Illustrates the computation of confidence intervals for effect sizes for some analysis of variance applications and shows how the use of intervals involving noncentral distributions is made practical by new software. (SLD)

Compared differential item functioning detection rates for logistic regression and analysis of variance for dichotomously scored items using simulated data and varying test length, sample size, discrimination rate, and underlying ability. Explains why the logistic regression method is recommended for most applications. (SLD)

Provides equations for computing confidence-interval estimates and conducting "F" tests for intraclass correlation coefficients defined using three-way random-effects models for crossed and nested designs. The estimates and tests use mean squares from analyses of variance. (Author/SLD)

Some results on how the Alexander-Govern heteroscedastic analysis of variance (ANOVA) procedure (R. Alexander and D. Govern, 1994) performs under nonnormality are presented. This method can provide poor control of Type I errors in some cases, and in some situations power decreases as differences among the means get large. (SLD)

Empirical estimates of the power and Type I error rate of the test of the classrooms-within-treatments effect in the nested analysis of variance approach are provided for a variety of nominal alpha levels and a range of classroom effect sizes and research designs. (SLD)

Univariate procedures proposed by M. Brown and A. Forsythe (1974) and the multivariate procedures from D. Nel and C. van der Merwe (1986) were generalized to form five new multivariate alternatives to one-way multivariate analysis of variance (MANOVA) for use when dispersion matrices are heteroscedastic. These alternatives are evaluated for Type I…

A method is presented for partitioning the sums of squares of treatments and interactions to reflect specific treatment and interaction contrasts. The method is illustrated by partitioning the sums of squares in a split-plot factorial analysis of variance design. (Author)

The analysis of variance model for estimating reliability in norm referenced tests is extended to criterion referenced tests. The essential modification is that the criterion or cut-off score is substituted for the population mean. An example and discussion are presented. (JKS)

A flexible Fortran program for computing one way analysis of variance is described. Requiring minimal core space, the program provides a variety of useful group statistics, all summary statistics for the analysis, and all mean comparisons for a priori or a posteriori testing. (Author/JKS)

Conventional paper-and-pencil testing was compared to computer-assisted testing on a pharmacy school pathology test. Individual items were speeded in the computer-assisted mode. In addition to responses, the number and pattern of changes in responses were analyzed. True-false, multiple-choice and matching items were used. (JKS)

A convenient, two-stage general linear regression approach to analysis of variance is described for use in univariate or multivariate designs involving one repeated measurement factor and one or more independent classification factors. A brief illustrative example is provided. (Author)

One group of subjects responded to an anxiety battery according to their general feelings and then according to their feeling under anxiety-related simulations. A control group received no special instructions. Analysis of variance of factor scores indicated that simulations altered level of state anxiety but not trait anxiety. (Author/JKS)

Systematically investigates the effects of test context on verbal analogy item difficulty, in terms of both simple percentage correct and easiness estimates from a parameter-invariant model (Rasch, 1960). (RC)