The Welch-James (WJ) and Improved General Approximation (IGA) tests for the within-subjects main and interaction effects in a split-plot repeated measurement design were investigated when least squares estimates and robust estimates based on trimmed means were used. Variables manipulated in the Monte Carlo study included the degree of multivariate…

The assumption of normality underlies much of the standard statistical methodology. Knowing how to determine whether a sample of measurements is from a normally distributed population is crucial both in the development of statistical theory and in practice. W. Ware and J. Ferron have developed a new test statistic, modeled after the K-squared test…

Covariance structure analytic techniques have become increasingly popular in recent years. During this period, users of statistical software packages have become more and more sophisticated, and more and more researchers are wanting to make sure that they are using the "best" statistic, whether it be for small sample considerations or…

To assess the effect on discriminant analysis in terms of correct classification into two groups, the following parameters were systematically altered using Monte Carlo techniques: sample sizes; proportions of one group to the other; number of independent variables; and covariance matrices. The pairing of the off diagonals (or covariances) with…

Two strategies, derived from J. P. Schaffer (1986), were compared as tests of significance for a complete set of planned orthogonal contrasts. The procedures both maintain an experimentwise error rate at or below alpha, but differ in the manner in which they test the contrast with the largest observed difference. One approach proceeds directly to…

Monte Carlo methods were used to investigate the robustness of techniques used in judging the magnitude of a sample correlation coefficient when observations are correlated. Empirical distributions of r, t, and Fisher's z were generated. A technique for controlling error rates in certain situations is suggested. (Author/BJG)

The effects of the violation of the assumption of normality in the conditional distributions of the dependent variable, coupled with the condition of multicollinearity upon the outcome of testing the hypothesis that the regression coefficient equals zero, are investigated via a Monte Carlo study. (Author/JKS)

An examination is made concerning the utility and design of studies comparing nonmetric multidimensional scaling algorithms and their initial configurations, as well as the agreement between the results of such studies. Various practical details of nonmetric scaling are also considered. (Author/JKS)

Several nonmetric multidimensional scaling random ranking studies are discussed in response to the preceding article (TM 503 490). The choice of a starting configuration is discussed and the use of principal component analysis in obtaining such a configuration is recommended over a randomly chosen one. (JKS)

The use of U-statistics based on rank correlation coefficients in estimating the strength of concordance among a group of rankers is examined for cases where the null hypothesis of random rankings is not tenable. (Author/BW)

A new measure of correlation and a measure of scale are proposed which are substantially more robust than their least squares counterparts. Increased robustness may also be obtained by use of equal regression weights, or knowledge of the theoretical structure of the weights. (Author/HG)

A Bayesian approach to the evaluation of person fit in item response theory (IRT) models is presented. In a posterior predictive check, the observed value on a discrepancy variable is positioned in its posterior distribution. In a Bayesian framework, a Markov Chain Monte Carlo procedure can be used to generate samples of the posterior distribution…

When multiple regression is used to develop a prediction model, sample size must be large enough to ensure stable coefficients. If sample size is inadequate, the model may not predict well in future samples. Unfortunately, there are problems and contradictions among the various sample size methods in regression. For example, how does one reconcile…

The analysis of interaction effects in multiple regression has received considerable attention in recent years, but problems with the valid identification of moderating variables have been noted by researchers. G. McClelland and C. Judd (1993), in their discussion of the statistical difficulties of detecting interactions and moderating effects,…

J. Barnette and J. McLean (1996) proposed a method of controlling Type I error in pairwise multiple comparisons after a significant omnibus F test. This procedure, called Alpha-Max, is based on a sequential cumulative probability accounting procedure in line with Bonferroni inequality. A missing element in the discussion of Alpha-Max was the…