Type I error rates of the likelihood ratio test for the detection of differential item functioning (DIF) were investigated using Monte Carlo simulations. The graded response model with five ordered categories was used to generate data sets of a 30-item test for samples of 300 and 1,000 simulated examinees. All DIF comparisons were simulated by…

C. L. Olson (1976, 1979) suggests the Pillai-Bartlett trace (V) as an omnibus multivariate analysis of variance (MANOVA) test statistic for its superior robustness to heterogeneous variances. J. Stevens (1979, 1980) contends that the robustness of V, Wilk's lambda (W) and the Hotelling-Lawley trace (T) are similar, and that their power functions…

The Type I error and power properties of the 2x2x2 analysis of variance (ANOVA) and tests developed by McSweeney (1967), Bradley (1979), Harwell-Serlin (1989; Harwell, 1991), and Blair-Sawilowsky (1990) were compared using Monte Carlo methods. The ANOVA was superior under the Gaussian and uniform distributions. The Blair-Sawilowsky test was…

Induced probabilities have been largely ignored by educational researchers. Simply stated, if a new or random variable is defined in terms of a first random variable, then induced probability is the probability or density of the new random variable that can be found by summation or integration over the appropriate domains of the original random…

P. O. Johnson and J. Neyman (1936) proposed a general linear hypothesis testing procedure for testing the null hypothesis of no treatment difference in the presence of some covariates. This is generally known as the Johnson-Neyman (JN) technique. The need for the hypothesis testing step (often omitted) as originally presented and the…

This Monte-Carlo study compared modified Newton (NW), expectation-maximization algorithm (EM), and minimum Cramer-von Mises distance (MD), used to estimate parameters of univariate mixtures of two components. Data sets were fixed at size 160 and manipulated by mean separation, variance ratio, component proportion, and non-normality. Results…

Psychology studies often have low statistical power. Sample size tables, as given by J. Cohen (1988), may be used to increase power, but they are based on Monte Carlo studies of relatively "tame" mathematical distributions, as compared to psychology data sets. In this study, Monte Carlo methods were used to investigate Type I and Type II…

Monte Carlo studies of statistical tests are prominently featured in the methodological research literature. Unfortunately, the information from these studies does not appear to have significantly influenced methodological practice in educational and psychological research. One reason is that Monte Carlo studies lack an overarching theory to guide…

An influential statistics test recommends a Levene text for homogeneity of variance. A recent note suggests that Levene's test is upwardly biased for small samples. Another report shows inflated Alpha estimates and low power. Neither study utilized more than two sample sizes. This Monte Carlo study involved sampling from a normal population for…

The Monte Carlo method was used, and the factors considered were (1) level of main effects in the population; (2) level of interaction effects in the population; (3) alpha level used in determining whether to pool; and (4) number of degrees of freedom. The results indicated that when the ratio degrees of freedom (axb)/degrees of freedom (within)…

INTERTAIL, the computer program which implements an approach to tailored testing outlined by Cliff (1975), was examined with errorless data in several Monte Carlo studies. Three replications of each cell of a 3 x 3 table with 10, 20 and 40 items and persons were analyzed. Mean rank correlation coefficients between the true order, specified by…

The effects of sample size, number of variables, and population value of the congruence coefficient on the sampling distribution of the congruence coefficient were examined. Sample data were generated on the basis of the common factor model, and principal axes factor analyses were performed. (Author/LMO)

This paper investigates three methods for comparing medians rather than means in studying two independent treatment groups. The method that gave the best results is based on a normal approximation of the distribution of the sample median where the variance is estimated using results reported by Maritz and Jarrett. (Author/JAZ)

The purpose of the current study was to investigate the relative performance of the parametric, rank, and normal scores procedures when the classical assumptions were met and under violations of these assumptions. This investigation included the normal scores as well as the rank test. (LMO)

Reviews the principles of Monte Carlo calculation and random number generation in an attempt to introduce the direct and the rejection method of sampling techniques as well as the variance-reduction procedures. Indicates that the increasing availability of computers makes it possible for a wider audience to learn about these powerful methods. (CC)