This paper describes two BASIC computer programs that calculate Hotelling's T-square either for one sample or for two samples. Output of the progams includes the Mahalanobis distance D-square, the F ratio associated with T-square, and its probability level. (Author)

Because multivariate statistics are increasing in popularity with social science researchers, the challenge of detecting multivariate outliers warrants attention. Outliers are defined as cases which, in regression analyses, generally lie more than three standard deviations from Yhat and therefore distort statistics. There are, however, some…

This paper reviews graphical and nongraphical procedures for evaluating multivariate normality by guiding the reader through univariate and bivariate procedures that are necessary, but insufficient, indications of a multivariate normal distribution. A data set using three dependent variables for two groups provided by D. George and P. Mallery…

This paper reviews the concept of experimentwise Type I error. While "testwise" alpha refers to the probability of making a Type I error for a single hypothesis test, "experimentwise" error refers to the probability of having made a Type I error anywhere within the study. Experimentwise error concerns are the basis for two…

Proposes a comprehensive approach, generalized constrained multiple correspondence analysis, for imposing both row and column constraints on multivariate discrete data. Each set of discrete data is decomposed into several submatrices and then multiple correspondence analysis is applied to explore relationships among the decomposed submatrices.…

Proposes a method for performing survival analysis with latent variables with dual aims of scoring subjects with respect to a psychological variable of interest and distinguishing the varying extents to which items are capable of measuring the construct. Illustrates this discrete proportional hazards model. (SLD)

Proposes a generalization of Cohen's kappa coefficient (J. Cohen, 1960) to address the problem of accounting for overall chance-corrected interobserver agreement among the multivariate ratings of several judges. The statistic's metric is conventional and in the univariate case it is equivalent to existing extensions of the kappa coefficient to…

We use clustering, an analysis method not presently common to the physics education research community, to group and characterize student responses to written questions about two-dimensional kinematics. Previously, clustering has been used to analyze multiple-choice data; we analyze free-response data that includes both sketches of vectors and…

This article describes the development of a classification system to measure the expertise levels of practicing paediatric rehabilitation therapists. Seventy-five therapists from five disciplines (physical, occupational, speech-language, behaviour, and recreational therapy) were involved, along with 170 peers, and 188 parents of children with…

According to some researchers canonical correlation results should be interpreted in part by consulting redundancy coefficients (Rd). This paper, however, makes the case that Rd coefficients generally should not be interpreted. Rd coefficients are not multivariate. Furthermore, it makes little sense to interpret coefficients not optimized as part…

Methods to compare population means in the univariate case and population mean vectors in the multivariate case are presented in terms of hypotheses tested by various procedures. Tests relevant to each hypothesis are described and compared in terms of maximizing power while controlling Type I error rates over the widest variety of conditions. (SLD)

Outlines the multivariate latent distribution specification and the corresponding interpretation issues. Also shows the univariate formulation and some alternative parameterizations that are useful in the estimation phase. Illustrates the procedures through an application to student ratings data. (SLD)

Proposes a Bayesian analysis of the multivariate linear model with polytomous variables. Shows how a Gibbs sampler algorithm is implemented to produce the Bayesian estimates. Illustrates the proposed methodology through examples using multivariate linear regression and multivariate two-way analysis of variance with real data. (SLD)

Provides an overall test statistic to evaluate significance that yields 100% (1-alpha) confidence intervals for all linear combinations of effect sizes, shows how the stepdown Finite Intersection Test (N. Timm, 1995) may be used to test whether effect sizes are simultaneously different from zero, and illustrates the procedure. (SLD)

Presents an application of the EM-algorithm to two problems of estimation and testing in a multivariate normal distribution with missing data. The two models are tested applying the log-likelihood ratio test. Solves the problem of different and nonmonotone patterns of missing data by introducing suitable transformations and partitions of the data…