K-means cluster analysis is known for its tendency to produce spherical and equally sized clusters. To assess the magnitude of these effects, a simulation study was conducted, in which populations were created with varying departures from sphericity and group sizes. An analysis of the recovery of clusters in the samples taken from these…

Developed a full maximum likelihood method for obtaining joint estimates of variances and correlations among continuous and polytomous variables with incomplete data that are missing at random with an ignorable missing mechanism. Simulation results and an empirical example illustrate the approach. (SLD)

Studied a multiple group model with ordinal categorical observed variables that are manifestations of underlying normal variables. Proposed to apply across-group stochastic constraints on thresholds to identify the model and used a Bayesian approach to analyze the model. Simulation findings and the analysis of a real data set show the usefulness…

Through simulation, showed that a multivariate test of interactions for aligned ranks in a split-plot design controlled Type I error rates for nonnormal data with nonspherical covariance structures. This method also performed well in the presence of a strong repeated measures main effect and demonstrated more statistical power than parametric…

Simulated test data (N=864 artificial data sets) with four different error conditions were used to study the recovery characteristics of the beta-flexible clustering method. Conditions under which the beta-flexible method provides good recovery are discussed. (SLD)

Three-dimensional response surface plots are presented for several hierarchical clustering methods and simulated core group data structures. Influence patterns explain some results from previous validation research on clustering methods and have significant implications for the choice of clustering methods in empirical research. (SLD)

Presents a simple method for generating simulated plasmodes and artificial test clusters with user-defined shape, size, and orientation. For "J" clusters, indicator validity is defined as the squared correlation ratio between the cluster indicator and J-1 dummy variables. Illustrates the method through simulation. (SLD)

Developed a sample-based nonanalytical degrees-of-freedom correction factor for situations sampling both subjects and conditions with measurement data departing from essentially parallel form. Assessed the application of this correction factor through a simulation study involving data sets with a range of design characteristics and manifesting…

Presents a Bayesian method for obtaining an interval estimate of the population squared multiple correlation from an incomplete multivariate normal data set. Estimates were constructed using Gibbs sampling. Simulation studies indicate that the method can yield accurate interval estimates of the population squared multiple correlation. (SLD)

Investigated the usefulness of multidimensional scaling (MDS) for assessing the dimensionality of dichotomous test data. Focused on two MDS proximity measures, one based on the PC statistic (T. Chen and M. Davidson, 1996) and other, on interitem Euclidean distances. Simulation results show that both MDS procedures correctly identify…

Presents and evaluates three estimators for "p," the proportion of success in predicting variable "Y," with nominal measurement, using predictor variables that also have nominal measurement. Showed through simulation that one estimator is always biased upward, and then proposed another possible estimator that involves using…

The search for cut-off criteria of fit indices for model fit evaluation (e.g., Hu & Bentler, 1999) assumes that these fit indices are sensitive to model misspecification, but not to different types of models. If fit indices were sensitive to different types of models that are misspecified to the same degree, it would be very difficult to establish…

The performance of five simple multiple imputation methods for dealing with missing data were compared. In addition, random imputation and multivariate normal imputation were used as lower and upper benchmark, respectively. Test data were simulated and item scores were deleted such that they were either missing completely at random, missing at…

Conducted simulations to explore methods for comparing bivariate distributions corresponding to two independent groups, all of which are based on Tukey's "depth," a generalization of the notion of ranks to multivariate data. Discusses steps needed to control Type I error. (SLD)

Proposed a method for extending the Bollen-Stine bootstrap model (K. Bollen and R. Stine, 1992) fit to structural equation models with missing data. Developed a Statistical Analysis System macro program to implement this procedure, and assessed its usefulness in a simulation. The new method yielded model rejection rates close to the nominal 5%…