We investigated methods of including covariates in two-level models for cluster randomized trials to increase power to detect the treatment effect. We compared multilevel models that included either an observed cluster mean or a latent cluster mean as a covariate, as well as the effect of including Level 1 deviation scores in the model. A Monte…

Standard least squares analysis of variance methods suffer from poor power under arbitrarily small departures from normality and fail to control the probability of a Type I error when standard assumptions are violated. This article describes a framework for robust estimation and testing that uses trimmed means with an approximate degrees of…

A maximum test in which the test statistic is the more extreme of the Brown-Forsythe and in which O'Brien's test statistics are developed, with estimated Type I error rates and power for all three tests. For study conditions, Type I error rates for the maximum test are near the nominal level. (SLD)

One approach to the analysis of repeated measures data allows researchers to model the covariance structure of their data rather than presume a certain structure, as is the case with conventional univariate and multivariate test statistics. This mixed-model approach, available through SAS PROC MIXED, was compared to a Welch-James type statistic.…

Type I error rates and power were estimated for 10 tests of variance equality under various combinations of the following factors: similar and dissimilar distributional forms, equal and unequal means, and equal and unequal sample sizes. (TJH)

For Y. Yao's and G. S. James' tests, Type I error rates were estimated for various combinations of the number of variables, sample-size and sample-size-to-variables ratios, and heteroscedasticity. These tests are alternatives to Hotelling's T(sup 2) and are intended for use when variance-covariance matrices are unequal for two independent samples.…

Alternative tests are presented for the between-by-within interaction null hypothesis and for two within-subjects main effects null hypothesis in a split plot design. Estimated Type I error rates for the interaction tests and for several tests of the second null hypothesis are reported. (SLD)

Parametric analysis of covariance was compared to analysis of covariance with data transformed using ranks. Using a computer simulation approach the two strategies were compared in terms of the proportion of Type I errors made and statistical power when the conditional distribution of errors were: (1) normal and homoscedastic, (2) normal and…

This paper examined the rank transformation approach to analysis of variance as a solution to the Behrens-Fisher problem. Using simulation methodology four parameters were manipulated for the two group design: (1) ratio of population variances; (2) distribution form; (3) sample size and (4) population mean difference. The results indicated that…

Sampling distributions for ten tests for comparing population variances in a two group design were generated for several combinations of equal and unequal sample sizes, population means, and group variances when distributional forms differed. The ten procedures included: (1) O'Brien's (OB); (2) O'Brien's with adjusted degrees of freedom; (3)…

Type I error rates of four multivariate tests (Pilai-Bartlett trace, Johansen's test, James' first-order test, and James' second-order test) were compared for heterogeneous covariance matrices in 360 simulated experiments. The superior performance of Johansen's test and James' second-order test is discussed. (SLD)