We develop semiparametric Bayesian Thurstonian models for analyzing repeated choice decisions involving multinomial, multivariate binary or multivariate ordinal data. Our modeling framework has multiple components that together yield considerable flexibility in modeling preference utilities, cross-sectional heterogeneity and parameter-driven…

Seven approaches to averaging reliability coefficients are presented. Each approach starts with a unique definition of the concept of "average," and no approach is more correct than the others. Six of the approaches are applicable to internal consistency coefficients. The seventh approach is specific to alternate-forms coefficients. Although the…

In the nineteenth century many people tried to seek a value for the most famous irrational number, [pi], by means of an experiment known as Buffon's needle, consisting of throwing randomly a needle onto a surface ruled with straight parallel lines. Here we propose to extend this experiment in order to evaluate other irrational numbers, such as…

Whether "more is ever too much" for the number of indicators (p) per factor (p/f) in confirmatory factor analysis (CFA) was studied by varying sample size (N) from 50 to 1,000 and p/f from 2 to 12 items per factor in 30,000 Monte Carlo simulations. For all sample sizes, solution behavior steadily improved (more proper solutions and more…

Conventional wisdom suggests the omnibus F-test needs to be significant before conducting post-hoc pairwise multiple comparisons. However, there is little empirical evidence supporting this practice. Protected tests are conducted only after a significant omnibus F-test while unprotected tests are conducted without regard to the significance of the…

This study examined the Type I error and power properties of the rank transform test when employed in the context of a balanced 2x2x2 fixed effects analysis of variance (ANOVA). Computer generated Monte Carlo methods were used to compare the Type I error and power properties to those used in the usual test. The results showed the rank transform…

Results showed that the canonical correlations are very stable upon replication. The results also indicated that there is no solid evidence for concluding that components are superior to the coefficients, at least not in terms of being more reliable. (Author/BJG)

Monte Carlo procedures are used to develop stress distributions using Kruskal's second stress formula. These distributions can be used in multidimensional scaling procedures to determine whether a set of data has other than random structure. (Author/JKS)

Four testing procedures for establishing the number of non-zero population roots in canonical analysis are investigated. Results of a Monte Carlo study indicate that three well-established procedures were effective, and a new procedure designed to correct a supposed flaw in the other procedures was ineffective. (JKS)

A general procedure is provided for comparing correlation coefficients between optimal linear composites. It allows computationally efficient significance tests on independent or dependent multiple correlations, partial correlations, and canonical correlations, with or without the assumption of multivariate normality. Evidence from Monte Carlo…

Showed that reliabilities in the .50 range can be obtained when raters rule out only 15-20% of the items on the Position Analysis Questionnaire as "Does Not Apply" and respond randomly to the remainder. (Author/ABB)

In order to make the parallel analysis criterion for determining the number of factors in factor analysis easy to use, regression equations for predicting the logarithms of the latent roots of random correlation matrices, with squared multiple correlations on the diagonal, are presented. (Author/JKS)

A special case of the homogeneity of effect size test, as applied to pairwise comparisons of standardized mean differences, was evaluated. Procedures for comparing pairs of pretest to posttest effect sizes, as well as pairs of treatment versus control group effect sizes, were examined. Monte Carlo simulation was used to generate Type I error rates…

The effect of a nonlinear regression term on the behavior of the standard analysis of covariance (ANCOVA) F test was investigated for balanced and randomized designs through a Monte Carlo study. The results indicate that the use of the standard analysis of covariance model when a quadratic term is present has little effect on Type I error rates…

Research was conducted to provide educational researchers with a choice of pairwise multiple comparison procedures (P-MCPs) to use with single group repeated measures designs. The following were studied through two Monte Carlo (MC) simulations: (1) The T procedure of J. W. Tukey (1953); (2) a modification of Tukey's T (G. Keppel, 1973); (3) the…