based on conventional variance ratios in the repeated measures design. Psychometrika, 1967, 32, 339-354.

Games, P. A. Multiple comparisons of means . American Educational Research Journal, 1971, 8, 531-565.

Games, P. A. , Winkler, H. B. , and Probert, D. A. Robust tests for homogeneity of variance. EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT, 1972, 34, 887-909.

Gartside, P. S. A study of methods for comparing several variances. Journal of the American Statistical Association , 1972, 67, 342-346.

Geisser, S. and Greenhouse, S. On methods in the analysis of profile data. Psychometrika, 1959, 24, 95-112.

Howell, J. F. and Games, P. A. The effects of variance heterogeneity on simultaneous multiple comparison procedures with equal sample size. British Journal of Mathematical and Statistical Psychology, 1974, 27, 72-81.

Huynh, H. and Feldt, L. S. Conditions under which mean square ratios in repeated measurements designs have exact F-distributions. Journal of the American Statistical Association, 1970, 65, 1582-1589.

10.

Kohr, R. L. and Games, P. A. Robustness of the analysis of variance, the Welch procedure, and a Box procedure to heterogeneous variances. Journal of Experimental Education, 1974, 43, 61-65.

11.

Welch, B. L. The generalization of students' problem when several different population variances are involved. Biometrika, 1947, 34, 28-35.

12.

Wiley, D. E. , Schmidt, W. H. , and Bramble, W. J. Studies of a class of covariance structure models. Journal of the American Statistical Association, 1973, 68, 317-323.

13.

Winer, B. J. Statistical principles in experimental design. (2nd ed.) New York: McGraw-Hill, 1971.